Dating Celtic Art: a Major Radiocarbon Dating Programme of Iron Age and Early Roman Metalwork in Britain

Abstract

This paper presents the first substantial set of radiocarbon determinations for the later Iron Age decorated metalwork known as Celtic art in Britain. Hitherto this material has been dated relatively on the basis of changes in decoration and form, which were then linked to materials with some absolute date. The latter process has tended to give relatively late dates, as most of the material with a firm date stems from the last century BC or after. This has meant that British Celtic art appears to be rather later than that on the Continent. Our results provide some tentative support for an earlier dating for at least some British material, more closely aligned to that on the Continent. Stead, building on earlier work, has developed a series of Stages or Styles for the decorations of Celtic art and we shall review these here. We also look at particular classes of artefacts and forms of deposition in the light of our new results. We conclude by drawing some contrasts between Bronze Age metalwork and that of Celtic art, reflecting on how far assumptions of a sequential series of changes, key to all forms of typology, are useful in the case of the latter.

Full text

This paper presents the first substantial set of radiocarbon determinations for the later Iron Age
decorated metalwork known as Celtic art in Britain. Hitherto this material has been dated
relatively on the basis of changes in decoration and form, which were then linked to materials
with some absolute date. The latter process has tended to give relatively late dates, as most of
the material with a firm date stems from the last century BC or after. This has meant that British
Celtic art appears to be rather later than that on the Continent. Our results provide some
tentative support for an earlier dating for at least some British material, more closely aligned to
that on the Continent. Stead, building on earlier work, has developed a series of Stages or Styles
for the decorations of Celtic art and we shall review these here. We also look at particular classes
of artefacts and forms of deposition in the light of our new results. We conclude by drawing some
contrasts between Bronze Age metalwork and that of Celtic art, reflecting on how far assump -
tions of a sequential series of changes, key to all forms of typology, are useful in the case of the
latter.
INTRODUCTION
In their Agenda for Action concerning the British Iron Age, Haselgrove et al. (2001,
4–6) highlight the need for radiometric dating of British Iron Age sites and materials.
Significantly, they pick out Celtic art as a key body of material in need of dating. From
the Early Bronze Age onwards chronologies of metalwork are problematic, both in
terms of providing links between metal objects and other forms of archaeological
evidence, and — particularly for the Iron Age — in terms of dating the entire corpus
of material and its stylistic changes. In the Iron Age, as in earlier periods, much
metalwork derives from rivers, bogs or chance finds that provide little reliable infor -
ma tion on archaeological context. For this reason, in many cases, the dating of the
objects themselves, or of material in direct association with them, is the only possible
route to an absolute chronology. In this paper we present the first substantial pro -
gramme of radiocarbon determinations for Celtic art from Britain (and indeed the first
large suite from anywhere in Europe) and discuss their implications for relations with
Dating Celtic Art: a Major Radiocarbon Dating
Programme of Iron Age and Early Roman
Metalwork in Britain
d. garrow, c. gosden, j. d. hill,
and c. bronk ramsey
Archaeol. J., 166 (2009), 79‒123
A3 Garrow:Layout 1 01/09/2010 15:03 Page 79

material on the Continent, as well as for internal relationships of material in Britain.
Overall, the results are thought-provoking, stimulating us to pose new questions con -
cerning this complex body of material.
The dating project discussed in this paper was designed to complement that carried
out by Needham et al. (1998), which looked at dates of metalwork from the Middle
Bronze Age through to the beginning of the Iron Age. Needham et al. took the typo -
logical stages of metalwork and the stylistically deduced dates attached to these and
compared them with radiocarbon determinations from a considerable number of
objects. In this paper, we too attempt to link changes in the form and decoration of
metal objects to radiocarbon determinations. Together the two projects allow an
under standing of the changes in British metalwork in the second and first millennia
bc (albeit with a gap in the Early Iron Age which is impossible to bridge using radio -
carbon dating, given the plateau in the calibration curve between 800 and 400 bc).
As we discuss below, until now the dating of most Celtic art in Britain has been
through typologies of form and decoration, ordered from early to late and given some
sense of absolute date either by links with Continental developments, or through
associa tion with coins and imports found from the Late Iron Age onwards. The pro -
cess of dating through links with these objects has resulted in a drift in dates towards
the Late Iron Age for many British Celtic art objects, with far fewer objects securely
dated in the fourth–second centuries bc. This has left a gap between the British and
Continental chronologies that is not seen in other forms of material culture such as
fibulae or coins (Haselgrove 1999). Although there are increasing numbers of radio -
carbon determinations from the British Iron Age, only a few of these are associated
directly with objects deemed to be Celtic art (although see Table 2). Even where
Celtic art has been dated, many dates were obtained in the 1970s when techniques for
both dating materials and calibration were less developed than now (e.g. Wainwright
1979, 125). Consequently, our key aim in this dating project has been to date as large
a number of objects as possible, either through material that was part of the original
object (primarily leather or wood), or through materials found in direct archaeological
association with the object (primarily bone). By running the samples in a single labora -
tory and using a standard set of calibration methods we hoped to achieve as much
consistency as possible. In addition to the dates obtained directly as a result of the pro -
ject described here, we have also included all of the recently obtained (and therefore
reliable) dates on Celtic art elsewhere (see Tables 1and 2).
In order to discuss both the dates and their broader context, we shall look briefly
first at general problems in dating the British Iron Age, as well as at Continental chro -
nol ogies and the nature of links between Britain and Europe. A consideration of the
stylistic schemes that have been applied to British Celtic art then follows. We then
move to the objects we have chosen for dating, including consideration of the objects
we tried to date but for various reasons could not. Note that original bibliographic
refer ences for each of the objects/contexts dated are provided in Table 1. The radio -
carbon determinations are then presented ranging from earliest to latest; these are next
discussed in terms of various classes of artefact and forms of decoration. We end with
a discussion which broadens out to consider more general issues of dating the Iron
Age and the place of Celtic art within it.
80 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 80

british iron age chronologies: britain’s relationship with
the continent
Discussion of the British Iron Age has been at times obsessively concerned with chro -
nology. Chronology has been controversial from at least the early 1960s because of the
differing assumptions that lie behind the various schemes. Most models up to and
including that of Hawkes (1959) saw invasions or migrations as being responsible for
the nature and timing of change, as new groups brought with them packages of
material culture, and forms of settlement and burial. Most models after Hodson (1964)
empha size indigenous and continual change in Britain, seeing a very limited role for
influ ence from the outside. Given the complexity of the evidence and of the nature
of chronological change, it is no longer seen as desirable to have a single detailed
scheme for all areas and periods — indeed, the complicated revisions to Hawkes’ ABC
scheme were one key reason for its demise. Now broad, general chronologies are
favoured, which can then be nuanced and given a more realistic amount of com -
plexity for particular sites and areas (see Collis 2008). The simplest such scheme,
which has achieved quick and wide acceptance, is that most clearly outlined in two
recent volumes on the Iron Age, constructed around a two-fold division into an
Earlier Iron Age (c.800 to 400/300 bc — Haselgrove and Pope 2007) and a Later Iron
Age (400/300 bc to ad 43 — Haselgrove and Moore 2007). After periods of some
acrimony, a broad consensus has emerged, shared generally by those interested in the
Iron Age, even if there are a considerable number of more detailed issues that still
need resolution.
At the heart of these chronological debates has lain the issue of Britain’s chronol -
ogical links with Europe. The pendulum of opinion has swung from large-scale
migra tions into southern England to development in relative isolation. Presently a
middle position of contact and interchange, without migration, is most popular.
Two apparently conflicting views of Britain’s relationship with Europe exist. On the
one hand Stead (1984) points out that there are very few definite imports from the
Continent prior to 150 bc. On the other, Haselgrove (2001) shows that for particular
periods, such as the Early and Late Iron Age and for particular areas, such as southern
Britain, there are good links with the Continent. Views that stress insularity have been
opposed to those emphasizing closer links between Britain and the Continent.
However, the either/or nature of this opposition does not provide realistic alternatives
for the Iron Age, at least in circumstances where more interesting cultural develop -
ments were taking place. As Haselgrove shows, and we shall discuss briefly below,
there are, on the one hand, very close links between northern France and southern or
eastern Britain. Similar forms of material culture and housing must indicate
interchange between them. On the other hand, there are very few material culture
imports from northern France in Britain; insular artefacts have a character and style of
their own, suggesting that a rather diffuse set of influ ences came into this country and
were quickly internalized. The case of Celtic art adds a further complication to this
picture of links with the Continent as they clearly existed alongside a strongly local
outlook. This is demonstrated by the fact that the Continental areas closest to Britain
81dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 81

table 1 Details of all objects sampled
Object County Project Context Sample Est. age of Location Reference Jope ‘Style’Lab no.
ID description wood (2000)
Ref. no.
Aylesford Kent CA03 Cremation Ash wood from old? BM Evans 1890 134–143 – OxA-18218
bucket bucket stave
Bermondsey London CA07 River Ash wood from old? M of L Cotton and – – OxA1-7332
dagger scabbard lining Green 2004
Boverton collar Glamorgan CA08 Inhumation Human bone found – NMW A. Gwilt, – VI OxA-17505
and bracelets inside bracelet pers. comm.
Brecon grave Powys CA09 Cremation Cremated human bone – NMW A. Gwilt, – VI OxA-17455
assemblage found with artefacts pers. comm.
Bury Hill Hants. CA11 Pit on settlement Animal bone from same – Chilcomb Cunliffe and – – OxA-17282
horse gear (P24 layer 9) layer in pit House, Poole 2000
Winchester
Bury Hill Hants. CA12 Pit on settlement Animal bone from layer – Chilcomb Cunliffe and – – OxA-17283
horse gear (P45 layer 10) immediately above in pit House, Poole 2000
Winchester
Cerrigydrudion Conwy CA13 Probable burial Rawhide originally – NMW Stead 1982 29–30 I OxA-X-2226-17
headpiece attached to rim of
headpiece
Deal spoons Kent CA14 Inhumation Human skull fragment – Dover Parfitt 1995 231 c – OxA-17284
(X2/D60) from inhumation
Deal warrior Kent CA15 Inhumation Human skull fragment – BM Parfitt 1995 205 o IV/V OxA-17506
(Grave 112) from inhumation
Ely horn cap Cambs. CA16 Stray find Ash wood from interior <5yrs BM J. D. Hill, – OxA-18190
of object pers. comm.
Garton Slack E. Yorks. CA17 Inhumation Bone from articulated – Hull Brewster 1971;264 g V OxA-17285
cart burial (Garton Slack pig skeleton found 1975
XI Barrow 2) within grave
Garton Slack E. Yorks. CA18 Inhumation Human arm bone from – Hull Brewster 1975 237 d–e V OxA-17286
mirror (Garton Slack inhumation
VII Barrow 2)
Garton Station E. Yorks. CA19 Inhumation Pig bone placed around – BM Stead 1991b127 i V OxA-17507
cartburial (GS6) inhumation
A3 Garrow:Layout 1 01/09/2010 15:03 Page 82

Grimthorpe E. Yorks. CA21 Inhumation Bone point found within – BM Stead 1968 92, 93, V OxA-17508
warrior grave 202 l–p
Gussage All Saints Dorset CA22 Pit on settlement Unidentified wood <5yrs BM Wainwright – – OxA-18191
metalworking (Pit 209, charcoal 1979
deposit layer 10a)
Gussage All Saints Dorset CA23 Pit on settlement Willow wood charcoal <5yrs BM Wainwright – – OxA-18192
metalworking (Pit 209 layer 12)1979
deposit
Kew tankard London CA24 River Yew wood from old? M of L Corcoran 1952,226 c – OxA-17435
tankard stave 98
Kirkburn chariot E. Yorks. CA25 Inhumation (K5) Pig bone placed around – BM Stead 1991b127 l V OxA-17509
burial inhumation
Kirkburn E. Yorks. CA26 Inhumation (K3) Human skull fragment – BM Stead 1991b127 j V OxA-17510
warrior from inhumation
Latchmere Green Hants. CA28a Cremation Cremated human bone – Andover Fulford and 248 g V OxA-17287
mirror found with mirror Creighton 1998
Latchmere Green Hants. CA28b Cremation Cremated human bone – Andover Fulford and 248 g V OxA-18355
mirror found with mirror Creighton 1998
Latchmere Green Hants. CA28c Cremation Cremated human bone – Andover Fulford and 248 g V OxA-18356
mirror found with mirror Creighton 1998
Llyn Cerrig Bach Anglesey CA44 Watery deposit Oak wood found inside old? NMW MacDonald – – OxA-17512
chariot pole iron chariot pole (POW) 2007b
Llyn Cerrig Bach Anglesey CA29a Watery deposit Ash wood shaft preserved <10 yrs NMW MacDonald – – OxA-17436
spear inside spearhead 2007b
Llyn Cerrig Bach Anglesey CA29b Watery deposit Ash wood shaft preserved <10 yrs NMW MacDonald – – OxA-17437
spear inside spearhead 2007b
Owslebury Hants. CA30a Inhumation Human rib fragment – Chilcomb Collis 1968 93 c–g V OxA-17288
warrior from inhumation House,
Winchester
Owslebury Hants. CA30b Inhumation Human rib fragment – Chilcomb Collis 1968 93 c–g V OxA-17289
warrior from inhumation House,
Winchester
Pentuan tankard Cornwall CA31 River Wood (probably yew) old? Truro Corcoran 1952,226 d – OxA-17366
from tankard stave 96
Salisbury Wiltshire CA43 Hoard Cow bone found in pit – BM Stead 2000 99 f–h V OxA-17511
(Netherhampton) into which the hoard
hoard pit was cut
A3 Garrow:Layout 1 01/09/2010 15:03 Page 83

Snettisham (Torc Norfolk CA33 Hoard Field maple wood <10 yrs BM Stead 1991a108–109 III/IV OxA-18219
241, Hoard F) charcoal inside torc
Snettisham (Torc Norfolk CA46 Hoard Lime wood fibre string <5yrs BM Stead 1991a108–109 III/V OxA-18222
SN/NZ, Hoard L) bound around torc
Snettisham (ingots Norfolk CA47 Hoard Alder wood charcoal old BM Stead 1991a108–109 – OxA-18223
in Hoard M) embedded within silver
lumps from hoard
Trawsfynydd Gwynedd CA39 Watery deposit Yew wood from tankard old? Liverpool Corcoran 228–9 a–d VI OxA-17439
tankard stave 1952, 97–98
Welshpool grave Powys CA40 Probable burial Yew wood from bucket old? NMW Boon 1961 183 a–d – OxA-17440
assemblage stave
suspect dates
Asby Scar sword/ Cumbria CA01 Stray find Ash wood from interior old? BM Stead 2006 – VI OxA-17581
scabbard (ash) of sword handle no. 203
Asby Scar sword/ Cumbria CA45 Stray find Lime wood from interior old? BM Stead 2006 – VI OxA-17582
scabbard (lime) of sword handle no. 203
Mortlake dagger London CA38 River Unidentified wood from old? M of L Jope 2000 12 a–b, – OxA-17367
scabbard 13 a–c
Snettisham (Torc Norfolk CA35 Hoard Lime wood fibre string <5yrs BM Stead 1991a108–109 III/V OxA-18220
6, Hoard L) bound around torc
Stanwick hoard N. Yorks. CA37 Hoard Charcoal found with <5yrs BM MacGregor 278 a, – OxA-18221
hoard 1962 285 e, 287
Stanwick scabbard N. Yorks. CA38 River Ash wood from scabbard old? BM Wheeler 1954 215 a–e – OxA-17438
failed dates
Bury Hill Hants. CA10 Pit in settlement Animal bone from same – Chilcomb Cunliffe and 127 f – OxA
horse gear (P24 layer 1) layer in pit House, Poole 2000
Winchester
Snettisham (Torc Norfolk CA32 Hoard Hazel wood charcoal <10 yrs BM Stead 1991a108–109 III/IV/ OxA
242, Hoard F) inside torc V
Snettisham (Torc Norfolk CA34 Hoard Lime wood fibre string <5yrs BM Stead 1991a108–109 – OxA-18164
17, Hoard G) bound around torc
table 1 (continued)
Object County Project Context Sample Est. age of Location Reference Jope ‘Style’Lab no.
ID description wood (2000)
Ref. no.
A3 Garrow:Layout 1 01/09/2010 15:03 Page 84

Welwyn bucket Herts. CA41 Cremation Unidentified wood from old? BM Stead 1967 – – OxA
stave of bucket
Previous dates
Balmaclellan Dumfries n/a Hoard Cloth wrapped around – NMS F. Hunter, 224, 252 – OxA-6742
hoard and hoard pers. comm.
Galloway
Bryher mirror and Isles of n/a Inhumation Human long bone – St Mary’s, Johns 2006 – V OxA-12095
sword/scabbard Scilly Scilly
Burnmouth Borders n/a Inhumation Human bone – NMS F. Hunter, 427–8VGrA-27301
spoons pers. comm.
Chertsey shield Surrey n/a Stray find Ash wood from shield ? BM Hedges et al. 69 – OxA-993
handle 1988
Ferrybridge W. Yorks. n/a Inhumation Human radius – Oxford Brown et al. – – NZA-20494
chariot burial Arch. North 2007
Ferrybridge W. Yorks. n/a Inhumation Human tibia – Oxford Brown et al. – – NZA-19423-5
chariot burial Arch. North 2007
Ferrybridge W. Yorks. n/a Inhumation Associated pig humerus – Oxford Brown et al. – – NZA-20496
chariot burial Arch. North 2007
Ferrybridge W. Yorks. n/a Inhumation Associated pig skull – Oxford Brown et al. – – NZA-21727-9
chariot burial Arch. North 2007
Garton Station E. Yorks. n/a Inhumation Human femur – BM J. D. Hill 127 iV SR-6928
chariot burial (GS6) pers. comm.;
see Stead 1991b
for context
Kirkburn E. Yorks. n/a Inhumation Human femur – BM J. D. Hill, 127 lV SR-6930
chariot burial (K5) pers. comm.;
see Stead 1991b
for context
Kirkburn E. Yorks. n/a Inhumation Human femur – BM J. D. Hill, 127 jIV/V SR-6929
warrior (K3) pers. comm.;
see Stead 1991b
for context
Newbridge Midlothian n/a Inhumation Wood from wheel rims ? NMS Carter and – – GO-107510
chariot burial Hunter 2003
Newbridge Midlothian n/a Inhumation Wood from wheel rims ? NMS Carter and – – GO-107512
chariot burial Hunter 2003
A3 Garrow:Layout 1 01/09/2010 15:03 Page 85

Wetwang Slack E. Yorks. n/a Inhumation Human bone – BM J. D. Hill, – V OxA-14112
chariot burial 1/(453) pers. comm.;
sword see Stead 1991b
for context
Wetwang Slack E. Yorks. n/a Inhumation Human bone – BM J. D. Hill, – V OxA-14113
chariot burial 2/(454) pers. comm.;
’canister’ see Stead 1991b
for context
table 1 (continued)
Object County Project Context Sample Est. age of Location Reference Jope ‘Style’Lab no.
ID description wood (2000)
Ref. no.
table 2 Radiocarbon measurements and calibrated date ranges
Object Project Lab no. Material d13CRadiocarbon Date ranges
ID measurement BP at 95.4%
Aylesford bucket CA03 OxA-18218 Wood –21.46 2137 ± 30 360–290 cal. bc (17.1%)
230–50 cal. bc (78.2%)
Bermondsey dagger CA07 OxA-17332 Wood –25.60 2557 ± 28 810–740 cal. bc (59.6%)
690–660 cal. bc (16.2%)
650–550 cal. bc (19.6%)
Boverton collar and bracelets CA08 OxA-17505 Bone –19.75 1855 ± 26 80–230 cal. ad (95.4%)
Brecon grave assemblage CA09 OxA-17455 Bone (crem.) –21.37 1905 ± 28 20–180 cal. ad (93.0%)
190–220 cal. ad (2.4%)
Bury Hill horse gear CA11 OxA-17282 Bone –21.34 2135 ± 28 370–160 cal. bc (94.6%)
130–120 cal. bc (0.8%)
A3 Garrow:Layout 1 01/09/2010 15:03 Page 86

Bury Hill horse gear CA12 OxA-17283 Bone –21.94 2175 ± 29 360–300 cal. bc (14.2%)
230–220 cal. bc (0.3%)
210–50 cal. bc (80.8%)
Cerrigydrudion headpiece CA13 OxA-X-2226-17 Rawhide –28.65 2329 ± 29 510–430 cal. bc (5.6%)
420–350 cal. bc (88.3%)
280–250 cal. bc (1.5%)
Deal spoons CA14 OxA-17284 Bone –20.34 2216 ± 28 380–200 cal. bc (95.4%)
Deal warrior CA15 OxA-17506 Bone –19.63 2158 ± 28 360–280 cal. bc (39.2%)
260–100 cal. bc (56.2%)
Ely ‘horn cap’ CA16 OxA-18190 Wood –27.67 2078 ± 25 180–30 cal. bc (95.4%)
Garton Slack cart burial CA17 OxA-17285 Bone –21.67 2172 ± 29 370–160 cal. bc (93.7%)
140–110 cal. bc (1.7%)
Garton Slack mirror CA18 OxA-17286 Bone –20.50 2215 ± 29 380–200 cal. bc (95.4%)
Garton Station cart burial CA19 OxA-17507 Bone –21.73 2198 ± 27 380–190 cal. bc (95.4%)
Grimthorpe warrior CA21 OxA-17508 Bone –21.89 2171 ± 26 360–160 cal. bc (94.4%)
130–120 cal. bc (1.0%)
Gussage All Saints CA22 OxA-18191 Wood –28.01 2120 ± 25 360–280 cal. bc (38.5%)
metalworking deposit 260–100 cal. bc (56.9%)
Gussage All Saints CA23 OxA-18192 Wood –25.61 2157 ± 26 340–330 cal. bc (1.3%)
metalworking deposit 210–50 cal. bc (94.1%)
Kew tankard CA24 OxA-17435 Wood –21.63 2131 ± 29 350–300 cal. bc (11.0%)
210–50 cal. bc (84.4%)
Kirkburn chariot burial CA25 OxA-17509 Bone –22.05 2146 ± 26 360–290 cal. bc (25.2%)
240–90 cal. bc (69.5%)
70–60 cal. bc (0.7%)
Kirkburn warrior CA26 OxA-17510 Bone –20.94 2161 ± 27 360–270 cal. bc (42.4%)
260–110 cal. bc (53.0%)
A3 Garrow:Layout 1 01/09/2010 15:03 Page 87

table 2 (continued)
Object Project Lab no. Material d13CRadiocarbon Date ranges
ID measurement BP at 95.4%
Latchmere Green mirror CA28a OxA-17287 Bone (crem.) –18.72 2145 ± 30 360–280 cal. bc (25.0%)
240–50 cal. bc (70.4%)
Latchmere Green mirror CA28b OxA-18355 Bone (crem.) –22.27 2169 ± 27 360–160 cal. bc (93.3%)
140–110 cal. bc (2.0%)
Latchmere Green mirror CA28c OxA-18356 Bone (crem.) –22.08 2162 ± 28 360–270 cal. bc (43.0%)
260–110 cal. bc (52.4%)
Llyn Cerrig Bach chariot pole CA44 OxA-17512 Wood –29.29 2247 ± 27 400–340 cal. bc (30.8%)
320–200 cal. bc (64.6%)
Llyn Cerrig Bach spear CA29a OxA-17436 Wood –26.13 2101 ± 30 210–40 cal. bc (95.4%)
Llyn Cerrig Bach spear CA29b OxA-17437 Wood –26.28 2109 ± 30 200–40 cal. bc (95.4%)
Owslebury warrior CA30a OxA-17288 Bone –19.66 2150 ± 29 360–280 cal. bc (30.4%)
240–90 cal. bc (64.1%)
80–60 cal. bc (0.9%)
Owslebury warrior CA30b OxA-17289 Bone –19.63 2120 ± 28 350–320 cal. bc (3.3%)
210–50 cal. bc (92.1%)
Pentuan tankard CA31 OxA-17366 Wood –24.55 2096 ± 27 200–40 cal. bc (95.4%)
Salisbury (Netherhampton) CA43 OxA-17511 Bone –21.42 2272 ± 28 400–350 cal. bc (50.9%)
hoard 300–210 cal. bc (44.5%)
Snettisham (Torc 241, Hoard F) CA33 OxA-18219 Wood –26.56 2181 ± 29 370–160 cal. bc (95.4%)
Snettisham (Torc SN/NZ, CA46 OxA-18222 Wood –27.43 2175 ± 31 370–160 cal. bc (93.8%)
Hoard L) 140–110 cal. bc (1.6%)
Snettisham (ingots in Hoard M) CA47 OxA-18223 Wood –26.12 1931 ± 29 1–130 cal. ad (95.4%)
A3 Garrow:Layout 1 01/09/2010 15:03 Page 88

Trawsfynydd tankard CA39 OxA-17439 Wood –23.45 2270 ± 30 400–350 cal. bc (46.9%)
310–200 cal. bc (48.5%)
Welshpool grave assemblage CA40 OxA-17440 Wood –19.18 1915 ± 30 1–140 cal. ad (92.9%)
150–170 cal. ad (1.3%)
190–210 cal. ad (1.2%)
Suspect dates
Asby Scar sword/scabbard (ash) CA01 OxA-17581 Wood –25.45 7635 ± 60 6610–6390 cal. bc (95.4%)
Asby Scar sword/scabbard (lime) CA45 OxA-17582 Wood –24.48 4745 ± 32 3640–3380 cal. bc (95.4%)
Mortlake dagger CA38 OxA-17367 Wood –26.34 1101 ± 26 880–1000 cal. ad (95.4%)
Snettisham (Torc 6, Hoard L) CA35 OxA-18220 Wood –28.22 2500 ± 31 790–510 cal. bc (95.4%)
Stanwick hoard CA37 OxA-18221 Charcoal –26.37 558 ± 25 1310–1430 cal. ad (95.4%)
Stanwick scabbard CA38 OxA-17438 Wood –25.78 3499 ± 32 1920–1740 cal. bc (95.4%)
Previous dates
(not from this project)
Balmaclellan hoard n/a OxA-6742 Cloth –24.2 1835 ± 55 340–50 cal. bc (95.4%)
Bryher mirror and sword/scabbard n/a OxA-12095 Bone –19.0 2098 ± 27 200–40 cal. bc (95.4%)
Burnmouth spoons n/a GrA-27301 Bone ? 2095 ± 35 340–330 cal. bc (0.4%)
210–20 cal. bc (94.2%)
10–1cal. bc (0.7%)
Chertsey shield n/a OxA-993 Wood ? 2300 ± 80 750–680 cal. bc (4.5%)
670–640 cal. bc (1.4)
600–160 cal. bc (89.6%)
Ferrybridge chariot burial n/a NZA-20494 Bone –20.3 2185 ± 35 380–160 cal. bc (95.4%)
Ferrybridge chariot burial n/a NZA-19423-5Bone –20.3 2168 ± 20 360–280 cal. bc (53.5%)
260–160 cal. bc (41.9%)
A3 Garrow:Layout 1 01/09/2010 15:03 Page 89

table 2 (continued)
Object Project Lab no. Material d13CRadiocarbon Date ranges
ID measurement BP at 95.4%
Ferrybridge chariot burial n/a NZA-20496 Bone –22.0 2330 ± 35 520–350 cal. bc (90.5%)
290–230 cal. bc (4.8%)
Ferrybridge chariot burial n/a NZA-21727-9Bone ? 2260 ± 20 400–350 cal. bc (47.1%)
300–210 cal. bc (48.3%)
Garton Station chariot burial n/a SR-6928 Bone –22.7 2145 ± 20 360–300 cal. bc (21.0%)
230–220 cal. bc (0.3%)
210–100 cal. bc (74.0%)
Kirkburn chariot burial n/a SR-6930 Bone –19.8 2210 ± 20 370–200 cal. bc (95.4%)
Kirkburn warrior n/a SR-6929 Bone –22.1 2175 ± 20 360–280 cal. bc (58.2%)
260–170cal. bc (37.2%)
Newbridge chariot burial n/a GO-107510 Wood ? 2350 ± 50 750–640 cal. bc (8.9%)
590–350 cal. bc (81.9%)
290–230 cal. bc (4.6%)
Newbridge chariot burial n/a GO-107512 Wood ? 2365 ± 40 740–690 cal. bc (5.8%)
670–650 cal. bc (1.0%)
550–370 cal. bc (88.6%)
Wetwang Slack chariot burial 1n/a OxA-14112 Bone –19.4 2152 ± 28 360–280 cal. bc (32.6%)
260–90 cal. bc (62.8%)
Wetwang Slack chariot burial 2n/a OxA-14113 Bone –20.2 2227 ± 30 390–200 cal. bc (95.4%)
Wetwang Village chariot burial n/a OxA-11993 Bone –20.6 2151 ± 21 360–290 cal. bc (30.3%)
230–110 cal. bc (65.1%)
A3 Garrow:Layout 1 01/09/2010 15:03 Page 90

do not have particularly marked distributions of material labelled art; the nearest
concentrations of Celtic art are found further east or south in Champagne and the
Ardennes.
As well as spatial complications, there are temporal ones. Celtic art is seen to arise
on the Continent in La Tène A, where it is connected to the coming into being of
the Celts, a much disputed ethnogenesis. The change from Hallstatt D to La Tène A
is freighted with significance, seeing the end of a world with links to the Bronze Age,
in the former case, to a new society, material culture and possibly languages which
formed the basis for so-called Celtic populations of today, but also new modes of
encounter with the Mediterranean that were ultimately to end with the expansion of
Rome. Celtic art was a key piece of evidence in the latter encounter and has been
viewed, from Jacobsthal (1944) onwards, as deriving also from influences from the
Mediterranean. The Reinecke chronology of Hallstatt and La Tène has never worked
well for Britain. This is partly because insular art is seen to be much later than its
Continental cousin, which flourished in La Tène A and B (c.450–250 bc), whereas the
British material has often been seen to date mainly from the second and first centuries
bc. The question of links with Continental chronologies is one we shall address here.
In order to do so we need to look both at Continental chronologies and at links with
Britain.
Although material designated Hallstatt D is not well represented in Britain, some
broader Continental trends are reflected, such as daggers replacing swords. As
Haselgrove (2001, 41–42) notes, towards the end date for Hallstatt D bow brooches
are found together with painted pottery with pedestal bases, which might show
influence from the Marne region. Like northern France and southern Belgium, Britain
also receives a smattering of Greek and Etruscan imports (Bradley and Smith 2007).
Early La Tène brooches (c.450–325 bc) are known in some numbers, but already these
have local features, such as hinges rather than Continental style springs. Early
La Tène swords with laddering, such as that from Orton Meadows, made an
appearance (Stead 2006). The period with apparently least Continental influence is
that between 325–150 bc (Haselgrove 2001, 43) when highly decorated bowls, jars
and saucepan pots in Britain have few parallels across the Channel. Brooch styles, most
famously the involuted series, are also restricted to Britain. The main shift in
chronology in recent years has been the earlier dates for such elements as coins and
wheel-turned pottery from northern France, many of which are also found in Britain
(Haselgrove 1999). With Gallo-Belgic A types now dating to 175 bc or possibly even
earlier in both areas (and with the very earliest gold coins stretching back to around
300 bc in both areas), many other aspects of our chronological schemes have been
pushed back in time. Not least are the earliest dates for wheel-turned pottery, which
may start in the earlier second century in some areas, although not until much later in
others.
The distribution of coins from the early second century until the end of the first
century bc shows very close connections between the Picardy region and the region
comprising Kent, Hertfordshire, Essex and southern Suffolk. One might also make the
same point about fibulae, although comparisons between northern France and south-
eastern England are less well-worked through. These distributions are also mirrored
91dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 91

by those of early Mediterranean imports, such as Dressel IA and IB amphorae
(Haselgrove 1996).
It is generally accepted that Celtic art arose in Continental Europe early in La Tène
A(c.450 bc) and was then introduced to Britain either through the movement of
people, as in older views up to and including those of Hawkes, or through the transfer
of materials and ideas, possibly through limited individual links and movements. In
trying to understand the start and development of Celtic art in Britain we need to
consider in outline recent Continental dates and debates.
Important new dates have been derived from timbers framing graves and from
wooden artefacts in the Austrian salt-production site of the Dürrnberg (Sormaz and
Stöllner 2005). When combined with the dendrochronology dates from the site of
Hallstatt (and elsewhere) a continuous set of curves was constructed covering the
period 815–124 bc (Sormaz and Stöllner 2005, 365–67), the full implications of which
will take some time to work through. Using this new curve Sormaz and Stöllner now
date the start of La Tène A1to 460 bc, with fully developed La Tène A forms and
accom panying Mediterranean imports coming in a decade or two later. This redating
is moving in the opposite direction to that of Trachsel (2004), who relies not just on
a redating of the Greek imports but also a dendrochronology date from the north
entrance of the Heuneburg to push the start of La Tène A back before 500 bc. How -
ever, the stratigraphic context of the wood dated at the Heuneburg has been the
subject of a critique by Brosseder et al. (2003) who argue against the reliability of this
date. Sormaz and Stöllner (2005, 370–72, Abb. 11) identify a series of areas that created
early La Tène A material forms, including the Champagne region, the Ardennes, the
middle Rhine, Western and Eastern Alpine areas, the middle Danube, Bohemia,
north-east Bavaria and northern Italy. In all of these regions, similar material forms are
found from around 450 bc, showing wide-ranging contacts but also subtle regional
variations.
The end of Celtic art is less clearly defined. Celtic Art is a polymorphous category,
but it is generally felt that much that defines it on the Continent comes to an end with
the end of La Tène C, so that decorated torcs, arm and leg rings, helmets, gold work,
complex fibulae and belt hooks or decorated swords finish around 150 bc. After that,
in the Late Iron Age world of La Tène D, coins, statuary, animal and human figurines
and bronze buckets follow some old traditions, but also take form and decoration in
new directions.
One of the difficulties of correlating British and European art styles is that much of
British art has traditionally been seen to start just when the Continental art stops (see
the dates given in the figure captions of Megaw and Megaw 2001). Using the
Reinecke chronology, Continental art in the areas mentioned above was at its peak
from La Tène A to C, whereas the British material saw some brief flourishes in La
Tène B, but only really got going in later La Tène C and D. Influences giving rise to
Celtic art in Britain supposedly come from the Continent (although it is very hard to
identify many actual imports), making it even harder to understand a possible
substantial time lag between the Continental developments and those in Britain. It is
to the dating of Celtic Art in Britain we now turn in order to try and throw some
light on some of these issues.
92 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 92

dating celtic art in britain
The main framework for dating British Celtic art comes from Jacobsthal’s (1944) work
on Continental material, which he started to amend for the British material but never
managed to complete. Various forms of Jacobsthal’s scheme have been developed by
others, most notably Jope, whose major work Early Celtic Art in the British Isles (2000)
was initially conceived in co-authorship with Jacobsthal. These labyrinthine works
have much of value on chronology in them (see Jope’s chronological chart for a
distillation of his thoughts — 2000, xii ff). But in many ways the most useable typo -
logical dating scheme is that put forward by Stead in two relatively brief works (Stead
1985a; 1985b). However, as Stead is first to acknowledge, the chronology, especially
for the early part of the sequence is uncertain: ‘The sequence of Early Celtic Art in
Britain relies on stylistic and typological factors founded on some tenuous Con tinental
parallels, for there are few associated finds’ (Stead 1985b, 27). The tenuous nature of
these parallels has led a number of British workers to eschew Jacobsthal’s scheme,
including Fox (1946; 1958) and Jope (2000). However, we have retained use of them
here as a convenient shorthand, to be changed and nuanced as we will attempt below.
Jacobsthal defined three styles for the Continent with a possible fourth for Britain.
These start with the Early Style, characterized by the use of palmettes and other
vegetal ornament thought to be derived ultimately from Greek art, but combining
these with continuing aspects of Hallstatt forms of decoration. For Britain, Stead ten -
ta tively attributes pieces such as a bronze chape on a dagger sheath from Wandsworth
and the headpiece from Cerrigydrudion to the Early Style (which Stead labelled Stage
I; Stead 1985a, 16–17, figs 15–17). Jacobsthal’s Waldalgesheim Style (Stead’s Stage II)
is seen as being more commonly present in Britain and is characterized by a flowing
wave tendril of linked triangles with rounded edges. This is found on diverse forms
such as the Brentford horn cap (Stead 1985a — inside front cover), a scabbard from
Standlake and an antler handle from Fiskerton (Stead 1985a, 18, fig. 19; Field and
Parker Pearson 2003). Jacobsthal’s following style, with its two variants of Plastic and
Sword Styles (Stead’s Stage III), is rare in Britain, but he argues that it was important
in giving rise to the disputed Insular Style (Stead’s Stage IV — although it is not
known whether Jacobsthal approved of the term or not) into which some of the better
known items of British Celtic art might be placed, including a large number of
scabbards (with the famous piece from the River Witham a striking example — Stead
1985a, fig. 24). Key motifs of the Insular Style include a rather loose form of the
Waldalgesheim tendril and the half-palmette. Stead, developing a scheme first put
forward by de Navarro (1952), which extended Jacobsthal’s styles to encompass the
full range of the British material, added Stage V to include many of the best-known
pieces of British Celtic art, encompassing torcs, early mirrors and the plaque from Llyn
Cerrig Bach. Stage V work is asymmetrical and curvilinear, as well as often having a
three-dimensional aspect to it with decoration raised above a surface. The use of
hatch ing is now common to create a variable set of textures, as is the termination of
ten drils in so-called trumpet voids. Engraving, chasing and repoussé work were all
used to achieve a variety of effects. The set of British styles is completed with VI,
which designates material that has an ancestral connection to earlier Insular styles but
93dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 93

also has a marked Roman influence in the use of new metals, such as brass, the greater
use of glass inlay and more formally geometrical motifs.
In terms of the dates assigned to British pieces on typological grounds, Stage II is
thought to be found in the decades centred on 300 bc, III and IV belong to the third
century, with IV perhaps extending into the second century, while V may itself start
before 200 bc. Motifs continue up to and possibly beyond the Roman conquest with
Stage VI, although there are significant changes in the first century ad, with more
sym metrical designs appearing. In the latter part of the Iron Age contacts with the
Roman world sometimes allow dating within a few decades, but before that British
metalwork can usually be dated no closer than within half centuries, and in many cases
even a century or two.
The typological basis for these schemes has been criticized most clearly by Mac -
donald (2007a). He has pointed out the existence of pieces that may have decorations
in more than one style, such as the terminal of the Clevedon torc, with a Stage V
triskele and a vegetal scroll that some might attribute to Stage 1(ibid., 332). He also
con siders problems of assigning dates to Styles (it should be noted that Stead has
termed his categories both Styles and Stages at different times): ‘The recognition that
the inception of Style V goes back to at least the third century bc, combined with the
identification of Style IV and V motifs on items decorated in “earlier” La Tène styles,
implies that the sequential character of the insular art classification may be largely
invalid’ (ibid., 333). Macdonald goes on to wonder whether some of the differences
in motifs may be geographical and not chronological, an idea that has been around at
least since Fox (1958) attempted to recognize regional schools in British Celtic art.
objects chosen for dating and rationale for that choice
The background to this dating programme lies in a three-year project (October
2005–September 2008) funded by the Arts and Humanities Research Council to look
at the social, political and aesthetic implications of later Iron Age and Roman metal -
work (The Technologies of Enchantment — AHRC reference number 112199; see Gosden
and Hill 2008 for a discussion of this project). A major element of this project was the
compilation of a database of all Celtic art found in Britain (this can be accessed at:
http://www.britishmuseum.org/research/research_projects/technologies_of_enchant
ment.aspx). The database will be updated each year, but in 2008 it incorporated 2580
objects — arm rings, torcs/collars, swords/scabbards, shields, horse gear, mirrors,
tankards, spoons, bowls/cauldrons/buckets, fire dogs, spoons, etc. It also includes
infor ma tion about their recovery contexts, the materials they were made from etc.
(see Garrow 2008 for a more detailed discussion of the database). It is impossible to
under stand the social role of objects and the changes in their forms of production, use
and deposition without a proper dating framework. As we have discussed, the dating
of items such as swords, shields, torcs, tankards and horse gear has in the past been
based almost entirely on stylistic attributes, whose dates are derived, by inference,
from Continental parallels (e.g. Jope 2000). A key aim of the Technologies of Enchant -
ment project has been to reintegrate Celtic art back into broader discussions of Iron
Age material (Garrow and Gosden in prep.). The effective dating of this material was
seen as a key element of that process.
94 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 94

In order to achieve a greater understanding of the dates at which particular objects
or object types were produced and/or deposited, we undertook a two-phase dating
audit of Celtic art. The first phase involved an investigation of every single object
within the project database thought to have been found in a potentially datable con -
text. In this case, the original site report relating to each of those objects was con -
sulted. If it was determined that the object had been found alongside datable pottery
or coins, this date was also assigned to the Celtic art object. In this way, 535 objects
(20.7% of the database) were given a contextually determined date. The second phase
was the radiocarbon dating programme that forms the main focus of this paper.
In choosing our samples (see Appendix 1for more details) we have picked a range
of different artefact types, including chariots and horse gear, swords and scabbards,
mirrors, buckets and tankards, shields, torcs, metalworking debris, and other finds
includ ing spoons and personal ornaments. We have also tried to ensure good coverage
across Britain (Illus. 1). It should be noted here that the National Museum of Scot -
land’s radiocarbon dating policy has ensured that most datable finds have been
sampled, and so we did not need to include any Scottish finds within our study; they
are, how ever, included in broader dating models and general discussions. As far as it
was pos sible, we also attempted to date objects from the full typological span of Celtic
art (Stages I to VI).
Ultimately, our sampling strategy was expedient and wide-ranging. Despite a
detailed audit of all potential samples, and close examination of museum collections,
relatively few objects were found that had datable materials within them or came from
closed contexts containing datable material. The sampling strategy certainly sought to
find datable material from most of the well-known and important objects. Unfor -
tunately, several of the objects we would very much have liked to have dated proved
to be undatable (see Appendix 2). Among these were objects that were too fragile to
sample effectively (e.g. the Kirkburn sword and the Fiskerton bone-handled tools),
those which had been found with burials but from which the human bone had subse -
quently been lost or become untraceable (Birdlip, Newnham Croft, Arras, etc.), and
those classic pieces which we hoped might, but unfortunately did not, have any datable
material at all (the Battersea shield, the Witham shield, the Waterloo helmet etc.).
In addition to these artefacts, a number of those we were able to date produced
spuri ous or suspect results. In some cases, this was likely to have been a consequence
of previous conservation treatments having contaminated the material (e.g. the Asby
Scar sword and the Stanwick scabbard). In others these results may have been due to
intrusive material having contaminated the deposit (e.g. the Stanwick hoard, which
gave a date of 1310–1430 cal. ad (95.4%), and the Snettisham torc G17 which gave a
post-ad 1950 date). In some cases it could possibly have been a consequence of the
object’s own life history (the Thames dagger is almost certainly Iron Age in date, but
could conceivably have been resheathed during the first millennium ad), and in others
it was impossible to explain (Snettisham torc L6may date to between 800 and 500
cal. bc but this seems somewhat unlikely).
Another key objective of the project was to sample a range of the burials which
pre-date the first century bc (and thus lie beyond the range of accurate dating by
association with other material culture) and which contained decorated objects. These
95dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 95

96 dating celtic art
illus. 1
a: Locations of all find spots of Celtic art in the Technologies of Enchantment project database, and b: those submitted for dating
A3 Garrow:Layout 1 01/09/2010 15:03 Page 96

97dating celtic art
illus. 2Distribution of all dates considered within this paper
A3 Garrow:Layout 1 01/09/2010 15:03 Page 97

include the chariot (and related) burials from east Yorkshire for which there is no
secure independent dating sequence (a number have already been dated in recent years
as a result of other projects, see Table 1). Other burials which the project sought to
date included anomalous burials such as those from Deal and Grimthorpe known as
warriors, both of whose typological dating was previously uncertain (Stead 1968;
1995, but see 2006).
Ultimately, this programme has included a wide range of object types and covers
the whole chronological range of early Celtic art in southern Britain. It includes
objects that were thought to be early (e.g. the Bermondsey dagger and the Cerrigy -
drudion headpiece) and late (e.g. the Brecon grave group and the Boverton collar). It
includes examples of object types with no previous secure dates (‘horn caps’ and
spoons) and redates several objects or contexts previously dated, either in order to
ensure comparability of results (e.g. Wetwang Slack, Kirkburn etc.) or because the
previ ous dates were conventional and thus had given large date ranges (e.g. the Garton
Slack mirror and Gussage All Saints Pit 209). When combined with the other recent
dates, our study provides for the first time a large sample of radiocarbon dates from
which to reassess the chronology of early Celtic art in Britain.
Associations between the materials dated and the objects we wanted to date were
of two types: firstly, from the object itself (e.g. the piece of rawhide once attached to
the Cerrigydrudion headpiece), and, secondly, from items in direct association with
the object of interest (in almost every case these were articulated bones from grave
con texts containing the object). Table 1and Appendix 1provide details. All deter -
mina tions were from single entities and not from massed or combined material.
Wher ever possible they were taken from short-growth materials. However, in a few
cases (notably tankards) we took the decision to obtain dates from what may well have
been old wood; while far from ideal, this at least gave us a terminus post quem for these
classes of objects. In all cases where it was important to establish the age, species or
type of material we were dating, appropriate specialist analysis was undertaken prior
to sampling.
radiocarbon dating
The new radiocarbon dates for this project were measured at the Oxford Radiocarbon
Accelerator Unit. Most samples were pre-treated using standard acid-base-acid
methods described in Hedges et al. (1989) while the cremated bone samples were
pre-treated using the acid digestion methods discussed in Naysmith et al. (2007). All
samples were then graphitized (Dee and Bronk Ramsey 2000) and AMS dated (Bronk
Ramsey et al. 2004).
the dates and their calibration
We attempted to obtain forty-four determinations in all. Four of these failed to give
a result, while six gave age ranges that fell outside the Iron Age and Roman periods,
and so have not been considered further. We will base our discussion on the remain -
ing thirty-four dates deemed to have been successful, in combination with those dates
which had been obtained by others recently enough to make them comparable with
98 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 98

our own (details of these are provided towards the bottom of Tables 1and 2). In the
following section, we will look at these dates in rough date order, then make some
com ments on the classes of objects and types of deposition, before turning to the
fraught issue of style.
All dates were calibrated with reference to the IntCal04 calibration curve (Reimer
et al. 2004) using the OxCal calibration program (Bronk Ramsey 2009a). A plot of all
these dates can be seen in Illus. 2.
It is important to point out at this stage that problems with the radiocarbon calibra -
tion curve ensure that dates for much Iron Age material are often rather less precise
than one would hope. The plateau in the curve between c.800 and 400 bc is well
known, causing considerable problems in that phase of the Iron Age. However, the
majority of our dates were expected to, and indeed did, fall within the last four
centuries bc (c.400–1cal. bc), and so were not affected by this particular issue.
However, the deviations in the radiocarbon calibration curve nevertheless did have a
substantial effect on the accuracy of our dating programme, ensuring that the many
objects which dated to between c.350 and 200 bc would always be essentially
indistinguishable.
In addition to simple calibration, Bayesian analysis was carried out using the
methods outlined in Bronk Ramsey (2009a). Three different models were considered
together. In all of them we treated the calibrated dates for most samples as the dates
for the use of the objects, the exception being those samples which were of wood.
Because the growth of the wood is quite likely to be significantly older than the use
of the object in question, all wood dates were treated as termini post quem for these
samples; the method employed for this was that recommended for charcoal in Bronk
Ramsey (2009b). Three main models were used to interpret the data:
Model 1: in which all of the material was treated as being from a single phase.
Model 2: a three-phase model. Phase I (= Style 1), immediately followed by a phase
that includes Styles II–V. Phase VI (= Style VI) follows after, but we allow in the
model for a gap between this phase and the preceding material.
Model 3: where each Style is a separate phase, following one after another, again
allowing for a hiatus before Phase/Style VI.
The results of this modelling exercise are shown in Illus. 3and the associated intervals
in Illus. 4. A number of interesting points are clear from this analysis. Each model
suggests that the samples dated all range from approximately 400 bc to some time in
the second century ad. Models 2and 3also suggest a potential hiatus before the final
Stage VI. Model 2suggests that it is possible that Stages I–V are all quite short-lived,
possibly being either limited to the late fifth and fourth centuries bc or the third and
early second centuries. However, it is much more likely that these styles are current
over the entire late fifth to early second centuries. Model 3also suggests this longer
chro nol ogical span for Stages I–V. Part of the uncertainty here comes from the fact
that we only have one sample from Stage I. In Models 2and 3it is clear from the
limited data we have that Stage VI was most likely to have been a phenomenon of the
later first/earlier second centuries ad, although we cannot rule out a much longer
dura tion for this Stage, extending back to the pre-Roman period.
99dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 99

100 dating celtic art
illus. 3Results of applying three different Bayesian models to the radiocarbon dates
considered in this project
A3 Garrow:Layout 1 01/09/2010 15:03 Page 100

101dating celtic art
illus. 4Plot showing the duration for each of the phases in the models considered for Illus. 3
and the modelled hiatus before Phase VI in Models 2and 3
A3 Garrow:Layout 1 01/09/2010 15:03 Page 101

As a sensitivity test, we also looked at the results of not allowing for a hiatus before
Phase VI in the latter two models. This did not significantly change the results, except
obvi ously for the final phase itself. Given that there are only three samples in this final
phase, and one of these is wood, it is not possible to learn very much about the dura -
tion of this final phase anyway. However, Models 2and 3both indicate that it is likely
there was a substantial hiatus, as can also be seen by looking at the calibrated dates
directly.
Model 2is probably the most realistic as we have no reason to suppose that Stages
II–V were chronologically distinct. However, interestingly Model 3was consistent
(Amodel =72) with the radiocarbon dates and so cannot be ruled out entirely. We will
return to the possible implications of this modelling below.
the radiocarbon determinations compared with typological
dating (see Tables 2 and 3)
The earliest determination obtained was that from ash wood in the scabbard lining of
the Bermondsey dagger (see Table 1for bibliographic references to individual sites/
objects). The date range of 810–550 cal. bc overlaps closely with what would be
expected for a Hallstatt D dagger. The next earliest determination was that from raw -
hide which formed part of the Cerrigydrudion headpiece. The decoration was dated
typologically by Jacobsthal to c.300 bc. Jacobsthal’s definition of the Waldalgesheim
Style, not a model of clarity, once allowed Jope to classify Cerrigydrudion as Wald -
algesheim art but it lacks the distinctive scrolls and tendrils of the other Stage II pieces.
Instead it belongs to Verger’s ‘Premier Style Continu’, ancestral to the Waldalgesheim
Style and thus Stage I, considered to be from the early fifth century to the second half
of the fourth. The radiocarbon date obtained confirms this date, pushing it back past
the date mooted by Jacobsthal and putting the Cerrigydrudion piece early in the Con -
tinental sequence.
A series of determinations have a wide span of possible dates from 400–200 cal. bc
(see above for discussion of the calibration curve during this period). The Salisbury
hoard had been placed at around 200 bc on typological grounds. The date obtained
— for the pit into which the hoard pit was cut — fits well with this suggestion. The
Trawsfynydd tankard has a handle whose symmetrical decoration is in keeping with a
date in the first century ad (i.e. Stage VI). The radiocarbon determination is much
earlier than this, almost certainly because old wood was used (it is worth noting in this
case that yew trees can live to be especially old). The Llyn Cerrig Bach chariot pole
date may also be influenced by the old wood problem; however, it is worth noting
that the three previously existing radiocarbon dates from associated animal bones there
have also produced similar date ranges (Macdonald 2007b, 168–69). However, the
nature of the deposit at Llyn Cerrig Bach is such that various objects do not need to
be con temporary and it is highly probable that chariots existed in the fourth and third
centuries bc, especially given the dates from Garton Station and Bury Hill discussed
below. Once again the date for the Deal spoons is arguably earlier than expected.
There is no reason to doubt this date, and a period of use prior to 200 bc becomes a
real possibility, so that evidence from Burnmouth, where the spoons were dated as
rather later (Fitzpatrick 2007), perhaps indicates a longevity of use for these objects.
102 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 102

Two dates from Garton Slack, which are largely overlapping between 380 and 190
cal. bc, reinforce Joy’s (2008a) impression that the east Yorkshire mirrors were early.
Fur ther indications of the early use of horse gear come from two determinations from
Bury Hill, where that from Pit 45, Layer 10 in particular indicates that bronze and iron
horse and chariot gear was in use in the fourth to second centuries bc. Two torcs from
Snettisham (Torc 241 from Hoard F and Torc SN/NZ from Hoard L) have identical
ranges of 370–160 cal. bc, which might potentially push back torc use and the
decorations of Stead’s Stage III/IV type found on them. Some of the very few
associations of artwork in hoards occur at Snettisham where two different art styles can
be distinguished, an earlier (Stage III/IV) and a later (Stage V). Two hoards contained
objects in both styles: Hoard L has one Stage III piece and four Stage V pieces; Hoard
F has four of Stage III/IV, three of Stage V and coins that may date from late in the
second century bc. The radiocarbon determinations arrived at here open up the pos -
sibility of a third century bc date for torcs and their accompanying coins.
The Grimthorpe warrior falls into the range between 360–160 cal. bc. The burial
has two pieces of decorated metalwork, a disc and scabbard. The disc is decorated with
repoussé lobes that create circular and fan-shaped voids. The scabbard has a chape
with openwork rungs featuring a Stage V bird-head and the chape-end is unique,
though typologically later than Kirkburn. On typological grounds, the burial is
thought to date to the second or first centuries bc. In combination with the radio -
carbon determination, this suggests a probable date early in the second century bc.
The Latchmere Green mirror, decorated with Stage V trumpet voids and bird-
heads, was found in a cremation burial; it was found with a pottery jar whose fabric
is said to be particularly common in the first half of the first century ad (Fulford and
Creighton 1998), and brooches that are likely to date to 75–25 bc. Three determina -
tions from the same sample gave ranges between the earlier fourth and the mid-first
centuries cal. bc. The latter part of this distribution fits relatively well with the
expected first century bc date of the burial (although see Appendix 1for a discussion
of the possibility that the bone had been curated).
Very similar spans of dates, between 360 and 100 cal. bc, were derived from an
important set of material which included the Deal, Kirkburn and Owslebury warriors,
the Kirkburn chariot burial (K5) and the evidence of metalworking at Gussage (Pit
209, Layer 12). Of the Yorkshire cemeteries sampled, Kirkburn has arguably the best
collections of metalwork. The scabbard in Burial K3has an open chape-end, a La
Tène B feature. The lower part of the front plate of the scabbard has a Stage IV design
and the upper part is a repair with a similar overall pattern but Stage V filler motifs.
As discussed above, the Deal warrior also has evidence for artefacts decorated with
both Stead’s Stages IV and V. Stead himself has been pushing back the date of these
Stages, possibly to the third century bc (see discussion below). This conclusion is
given further weight by our radiocarbon dates (which could even indicate some
currency as far back as the fourth century bc).
Two Yorkshire cart burials with involuted brooches had previously been radio -
carbon dated: Ferrybridge and Wetwang Village (see Table 2). The involuted brooch
is thought to have developed from the La Tène C brooch no earlier than the second
half of the third century bc, and the Ferrybridge and Wetwang brooches are late in
103dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 103

this typological sequence. Horizontal stratification in the Wetwang Slack cemetery has
confirmed the relative chronological validity of their typological sequence. In addition
to these burials, the cart burial (K5) at Kirkburn has linch-pins with Stage V decora -
tion, triskeles in relief with bird-head terminals and trumpet voids. An adjoining burial
(K6) in this small family group has a La Tène I hollow ring. Stylistically, the whole
group could belong in the third century bc, or not much later. Like the Kirkburn
linch-pins, the central terret from Garton Station (GS6, another cart burial) is decor -
ated with relief lobes and bird-heads, Stage V art. Stylistically, the burial could have
been said to belong anywhere between the late third century and the first century bc.
The resolution of the radiocarbon dates for these chariot burials is significantly affected
by the flatness of the calibration curve at that time. Nevertheless, the determinations
obtained suggest dates between the fourth and second centuries cal. bc for all four.
The Owslebury warrior has relatively close typological dating through the unique
copper-alloy shield-boss, which bears comparison with iron examples from Alesia and
Grave 59 at Lamadelaine (Collis 1973), and, more particularly, the belt hook which
can be linked with Continental examples of La Tène D1and D2, hence around 150–1
bc (Bataille 2001). The Owslebury burial was the only inhumation, and perhaps the
earliest burial, in a cemetery which is thought to have started in the first century bc
and extended into the second century ad (Collis 1968). While our radiocarbon results
could suggest the possibility of an earlier date for the warrior, it is important to note
that the later part does overlap with the expected first century bc date.
The other southern English burials to have been sampled are both from Deal. An
inhumation burial (X2), which included the pair of bronze spoons, was discovered in
a quarry in 1902. When the adjoining area was excavated in 1987–88 an inhumation
cemetery was revealed with grave goods including at least two La Tène C brooches
(one involuted) and fragments of La Tène D brooches. A date in the second century
or the first half of the first century bc seemed likely on typological grounds. However,
the radiocarbon date was rather earlier, suggesting a range between 380 and 200 cal.
bc. The other Deal burial (Grave 113 — the Deal warrior) has a rich collection of
grave goods: a decorated brooch, a sword in its decorated scabbard and three other
decorated artefacts. The brooch might be described as an oddity, belonging to Hull
and Hawkes’ Type 2B — ‘decorative forms, peculiar to Britain’ (the Newnham Croft
brooch belongs to the same type and is associated with an armlet with Stage II
Waldalgesheim decoration). Two other objects from the Deal grave, the suspension
ring and the crown, have decoration classified as Stage IV. The scabbard is a British
version of La Tène C and its decoration includes Stage V motifs. The umbo has Stage
V motifs arranged in a way that Fox would have classified as ‘a late piece’ (cf. the Old
Warden mirror). This grave was presumed to date to late in the third century or early
in the second century cal. bc on typological grounds, a suggestion which the radio -
carbon date broadly supports.
The date span obtained for the Aylesford bucket, of between 230 and 50 cal. bc,
fits comfortably with its expected first century bc date, especially when the fact that
old wood may have been used in its manufacture is taken into account. The old wood
problem is also likely to have ensured that the dates for the Kew and Pentuan tankards
are earlier than the period in which they were actually in use. The date ranges
104 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 104

obtained for these objects, of 350–50 cal. bc and 200–40 cal. bc, suggests a date for
their use towards the latter part of the Iron Age, and potentially into the Romano-
British period. The Ely ‘horn cap’ has a similar range of dates (180–30 cal. bc).
One set of objects produced dates that were rather later than expected. The silver
ingots in Snettisham Hoard M have dates falling within the Roman period, perhaps
indicating the continued use of this site post-Conquest. The dates from Welshpool,
Brecon and Boverton all fall in the first to second (or in one case, even early third)
centuries ad. The latter two of these contain material judged to be Stage VI in Stead’s
scheme, showing that the use and deposition of such material may have continued
rather longer than previously thought. This fact is particularly important given the
amount of material in the project database overall that can also be assigned to Stage
VI (Garrow 2008, 30–33).
the dates of some classes of material
Given the large probability distributions of many dates, it is hard to pin down precisely
when many types of artefacts were first made and used. However, the determinations
presented here open up the possibility of earlier dates for a range of important classes
of material. These include: horse and chariot gear, torcs, iron mirrors and spoons. The
distributions for the dates of many of these objects fall in large part before 200 cal. bc,
often stretching back before 300 cal. bc, making conservative estimates for the start of
horse gear or torcs, for instance, as first occurring within the second or even first
centuries bc most unlikely.
There are six sets of dates for horse and chariot gear: Bury Hill (two dates), Garton
Slack, Garton Station, Gussage All Saints metalworking (two dates), Kirkburn chariot
burial and Llyn Cerrig Bach. The others were on animal bone or charcoal from short-
lived wood. The Gussage All Saints dates we have obtained are older than those in the
original report (both on charcoal); the discrepancy may be due to refinements in pro -
cessing and dating samples. On the basis of the dates presented here one would
conclude that a fourth-century beginning for La Tène horse gear and chariotry is pos -
sible and that both were certainly common in the south of Britain as well as the north
before 200 bc. It is worth noting that these results fit very well with those obtained
from other recently dated chariot burials (see Table 2) at Ferrybridge (375–155 cal.
bc), Newbridge (520–370 cal. bc) and Wetwang Village (355–110 cal. bc).
Torcs are another key class of artefacts that may actually pre-date 200 bc (at least
those buried at Snettisham). Dates were obtained from Hoards F, L and M; in two
cases these were young wood charcoal and in one, limewood string. The dates from
Hoards F and L were almost identical, with a main probability distribution between
370 and 160 cal. bc, so that even a conservative reading of this span would put them
in the early second century bc. The date for Hoard M, in the first century ad,
indicates that deposition may well have continued into the Romano-British period.
The date from the Salisbury hoard was approximately what was expected. The
miniature shields in the hoard were probably similar in form to that from Grave 112
at Mill Hill, Deal which also has a span of dates between the fourth and second cen -
turies bc (Stead 1995). It is also worth noting that the Chertsey shield was dated to
between 750 and 165 cal. bc (Hedges et al. 1988).
105dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 105

One intriguing early date was that from Grave X2at Mill Hill, Deal (containing
the spoons) as this had a distribution in the fourth and third centuries cal. bc. We
know little about the dates of spoons, but most were thought to be later. The previ -
ously stated expectation for the Mill Hill spoons was second century or first half of the
first century bc (Fitzpatrick 2007, 294). A burial with spoons at Burnmouth was
recently dated to 335–5cal bc. The most likely date for the spoons from Pogny/La
Chausée-sur-Marne, accompanying an inhumation burial, is between 280–220 bc
(Fitzpatrick 2007, 294). Some spoons were thought to be earlier on grounds of their
decoration, such as the find from Andover or that on the pair from Weston-super-
Mare, North Somerset, suggesting to Fitzpatrick (2007, 293) dates between the fourth
and third centuries bc. The date reported on here from Mill Hill fits into the earlier
part of this overall range, confirming Fitzpatrick’s conclusion that spoons appear
‘perhaps in the fourth or third century bc, or slightly earlier, and continued into the
Late Iron Age’ (ibid., 294).
The Pentuan, Trawsfyndd and Kew tankards all have early dates. It is very likely
that the age of the yew used for the staves has influenced these results significantly.
One class of deposition, rather than artefact, which is intriguing, is that of the so-
called warrior burials, represented here by Deal, Grimthorpe, Kirkburn and
Owslebury (two dates). Grimthorpe has been dated on an associated bone point, the
others on human bone. For all these burials the spans we have encompass the fourth
to second centuries bc; only for Owslebury is there even the possibility of a first-
century bc date. The prior expectations for the Deal and Kirkburn burials were for
third- or second-century dates. For Grimthorpe and Owslebury there were expecta -
tions of a date in the first century bc. Once again the spans of the dates are broad, but
they do open up the possibility that warrior burials may also be rather earlier than we
have thought hitherto.
The dates of mirrors have been much debated (Joy 2008b). The probability dis -
tribu tion for the iron mirror from Garton Slack in the fourth and third centuries bc
helps confirm the early occurrence of these types in east Yorkshire, with no evidence
of mirrors this early elsewhere in Britain.
‘Horn caps’ are one of the more debated types as it is not agreed what these were
used for, nor how old they were. The Ely horn cap’s distribution encompasses the
earlier second to later first centuries, which occasions no great surprise.
A number of general conclusions can be made at this stage, although they are all by
no means certain. The first is that important classes of artefacts such as chariot and
horse gear, torcs, iron mirrors, hide-shaped shields and spoons might have been in use
between 400–300 bc. If this were true, they fall within La Tène B in the Reinecke
chronology, making them contemporary with the early high point of Continental
Celtic art. It must immediately be emphasized that the form and decoration of these
objects was peculiar to Britain, so we are looking at parallel developments and general
influ ence, rather than the British material directly copying that on the Continent.
Secondly, the latest types of collars, mirrors and horse gear have a much longer
currency than their equivalents on the Continent, lasting well into the Romano-
British period. This indicates that the nature of metalwork within Romano-British
culture was quite different from that elsewhere in the empire.
106 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 106

dating the celtic art decorative ‘styles’
Previous discussions of dates have focused not just on classes of artefacts, but even
more fully on the types and stages of decoration. Stead’s scheme — of six successive
phases, ‘styles’ or ‘stages’ — has been discussed above and it is to this we turn when
think ing about the implications of our dates for the understanding of stylistic change
(see Table 3). Two aspects of Stead’s scheme appear to be confirmed, albeit on the
basis of a small number of determinations. Cerrigydrudion, long held to be one of the
earliest extant pieces in the British sequence (Stead 1982) is confirmed in this position
by our results, dating from the beginning of the fourth century and possibly even the
end of the fifth. At the other end of the sequence, Stage VI pieces were, in the main,
dated to the first century ad or later.
For Stages III to V, however, the dates were much less supportive of an overall
succes sive sequence. Some objects with Stage V decoration (such as the Garton
Station linch-pins, the Garton Slack chariot burial, aspects of decoration on the Deal
warrior’s equipment and the berried rosettes, triskeles and bird’s heads on the linch-
pins, terrets and strap unions of the Kirkburn chariot burial, K5) were given dates
prior to the first century bc. These dates covered approximately the same time spans
as those assigned to the Snettisham torcs (Hoards F and L) or the Kirkburn warrior,
which had aspects of Stage IV decoration. As Stead himself notes, individual aspects
of the Deal warrior’s equipment might be assigned to different Stages (see also
Macdonald 2007a) — the sword has in itself aspects of both La Tène B and La Tène
C swords. Trumpet voids are found on both the top of the scabbard in repoussé and
on the open-work motifs of the shield. These have been seen as characteristic of Stage
V and hence of the first cen tury bc. However, it has also been suggested on the basis
of stylistic analysis that some of the Deal artefacts might be as early as the third century
bc (Stead 1995, 95). We can see from the radiocarbon dates that the latter appears to
be an accurate date.
On the basis of the dates presented and analysed through Bayesian modelling here,
it is hard to separate Stages III to V. While it might be argued that this is partly a
function of the relatively small number of determinations and the nature of the
material we have been able to date, we are nevertheless tentatively inclined to con -
clude that decorated metalwork in Britain between the fourth and early first centuries
bc did not follow a recognizable progression of styles. While Stage II probably did first
appear at an earlier date than Stage III, for example, and Stage IV seems to have
appeared earlier than Stage V, they were not strictly successive — i.e. Stage IV did
not replace Stage III and so on. Rather, we would suggest that there was a gradual
accum ulation of motifs, and while some were earlier and others later, both could be
combined together on later pieces. This finding accords with Collis’s recent careful
treatment of chronologies on a European-wide scale in which he says that a horizon
is marked by the beginning of an artefact or attribute type but may have no clear end
(Collis 2008). By contrast, the forms of some objects, notably fibulae and swords, do
change in a regular and recognizable manner (Haselgrove 1997; Stead 2006). In this
case there is, therefore, an interesting disjunction between changes in form, which had
some regularity, and changes in decoration, which did not.
107dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 107

108 dating celtic art
table 3 Comparison of previous, stylistically derived dates with calibrated radiocarbon dates obtained as part of this study
Site Artefacts and materials Motif Technique Stage Previous Our date
date
Bermondsey Dagger – cast and sheet bronze Ha. D C8–C6
Cerrigydrudion ‘Headpiece’ – sheet bronze Palmettes, scrolls, Engraving, hatching I C4C5–early C4
hatching
Salisbury hoard Miniature shields, cauldrons etc. Lobes, triangular Engraving, hatching V? C3/2C4–C3(tpq)
– sheet bronze motifs, trumpet voids
Trawsfyndd Tankard – wood, cast metal Triskele, S-shapes Casting VI C4–C3
handle and sheet metal body and trumpet shape
Llyn Cerrig Bach Wood, bronze C4–C3
chariot
Deal spoons Bronze – sheet and cast C4–C3
Garton Slack mirror Iron C4–C3
Garton Station Iron tyres, nave hoops, horse bits; Terrets – lobes, Cast-on decoration V C3C4–C3
cart (GS6) one bronze and iron terret, four trumpet shapes,
bronze terrets and one linch pin berried rosettes
Bury Hill P45/L10 Bone and bronze cheekpiece, iron Circles Stamping C4–C2
bridle bits, nave bindings, split ring,
cleat and fragment
Snettisham, Torc 241,III–IV C2–C1C4–C2
Hoard F
Snettisham, Torc III–IV C4–C2
SN/NZ. Hoard L
Garton Slack Cart Five terrets, bronze and wood V C4–C2
Burial XI ?whip pommel, three-link horse
bit, two strap rings
Grimthorpe warrior C2C4–C2
Latchmere Green Sheet and cast bronze Triskeles, peltae, Engraving V Late C1C4–C1(3dates)
mirror circles, basketry
A3 Garrow:Layout 1 01/09/2010 15:03 Page 108

109dating celtic art
Deal warrior Sword, scabbard, ‘crown’, shield Cusps, lobes, S-shapes, Casting, repoussé, II/IV–V C3/2C4–C2
binding and openwork decoration, wave tendrils, trumpet openwork
brooch, suspension ring, strap end voids
Kirkburn warrior K3Iron sword in composite scabbard, Cusp, lobe, tendril Enamelling, IV–V C3C4–C2
three iron spearheads engraving, cast
Gussage, Pit 209, Clay moulds C1C4–C2
Layer 12
Owslebury warrior Leaf-shaped iron spearhead, iron V C2C4–C2
sword in wooden scabbard
Kirkburn chariot K5Iron tyres, bronze nave hoops, Berried rosettes, trisceles, Cast V C4–C1
bronze and iron linch pins, bronze bird heads, lipped terrets,
terrets and mini-terrets, bronze and dots
iron horse bits, bronze strap unions,
toggles, mail coat
Aylesford bucket Bronze – sheet and cast, wood Animals, peltae, Repoussé, cast C1C3–C1
S-motifs, lobes
Kew tankard Bronze – sheet and cast, wood Circles, whirligigs Cast C3–C1
Llyn Cerrig Bach Iron spearhead, wood C2–C1
spear
Pentuan tankard Bronze – sheet and cast, wood C2–C1
Ely horn cap Bronze cast Tendril C2–C1
Snettisham ingots, C1–C2ad
Hoard M
Welshpool Ewer, patera, piece of cauldron, Cast, forged C1–C2ad
bucket escutcheon (all bronze).
Fire dog, stand on tripod base – iron
Brecon grave Mirror, mini-terrets, toilet set Sheet, cast VI C1–C2ad
assemblage
Boverton Collar and bracelets Cast VI C1–C2ad C1–C3ad
A3 Garrow:Layout 1 01/09/2010 15:03 Page 109

A further feature of our determinations is that none falls in the later first century bc
or earlier first century ad, creating a gap of about one hundred years between pieces
of Stages III–V and those of Stage VI. This hiatus is probably, at least in part, a conse -
quence of the size and nature of our sample. Equally, it is important to take the
potential biases of depositional practice into account — the fact that very few Celtic
art objects were deposited during this period does not necessarily mean that they were
not being made and used at that time. However, it is worth noting in this light that
there were a number of artefacts that were expected to fall within this century, but
did not. Indeed, many commentators from Fox (1958) to the Megaws (2001) have
expected that the majority of the British material would fall within this period. Even
those such as Stead (1985a) or Jope (2000) who entertained earlier dates for many
pieces would not have predicted the apparent hiatus in the late first century bc and
the early first century ad.
Certainly, more thought and work needs to be directed at this possible gap. How -
ever, our results do at the very least raise the question of a possible temporal break
between the Iron Age forms and decorations and those of the Romano-British period.
If true, it would also cause us to question why such a break occurred and whether it
might have been the attraction of new forms of material culture, such as Roman
imports in pottery or metal, or maybe the newly rich imagery of coins (Creighton
2000; 2006) that ushered in novel sensibilities or forms of consumption.
final thoughts
In their dating of Bronze Age metalwork, Needham et al. (1998) conclude that the
main phases of metalwork generally identified (Acton/Taunton, Penard, Wilburton
and Ewart Park) did indeed exist in sequence, although the start of some phases,
especially Wilburton, was pushed back a little (Needham et al. 1998, 82, illus. 15). In
our current programme, there are indications that some elements of British Celtic art
might be pushed back, aligning it better with the typologically derived dates from the
Continent. However, support for a simple sequence of styles or stages is less apparent.
We have to allow for the possibility that something more complicated was happening
with Celtic art, so that motifs and styles may have accumulated through time, with
earlier motifs being joined by later ones.
A fair amount of Iron Age metalwork, notably coins and fibulae (Haselgrove 1987;
1997) and swords (Stead 2006), does behave in good typological fashion with styles
following each other in sequence, just as had been true of Bronze Age metalwork. It
does not follow that all metalwork acted in the same manner and it may be that one
artefact could have incorporated motifs first found in different periods. If this is true,
we can maintain both Stead’s stages — as some motifs are later than others — but also
Macdonald’s (2007a) criticisms of them (echoed in some of Stead’s own observations)
that motifs from different stages can be found on the same artefact (a trait also noted
on some Continental objects, such as the scabbard from Cernon-sur-Coole, Marne).
On the basis of the dates we have, it is impossible to distinguish chronologically
Stead’s Stages IV and V. Dates for material in both range between 400 and 150 bc,
with some Stage V objects, such as the terrets from Garton Station or the Kirkburn
110 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 110

(K5) linch-pin, having dates overlapping with Stage III or IV torcs from Snettisham.
It is intriguing that Stages IV and V, seen as the real mark of British Celtic art, with
their basketry hatching, use of lobes, triskeles and berried rosettes, may well have been
flourishing as early as the period between 325–150 bc, identified by Haselgrove as the
time when contacts with the Continent were least evident. It is possible that British
decora tion achieved its distinctive forms in a period of relative isolation.
As discussed, a further interesting facet of our results is that very few determinations
fall in the century after 50 bc, so that there might be a temporal gap between the end
of Stages IV and V and the start of VI. This result needs to treated with caution at the
moment, as it might be a function of what we have been able to date (more results
from the contexts in which southern mirrors are found might fill this gap, for
instance), and of course of what was actually deposited. However, we must also be
aware that the production of Celtic art might not have been continuous, so that older
forms might have been edged out by new imported types of metal and/or the
increased importance of fibulae and coins in the late Iron Age. If this is true, it would
increase the novelty and importance of Stead’s Stage VI, prompting questions about
its (re-)emergence.
To develop these arguments more fully we would need to dive into the social uses
to which Celtic art was put in Britain, a topic beyond this paper (but see Garrow and
Gosden in prep.). However, we can make the following observations. Needham et al.
(1998) have made the observation that Late Bronze Age metalwork reflects a concern
with quantity, given the size and frequency of hoards especially in the Ewart Park
phase. We would complement this by putting forward the possibility that Celtic art
derived from a concern with quality, whereby a much smaller number of items were
made and deposited, but that many of these exhibit considerable complexity of form
and decoration. Unlike the Middle and Later Bronze Age, standardized forms were
very rare; in contrast to the many examples of identical socketed axes, for example,
no two swords are exactly alike. This is partly because swords and scabbards were
made from a number of different components, which could be put together differently
in their initial making, but partly also because they sometimes had complex histories.
Scabbards often show signs of repair or wear, as do other objects such as the
Newnham Croft bracelet or the Mill Hill ‘crown’. On the other hand, there seems to
be a marked emphasis on novelty and innovation in the decorations applied to Iron
Age objects. One of the problems of constructing schemes of change for decoration
is that general resemblances between the decorative schemes applied to objects can be
perceived, but very rarely are identical motifs used in the same combination. Even
relatively simple artefacts like terrets show great ranges of variation and everyone who
discusses their variability is compelled to come up with a slightly different typology.
The range of terrets produced at Gussage All Saints is very considerable and maybe
everyone supplied from that site wanted a style slightly different from that of their
neighbours. There are also indications that the composition of some hoards was quite
different in the Iron Age from those of the Bronze Age. The conundrum posed by
the Salisbury hoard is not easily ignored. In one collection of material there are arte -
facts from the whole span of metalwork in Britain, including flat axes from around
2400 bc to miniature cauldrons and shields thought to date to around 200 bc or
111dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 111

earlier, but still spanning at least two millennia. It may be that people in the Iron Age
com bined material that they had found by chance in Bronze Age contexts, but it is
also possible that old material had been curated in some way to be buried in the later
Iron Age. Stead (2000) points to other hoards containing old material at the time of
deposit and Hingley (2006) has provided instances of the Iron Age use of Bronze Age
or Neolithic material. Many of the objects designated as Celtic art seem to have been
old when deposited.
Iron Age chronologies require not just more dates and sound archaeological con -
texts, but thought about the sequences of continuity and change created by society at
the time. Any particular origin point for Celtic art is likely always to elude us for a
great range of reasons. But it is reassuring that the artefacts judged to be early on
grounds of form or decoration produced early dates. The early fourth (or even fifth)
century date for Cerrigydrudion occasions no real surprise, and indicates that Britain
was in step with the rest of Continental Europe in its production of novel metalwork.
Inevitably, this dating programme has in some senses posed more questions than
answers. However, we hope that by obtaining a large number of dates, by confirming
suspicions as to the successiveness of Stead’s stages, and by placing all of these within
a Continental European framework, we have contributed substantially to our under -
standing of the chronology, and of course fascination, of Celtic art. To our mind, our
results reinforce the point made at the beginning of this paper — that many aspects of
our Iron Age and Romano-British chronologies are of a dubious nature, based as they
are on typological assumptions and hindered by poor or poorly understood archaeo -
logical contexts. Radiocarbon dating is not a sole answer to chronological problems
in these periods, but it is a most useful element in our search for understandings of
sequence, change and the use of history in the past.
APPENDIX 1
contextual details for the samples successfully dated and
implications of the dates obtained
Please note that all radiocarbon determinations listed in Appendix 1are quoted at 95% confidence.
Bibliographic references for the original reports relating to each artefact/context can be found in Table 1.
Aylesford bucket
The bucket was found in 1886, along with a group of other finds. The bucket itself actually
contained a pile of cremated bone, one burial within a much wider cemetery. The burial is
one of the best known of a group of rich cremation burials in south-eastern England. While it
can be dated approximately to the first century bc by the brooches and bronze vessels found
with it, we felt it was important to try to establish how early in the first century the burial had
been deposited, and whether or not the bucket, the other finds and the dead individual were
actually of similar ages. We originally hoped to date a sample of ash wood from the bucket and
a sample of the cremated bone with which it was associated. However, since its original
recovery, the bone has been lost. Unfortunately the sample (taken from a stave of the bucket)
which was dated gave a broad date range (360–50 cal. bc) on old wood, rendering it largely
uninformative.
112 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 112

Bermondsey dagger
This dagger, found in its scabbard, was discovered as a stray find on the foreshore of the River
Thames in Bermondsey, London in 2005. It appears to be decorated with Hallstatt D style art,
the phase immediately pre-dating the origins of Celtic art. We felt it was important to establish
a date in Britain for these origins, and also to assess whether this object dates to the same period
as those decorated in a comparable style on the Continent. The age of the wood (ash) sampled
(from the scabbard lining) for dating could not be established. It thus provides a terminus post
quem only for this object. The determination obtained (810–550 cal. bc) suggested a date in the
eighth–sixth centuries bc, falling within the range expected for objects of this kind.
Boverton collar and bracelets
The Boverton neck-collar was found along with one bracelet by metal detectorists near
Boverton, Vale of Glamorgan in July 2005. A small excavation around the find spot by staff
of the National Museum of Wales and the Portable Antiquities Scheme, Wales, revealed no
sign of a clear vertical grave cut. However, additional human bone fragments were found in
the exca vated test pit, lying at the base of a hollow. This provides clear supporting evidence
for the existence of a recently disturbed inhumation. In addition, a second, matching, bracelet
was found beside this with radius and ulna fragments still projecting through the bracelet hoop.
The collar belongs to a class of neck ornament increasingly well known from south-west
England, Wales and Scotland, generally dated on stylistic grounds to the later first and early
second cen turies ad. The radiocarbon date derived from the Boverton sample was the first
independent one for a neck-collar of this type. The strip bracelets, or armlets, with C-shaped
cross sections, slightly expanded terminals and pointillé decoration, find clear parallels in well-
dated first-century ad contexts in southern England and East Anglia. The determination
obtained (80–230 cal. ad) from a human arm bone found inside the bracelet overlapped with
the stylistically derived date, but suggested that it was deposited towards the latter part of the
expected range. Accepting this later date of burial, the artefacts are now being interpreted as
heirlooms, made during the first century ad and remaining in circulation for fifty to one
hundred years before eventual burial with the deceased person. This interpretation is consistent
with the observation that one of the bracelets had been carefully repaired during its use-life
(A. Gwilt, pers. comm.).
Brecon grave group
This group of grave goods — which included a mirror, two mini terrets, a copper-alloy toilet
set, two pottery vessels, and a pottery lamp — was found by a metal detectorist in 2005.
Unfortunately, no contextual information about the find circumstances can be released at
present. The importance in dating this group of finds lies in its combination of Iron Age and
Roman artefact types. Along with the Welshpool burial (below), this hybrid mixture provides
an interesting analogue for the rich burials in south-east England. It was thought likely that this
burial, away from the south-eastern core zone, was actually rather later, probably dating to the
late first century ad. The determination obtained (20–220 cal. ad) from cremated human bone
is consistent with this.
Bury Hill horse gear
B. W. Cunliffe’s excavations at Bury Hill in 1990 revealed a series of features within the hill -
fort. Two pairs of adjacent pits (P23/P24 and P45/P57) produced large collections of bronze
and iron metalwork, including bridle bits, harness rings, wheel parts etc. Artefacts that formed
part of horse harness sets are by far the most numerous class of objects associated with Celtic
113dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 113

art. Despite this, very few have come from contexts that can be independently dated. Those at
Bury Hill therefore offered an opportunity to date a number of classic Late Iron Age decorated
objects. The artefacts did not appear to have been deposited on a single occasion, but were
distri buted throughout each pit’s fills. We therefore dated samples of animal bone from three
different contexts: the top (1) and bottom (9) layers of P24, and the basal layer (10) of P45.
One sample failed, but the other two determinations (370–120 cal. bc and 360–50 cal. bc)
suggested a date in the fourth–second centuries bc, arguably placing the deposit slightly earlier
than had been expected.
Cerrigydrudion headpiece
The Cerrigydrudion headpiece was found as a result of stone quarrying in 1925. It was found
within a stone cist, but there were no direct traces of a burial. The headpiece is thought to be
among the earliest examples of British Celtic art, on the basis of the palmettes incorporated
into its design. We felt that it was important to verify this stylistic assumption independently.
The determination obtained (510–250 cal. bc at 95%; 405–380 cal. bc at 68%) from the rawhide
associated and thought to be from the original lining of the headpiece confirmed an early date,
probably placing the object slightly earlier than expected at the beginning of the fourth century
bc.
Deal spoons
The Deal spoons were excavated in 1902 in a large cemetery subsequently re-investigated in
1984–89. In total, twenty Iron Age spoons have been found to date, most of them in matching
pairs. Other than a pair from Burnmouth in south-east Scotland (already dated to the
second–first centuries bc), those from Deal are the only examples to come from a secure,
datable context. They therefore offered an excellent opportunity to date this enigmatic class of
artefact at the other end of their geographical distribution. The determination obtained
(380–200 cal. bc) on a fragment of human skull placed them earlier than those from
Burnmouth, and earlier than expected, within the fourth–third centuries bc.
Deal warrior
The Deal warrior is one of the most impressive Iron Age burials found in Britain. In association
with this extended inhumation were an iron sword, the decorated bronze fittings from a
scabbard, a circular bronze ring and brooch decorated with coral beads, the bronze bindings
and fittings for a shield, and an elaborately decorated bronze crown or headband. The objects
associated with the Deal warrior vary substantially in terms of their stylistic and decorative
attributes: some would be considered typologically early but others late. It was therefore con -
sidered extremely important in terms of stylistic chronologies, as well as in terms of under -
standing Late Iron Age social identities in general, to establish an independent date for this
burial. The determination obtained (360–100 cal. bc) on a fragment of human skull places the
burial within the fourth–second centuries bc. Unfortunately, the broad span of the radiocarbon
date did not enable us to provide a much more accurate date than stylistic chronologies had
suggested in this case.
Ely ‘horn cap’
This object was found as a stray find during the nineteenth century near Ely, Cambridgeshire.
The function of this class of objects (which have come to be termed ‘horn caps’ because they
were originally thought to have acted as capping pieces on the upturned ends of chariot yokes)
is not known. In total, sixteen have been found across southern Britain, but only one was
114 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 114

datable contextually (from Maiden Castle which gave a date of c.50 bc–ad 50). The object
from Ely is unique in containing a datable fragment of wood and the importance of using this
to establish another independent date for this class of objects was clear. The determination
obtained (180–30 cal. bc) on (probably young) ash wood found inside the object suggested a
date in the second–first centuries bc.
Garton Slack cart burial
Excavations at Garton Slack were undertaken in advance of gravel quarrying during the 1960s
and 1970s. Among a vast array of impressive archaeology, the site produced two important Iron
Age burials: a cart or chariot burial accompanied by bronze and coral horse gear (including
bridle bits, strap junctions and terrets), and an inhumation accompanied by a bronze and iron
mirror. These burials form part of what has subsequently become a very well-known tradition
of inhumation within square barrows in east Yorkshire. On the basis of brooch chronologies
(derived from Continental comparisons), it has been suggested that these burials began as early
as the fourth century bc, and carried on until at least the first century bc. Despite the
outstanding quality of many of the artefacts associated with them, and the importance of
establishing an accurate chronology for the burial rite in itself, until recently very few
independent dates indeed had been acquired for the burials (recent unpublished work initiated
by Mandy Jay has changed this picture significantly; see Table 2). Our dates were targeted
specifically at those burials associated with important or unusual pieces of material culture. The
determination we obtained for the chariot burial (370–110 cal. bc) on bone from an articulated
pig skeleton found within the grave matched the date span previously assumed fairly closely,
suggesting that it fell within the earlier part of that period.
Garton Slack mirror
The mirror, which accompanied a female inhumation in a square barrow very close to the
chariot burial, was the only artefact found in the grave. A small number of iron-plate mirrors,
including one in the ‘Lady’s Barrow’ at Arras, have been found in East Yorkshire. It is thought
likely that these represent the earliest mirror type in Britain, pre-dating their often highly
decorated bronze counterparts in southern England by at least a century. As yet, however, this
impression had not been confirmed independently. As well as contributing to an understanding
of the east Yorkshire burials generally, it was considered important to establish when these
objects were first used in Britain, and how close in time they were to those in the south. While
a radiocarbon date for this burial was obtained during the 1970s, it is generally regarded as
questionable (Joy 2008a, appendix B). The determination obtained (380–200 cal. bc) from a
human arm bone for the mirror burial matched the date span previously assumed for east
Yorkshire burials, again suggesting that it lay within the earlier part of that period. It also
confirmed that the object had been deposited at least a century prior to the earliest known
bronze mirrors in the south.
Garton Station cart burial
The Garton Station and Kirkburn (see below) cart or chariot burials were excavated for
research purposes by a British Museum team led by Ian Stead in 1984–87. Their finds assem -
blages included a broad range of iron and bronze horse gear (including terrets, bridle bits,
linch-pins, etc.). As discussed above, it was considered important to establish an independent
chro nology for the east Yorkshire cart burials. The determination we obtained for the cart
burial (380–190 cal. bc) on pig bone which had been placed around the inhumation matched
the date span previously assumed fairly closely, again suggesting that it lay within the earlier
part of that period.
115dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 115

Grimthorpe shield, sword, etc.
The Grimthorpe burial was discovered in 1868 in an oval pit. Within it was a loosely crouched
inhumation (a young male) with the bronze fittings from a shield on his chest and an iron
sword with a bronze scabbard, a bronze spear ferrule and an iron spearhead to his left-hand
side; sixteen bone points were found distributed throughout the grave. This burial is one of a
small group of Iron Age warrior burials whose date span had not previously been established.
The sword falls typologically into Stead’s Group E (‘earlier swords and scabbards in the north’).
The determination obtained (360–120 cal. bc) suggested a date in the fourth–second centuries
bc, the latter part of that range being approximately in line with the date previously expected
(Stead 2006, 63).
Gussage All Saints horse gear moulds
Wainwright’s excavations of the settlement at Gussage All Saints revealed a series of deposits
relating to Iron Age iron and bronze working across the site. The most impressive was found
in Pit 209; this contained over 600 crucible fragments and 7000 mould fragments, used to make
a wide variety of horse gear. The deposits within the pit were highly stratified, there being
twenty-six in all. Although they were found throughout the pit, mould fragments from
different deposits did not generally fit together; it was therefore suggested that each one repre -
sented the residue of a separate episode of metalworking. The length of time over which this
metalworking occurred had not previously been established. The Gussage All Saints finds
repre sent the most impressive evidence for Iron Age metalworking found so far in Britain.
Impor tantly, these artefacts relate clearly to the manufacture of items which might be con -
sidered classic examples of Celtic Art. The deposit represented a rare opportunity to establish
a date for their actual manufacture. While two radiocarbon dates had already been obtained
(Q1206/7), these were taken in the 1970s from old wood charcoal and had relatively large
errors. Consequently, it was thought important to obtain a new date from young wood char -
coal. Initially we had hoped to date samples from the top and bottom of the feature in order
to gain an idea of the time span over which the deposits had accumulated. However, due to a
general absence of young wood charcoal from the lowest fills, we had to take both samples
from the upper half of Pit 209 (Layers 10a and 12). The determinations which resulted were
broad and essentially indistinguishable (360–100 and 340–50 cal. bc), suggesting a date in the
fourth–first centuries bc. These fell towards the earlier part of the period suggested by those
dates obtained in the 1970s which ranged from 355 bc–ad 80.
Kew tankard
This tankard was found in the River Thames by Kew Bridge during the nineteenth century.
Iron Age tankards are often difficult to date stylistically, and so we had little idea of the period
over which they were used and deposited in Britain. Although in several cases much of the
original tankard wood survives, not one example had yet been radiocarbon dated. We dated a
sample of yew wood from one of the staves. This produced a broad date range in the
fourth–first centuries bc (350–50 cal. bc). Owing to the fact that the sample may have been
old wood, this must be considered a terminus post quem. It nevertheless does tell us that the
object was made during the Late Iron Age or Early Roman period.
Kirkburn cart burial
The Kirkburn cart or chariot burial was excavated for research purposes by the British Museum
team in 1984–87. Along with an impressive range of mostly bronze horse gear, the finds
116 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 116

assemblage included an extremely rare example of iron chain mail. As discussed above (see
Garton Station) it was thought very important to establish a larger series of radiocarbon dates
for east Yorkshire cart burials. The determination obtained (360–60 cal. bc) from pig bones
placed around the inhumation suggested a date in the fourth–first centuries bc, matching other
east Yorkshire determinations.
Kirkburn warrior
The Kirkburn warrior was found within a square barrow immediately adjacent to the chariot
burial. The grave goods included three iron spearheads, as well as possibly the most elaborately
decorated Iron Age sword/scabbard in Europe. It was, first and foremost, considered important
to date the Kirkburn sword because it is such an excellent example of Celtic art. However,
establishing a date for this warrior burial would also provide a temporal context for other
similar burials elsewhere in England. In order to approach the issue of whether this object had
an extended life history prior to its deposition, we originally intended to date the sword itself
in combination with a sample from the skeleton it was buried with. Unfortunately, the sword
was too fragile to sample. The determination obtained (360–110 cal. bc) on a human skull
fragment suggested a date in the fourth–second centuries BC, matching other east Yorkshire
determinations.
Latchmere Green mirror
This mirror was found by a metal detectorist in 1992. A team of archaeologists from Reading
University was called in immediately to investigate the find spot. Their excavation revealed a
small pit containing a pedestalled jar filled with cremated bone. The hole dug to retrieve the
mirror was clearly visible. Subsequent analysis revealed the presence of two individuals (an
adult and a young child), along with several pig bones. The Latchmere mirror represents an
example of the south-eastern type of mirror, in this case found at the very edge of the south-
eastern region in Hampshire. It was considered important to establish whether this outlier of
the south-eastern distribution was buried around the same time as those in the core zone, or
closer in time to the earlier western mirrors. The first determination obtained (360–50 cal. bc;
360–110 cal. bc at 68%) on cremated human bone suggested a date potentially earlier than the
75–25 bc date suggested by the accompanying brooches and pottery vessels. Consequently,
two further determinations on cremated human bone were obtained. Intriguingly, these
suggested a date span of 360–110 cal. bc at 95% confidence. It is important not to make too
much of this discrepancy, as the latter part of the first determination does overlap with the date
expected from the brooches. However, given that the second and third determinations
suggested a latest possible date within the second century bc, it is also important at least to
consider the possibility that one of the two individuals (the one we dated) had been cremated
some decades before the double burial was actually deposited.
Llyn Cerrig Bach chariot and spear
The Llyn Cerrig Bach hoard is one of the most important assemblages of Iron Age metalwork
in Britain. Discovered in 1943 on the west coast of Anglesey, north-west Wales, the circum -
stances in which it was found were far from ideal archaeologically. In order to stabilize the sand
dunes close to an RAF airfield, a nearby peat bog was excavated, dried, spread across the dunes
and sewn with grass seed. Over the course of the months which followed, it was gradually
realized that the peat had contained a large assemblage of iron and bronze artefacts, which had
been distributed across the locality. Current thinking suggests that the metalwork at Llyn
Cerrig Bach may have been deposited in a watery context over a period of several centuries,
117dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 117

from perhaps as early as the fourth century bc to early in the first century ad. This theory was
estab lished partly as a result of radiocarbon dating carried out in 1997 on a number of animal
bones which were presumed to have been deposited along with the metalwork; these gave a
very broad span of calibrated dates ranging from 800 bc to ad 60 (OxA 6390-2) (Macdonald
2007b). The fact that the spear and cart or chariot pole that we dated contained preserved
wood within them presented us with a unique opportunity to date two of the objects within
this important assemblage directly, rather than relying on indirectly associated finds. The deter -
mination obtained for the cart pole (390–200 cal. bc) suggested a date in the fourth–third
centuries bc; however, this was taken from an old oak sample. The dates (taken from a young
wood sample) for the ash spear shaft lie within the second–first centuries bc (210–40 cal. bc).
These dates provide informative fixed points, associated with specific artefacts within this
assem blage, around which to work in future.
Owslebury warrior
This burial was excavated during the 1960s. It was found within the earlier of two burial
enclosures, adjacent to a major Late Iron Age and Roman settlement at Owslebury in Hamp -
shire. The most significant burial was a male inhumation, accompanied by a sword, spearhead
and shield; interestingly, it appeared to have formed a focus for a series of later cremation
burials. The determinations obtained (360–50 cal. bc and 210–50 cal. bc) from a human rib
frag ment suggested a date in the second–early first centuries bc, towards the earlier part of the
period supposed on the basis of stylistic dating, but approximately in line with the other
warrior burials dated.
Pentuan tankard
This tankard was found as a stray find within a tin stream close to St Austell in Cornwall. As
discussed above, tankards are difficult to date stylistically, and although a number have been
found across Britain, none had previously been radiocarbon dated. As with the Kew tankard,
we dated a sample of wood (probably old and probably yew) from one of the tankard staves.
This produced a date in the second–first centuries bc (200–40 cal. bc). Due to the nature of
the sample, this must be considered a terminus post quem. It does, however, suggest that the
object was made towards the end of the Iron Age or in the Early Roman period.
Salisbury hoard
The hoard from Netherhampton, near Salisbury is the largest collection of prehistoric metal
objects ever found in Britain. It was first recovered as a result of metal detecting in the 1980s,
and its contents were sold piecemeal on the antiquities market. The fact that they had originally
been found together was established only as a result of Ian Stead’s painstaking detective work,
after having been shown twelve miniature shields found within it. The hoard contained a range
of bronze artefacts spanning the entirety of the Bronze Age and the Iron Age. Excavations at
the site by the British Museum in 1993 established that these objects had been deposited in a
large pit close to an Iron Age settlement. Although nothing datable survived from the hoard
itself, we felt that it was nevertheless worth obtaining a definitive date from material within the
pit, the hoard having subsequently been placed in the ground. The determination we obtained
(400–200 cal. bc) suggested a date for the pit in the third–second centuries bc.
Snettisham torcs
The hoards of torcs from Snettisham are probably the best known Iron Age artefacts in Britain.
Following the initial find of three hoards discovered in 1948, a further series of hoards was
118 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 118

excavated by the British Museum in 1990. In total, seventy-five complete torcs and fragments
from one hundred more have been found. An approximate time span during which the torcs
are presumed to have been deposited (of between 150 and 1 bc, or possibly even earlier,
accord ing to recent revisions of the coin dates (C. Haselgrove, pers. comm.)) has previously
been assumed on the basis of diagnostic associated coins. However, previously, only a single
radio carbon date had been obtained for the site (on a lump of charcoal in Hoard M; the date
was not an AMS one and the error range was very broad). The presence of several other datable
items, in four separate hoards at Snettisham, offered a previously unrealized opportunity to
estab lish an independent date for these impressive collections of artefacts. It was also hoped that
we would be able to answer some of the questions relating to the length of time that torcs were
deposited in that locality. We obtained six determinations from five different hoards. One
sample failed and another gave a modern reading. Two further determinations suggested dates
in the fourth–second centuries bc for torcs (of simple twisted-bronze type) in Hoards F
(370–160 cal. bc) and L (370–110 cal. bc), placing those items firmly towards the earlier part
of the period of deposition previously assumed. A third determination (1–130 cal. ad)
suggested a date in the first or early second centuries ad for the silver lumps in Hoard M,
suggesting that deposition at the site may well have continued into the Roman period (a
suggestion also supported by recent coin finds; Jody Joy, pers. comm.). A fourth determination
(790–510 cal. bc) suggested a date in the eighth–fifth centuries bc for a torc (of simple twisted-
bronze type) in Hoard L; while such an early date is possible, it does seem very early — the
sample may perhaps have been adversely affected by conservation treatments.
Trawsfynydd tankard
This tankard was found during peat cutting before 1850. With its elaborate openwork handle,
it is perhaps the most visually impressive of all the British tankards. As discussed above, it was
considered important to establish a good set of dates for these objects. The fourth–second
centuries bc date (400–200 cal. bc) obtained from one of the tankard staves closely matched
those from the other tankards. However, as the stave was made of old yew wood, this date
must be considered a terminus post quem and cannot necessarily be viewed as closely indicative
of the tankard’s date of manufacture.
Welshpool grave group
This group of objects — which included a wooden bucket with bronze ox-head mounts, an
iron fire dog, a further iron stand, four imported bronze vessels and two glass vessels — was
found by workmen laying a drain in 1959. Unfortunately, further excavations revealed no
further information about the objects’ contexts although it can, perhaps, be presumed that the
finds came from a rich cremation burial. As a group, the Welshpool finds represent a very
significant collection of Celtic and Roman objects found together. Consequently, they provide
an interesting analogue for the better-known rich burials in south-eastern England (e.g.
Aylesford, above). Despite the presence of Roman finds, the group could previously be dated
no more precisely than pre-ad 150. The determination obtained (on old yew wood from a
bucket stave) suggested a date in the first–second centuries ad (1–210 cal. ad). While this can
is only a terminus post quem, it nevertheless indicates that the burial was probably deposited
during the first century ad or later.
119dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 119

APPENDIX 2
objects that could not be dated
object reason
Arras ‘Charioteer’s barrow’ Associated bone lost/untraceable
Arras mirror burial (Lady’s barrow) Associated bone lost/untraceable
Aylesford bucket cremation Associated bone lost/untraceable
Aylesford tankard Associated bone lost/untraceable
Battersea shield No datable material survives
Birdlip mirror Associated bone lost/untraceable
Fiskerton file Too fragile to sample
Fiskerton pull-saw Too fragile to sample
Fiskerton sword No datable material survives
Ghegan Rock bone comb Too fragile to sample
Hamperden End shears Associated pottery dates adequate
Kirkburn warrior’s sword Too fragile to sample
Langbank bone comb Too fragile to sample
Newnham Croft arm ring/headpiece Associated bone lost/untraceable
North Grimston anthropoid dagger Associated bone lost/untraceable
Shepperton Ranges sword Too fragile to sample
Shouldham anthropoid handle Associated bone lost/untraceable
Waterloo helmet No datable material survives
Witham shield No datable material survives
ACKNOWLEDGEMENTS
This study could not have been attempted, and certainly would not have been completed, without the
assistance, time and goodwill of many people. First of all, we would like to thank those at museums,
units and universities around Britain for first locating, and then allowing us to sample, the objects
discussed here: David Allen (Andover Museum), Luke Currell, Duncan Hook, Jody Joy, Nigel Meeks,
Jim Peters and Peter Rea (British Museum), Keith Parfitt (Canterbury Archaeological Trust), Peter
Woodward (Dorset County Museum), Jon Iveson (Dover Museum), Paula Gentil (Hull and East
Riding Museum), Christine Longworth and Tracey Seddon (National Museums Liverpool), Jon
Cotton, Rose Johnson and Rebecca Lang (Museum of London), Jane Marley and Laura Ratcliffe
(Royal Cornwall Museum, Truro), and Richard Brewer, Evan Chapman, Mary Davis, Jody Deacon
and Adam Gwilt (National Museum Wales). Special thanks are due to Jody Joy who had to work
especially hard to keep up with our almost endless demands at the BM. It is also important to
acknowledge those who helped us only to find out that certain objects could not be dated: Louise
Allen (Gloucester City Museum), John Davies (Norwich Castle Museum), Jennifer Foster (University
of Reading), Dawn Heyworth (Museum of Lincolnshire Life), Fraser Hunter (National Museum of
Scotland), Chris Knüsel (University of Bradford), Jay Stock (Duckworth Laboratory, University of
Cambridge), Anne Taylor (Museum of Archaeology and Anthropology, University of Cambridge) and
Tom Wilson (Network Archaeology). We would also like to thank Diane Baker for her assistance at
the laboratory in Oxford, John Collis for his assistance with the Owslebury ‘burial with weapons’,
Mandy Jay for allowing us to cite the dates she had previously obtained for various east Yorkshire
burials, Mansel Spratling for sharing his knowledge, wisdom and comments on many aspects of this
project, and Ian Stead for his help with Snettisham, with typological and stylistic dating issues, and
120 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 120

much more besides. Vital specialist analyses were carried out by Esther Cameron (Oxfordshire
Museums Resource Centre; rawhide), Caroline Cartwright (British Museum; plant materials), Louise
Loe (Oxford Archaeology; human bone), Dan Miles (RLAHA, University of Oxford; wood) and
Quita Mould (Barbican Research Associates; rawhide). We are also very grateful to John Collis,
Andrew Fitzpatrick, Adam Gwilt, Colin Haselgrove, Jody Joy, Mansel Spratling and Ian Stead for their
detailed and thoughtful comments on earlier versions of this paper. The radiocarbon dates were
generously funded by the NERC and the AHRC via the ORADS programme. We are particularly
grateful to the staff of the Oxford Radiocarbon Accelerator Unit, University of Oxford, for their
careful analytical work on the samples that were dated. Tom Higham helped with the sampling of
some of the material for the project. The Arts and Humanities Research Council (AHRC reference
number 112199) also funded the project as a whole for which we are very grateful.
BIBLIOGRAPHY
Bataille, G. 2001. Les agrafes de ceinturon du sanctuaire de la Villeneuve-au-Châtelot (Aude),
Archäologisches Korrespondenzblatt, 31, 443–60
Boon, G. 1961. Roman antiquities at Welshpool. Antiq. J., 41, 13-31
Bradley, R. and Smith, A. 2007. Questions of context: a Greek cup from the river Thames, in Gosden
et al. 2007, 30–42
Brewster, T. 1971. The Garton Slack chariot burial, East Yorkshire. Antiquity, 45, 289–92
Brewster, T. 1975. Garton Slack, Current Archaeol., 51, 104–16
Bronk Ramsey, C. 2009a. Bayesian analysis of radiocarbon dates. Radiocarbon, 51(1), 337–60
Bronk Ramsey, C., 2009b. Dealing with outliers and offsets in radiocarbon dating, Radiocarbon, 51(3),
1023–45
Bronk Ramsey, C. Higham, T. and Leach, P. 2004. Towards high-precision AMS: progress and
limitations, Radiocarbon, 46, 17–24
Brosseder, U., Sauter, E. and Schwentzer, S. 2003. Bermerkungen zur Heuneburg and ihrer
Publikation, Prähistorische Zeitschrift, 78, 60–98
Brown, F., Howard-Davis, C., Brennand, M., Boyle, A., Evans, T., O’Connor, S., Spence, A.,
Heawood, R. and Lupton, A. 2007. The Archaeology of the A1(M) Darrington to Dishforth DBFO
Road Scheme, Lancaster: Oxford Archaeology North, Lancaster Imprints 12
Carter, S. and Hunter, F. 2003. An Iron Age chariot burial from Scotland, Antiquity, 77, 531–35
Collis, J. R. 1968. Excavations at Owslebury, Hants: an interim report. Antiq. J., 48, 18–31
Collis, J. R. 1973. Burials with weapons in Iron Age Britain, Germania, 51, 121–33
Collis, J. R. 2008. Constructing chronologies: lessons from the Iron Age, in A. Lehoërff (ed.) Construire le
Temps. Histoire et méthodes des chronologies et calendriers des derniers millénaires avant notre ère en Europe
occidentale, Actes du XXXe colloque international de HALMA-IPEL, 7–9décembre 2006, 85–104,
Glux-en-Glenne: Collection Bibracte, 16
Corcoran, J. 1952. Tankards and tankard handles of the British Early Iron Age, Proc. Prehist. Soc., 18,
85–102
Cotton, J. and Green, A. 2004. Further prehistoric finds from Greater London, Trans London and
Middlesex Archaeol. Soc., 55, 119–51
Creighton, J. 2000. Coins and Power in Late Iron Age Britain, Cambridge: Cambridge University Press
Creighton, J. 2006. Britannia: the Creation of a Roman Province, London: Routledge
Cunliffe, B. and Poole, C. 2000. Bury Hill, Upper Clatford, Hants, 1990, Oxford: Oxford University
Committee for Archaeology
de Navarro, J. M. 1952. The Celts in Britain and their art, in M. P. Charlesworth and M. D. Knowles
(eds) The Heritage of Early Britain, 56–82, London: G. Bell and Sons Ltd
Dee, M. and Bronk Ramsey, C. 2000. Refinement of graphite target production at ORAU, Nuclear
Instruments and Methods in Physics Research, 172B, 449–53
Evans, A. J. 1890. On a Late-Celtic Urn-field at Aylesford, Kent, and on the Gaulish, Illyro-Italic, and
Classical connexions of the forms of pottery and bronze-work there discovered. Archaeologia, 52,
315–88
121dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 121

Field, N. and Parker Pearson, M. 2003. Fiskerton. An Iron Age Timber Causeway with Iron Age and Roman
Votive Offerings: the 1981 Excavations, Oxford: Oxbow Books
Fitzpatrick, A. 2007. Druids: towards an archaeology, in Gosden et al. 2007, 287–315
Fox, C. 1946. A Find of the Early Iron Age from Llyn Cerrig Bach, Anglesey, Cardiff: National Museum of
Wales
Fox, C. 1958. Pattern and Purpose: A Survey of Early Celtic Art in Britain, Cardiff: National Museum of
Wales
Fulford, M. and Creighton, J. 1998. A late Iron Age mirror burial from Latchmere Green, near
Silchester, Hampshire, Proc. Prehist. Soc., 64, 331–42
Garrow, D. 2008. The space and time of Celtic Art: interrogating the ‘Technologies of Enchantment’
database, in Garrow et al. 2008, 15–39
Garrow, D. and Gosden, C. in prep. A Technology of Enchantment? Celtic Art in Britain 400 BC to AD 100,
Oxford: Oxford University Press
Garrow, D., Gosden, C. and Hill, J. D. (eds) 2008. Rethinking Celtic Art, Oxford: Oxbow Books
Gosden, C. and Hill, J. D. 2008. Introduction: reintegrating ‘Celtic’ art, in Garrow et al. 2008, 1–14
Gosden, C., Hamerow, H., de Jersey, P. and Lock, G. (eds) 2007. Communities and Connections: Essays
in Honour of Barry Cunliffe, Oxford: Oxford University Press
Haselgrove, C. 1987. Iron Age Coinage in South-east Britain: The Archaeological Context, Oxford: Brit.
Archaeol. Rep. Brit. Ser., 174
Haselgrove, C. 1996. Roman impact on rural settlement and society in southern Picardy, in N. Roymans
(ed.) From the Sword to the Plough: Three Studies on the Earliest Romanisation of Northern Gaul, 127–87,
Amsterdam Archaeol. Stud., 1
Haselgrove, C. 1997. Iron Age brooch deposition and chronology, in A. Gwilt and C. Haselgrove (eds)
Reconstructing Iron Age Societies, 51–72, Oxford: Oxbow Books
Haselgrove, C. 1999. The development of Iron Age coinage in Belgic Gaul, Numismatic Chronicle, 159,
111–68
Haselgrove, C. 2001. Iron Age Britain and its European setting, in J. R. Collis (ed.) Society and Settlement
in Iron Age Europe, 37–72, Sheffield: J. R. Collis Publications
Haselgrove, C. and Moore, T. 2007. The Later Iron Age in Britain and Beyond. Oxford: Oxbow Books
Haselgrove, C. and Pope, R. 2007. The Earlier Iron Age in Britain and the Near Continent, Oxford: Oxbow
Books
Haselgrove, C., Armit, I., Champion, T., Creighton, J., Gwilt, A., Hill, J. D., Hunter, F. and
Woodward, A. 2001. Understanding the British Iron Age: An Agenda for Action. A Report for the Iron
Age Research Seminar and the Council of the Prehistoric Society, Salisbury: Wessex Archaeology
Hawkes, C. 1959. The A B C of the British Iron Age, Antiquity, 33, 170–82
Hedges, R., Housley, R., Law, I. and Perry, C. 1988. Radiocarbon dates from the Oxford AMS
system: Archaeometry datelist 7, Archaeometry, 30, 155–64
Hedges, R. E. M., Law, I. A., Bronk, C. R. and Housley, R.A. 1989. The Oxford accelerator mass
spectrometry facility: technical developments in routine dating, Archaeometry, 31 (2), 99–113
Hingley, R. 2006. The deposition of iron objects in Britain during the later prehistoric and Roman
periods: contextual analysis and the significance of iron, Britannia, 37, 213–57
Hodson, F. R. 1964. Cultural grouping within the British pre-Roman Iron Age, Proc. Prehist. Soc., 30,
99–110
Jacobsthal, P. 1944. Early Celtic Art, Oxford: Clarendon
Johns, C. 2006. Iron Age sword and mirror cist burial from Bryher, Isles of Scilly, Cornish Archaeol.,
41‒42, 1–80
Jope, E. M. 2000. Early Celtic Art in the British Isles, Oxford: Clarendon
Joy, J. 2008a. Reflections on the Iron Age: Biographies of Mirrors, unpublished PhD thesis, University of
Southampton
Joy, J. 2008b. Reflections on Celtic Art: a re-examination of mirror decoration, in Garrow et al. 2008,
78–99
Macdonald, P. 2007a. Perspectives on insular La Tène art, in Haselgrove and Moore 2007, 329–38
Macðonald, P. 2007b. Llyn Cerrig Bach: A Study of the Copper Alloy Artefacts from the Insular La Tène
Assemblage, Cardiff: University of Wales
MacGregor, M. 1962. The Iron Age metalwork hoard from Stanwick, Proc. Prehist. Soc., 28, 17–57
122 dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 122

Megaw, J. V. S. and Megaw, R. 2001. Celtic Art from its Beginnings to the Book of Kells, London: Thames
and Hudson
Naysmith, P., Scott, E. M., Cook, G. T., Heinemeier, J., van der Plicht, J., Van Strydonck, M.,
Bronk Ramsey, C., Grootes, P. M. and Freeman, S. P. H. T. 2007. A cremated bone
intercomparison study, Radiocarbon, 49(2), 403–08
Needham, S., Bronk Ramsey, C. R., Coombs, D., Cartwright, C. and Pettit, P. 1998. An
independent chronology for British Bronze Age metalwork: the results of the Oxford radiocarbon
accelerator programme, Archaeol. J., 154, 55–107
Parfitt, K. 1995. Iron Age Burials from Mill Hill, Deal, London: British Museum Press
Reimer, P. J., Baillie, M. G. L., Bard, E., Bayliss, A., Beck, J. W., Bertrand, C. J. H., Blackwell, P.
G., Buck, C. E., Burr, G. S., Cutler, K. B., Damon, P. E., Edwards, R. L., Fairbanks, R. G.,
Friedrich, M., Guilderson, T. P., Hogg, A. G., Hughen, K. A., Kromer, B., McCormac, G.,
Manning, S., Bronk Ramsey, C., Reimer, R. W., Remmele, S., Southon, J. R., Stuiver, M.,
Talamo, S., Taylor, F. W., van der Plicht, J. and Weyhenmeyer, C. E. 2004. IntCal04
terrestrial radiocarbon age calibration, 0–26 cal kyr BP, Radiocarbon, 46(3), 1029–58.
Sormaz, T. and Stöllner, T. 2005. Zwei Hallstattzeitliche Grabkammern von Dürrnberg bei Hallein —
neue dendrochronologische Ergebnisse zum übergang von der Hallstatt- zur Frühlatènezeit,
Archäologisches Korrespondenzblatt, 35, 361–76
Stead, I. 1967. A La Tène burial III from Welwyn Garden City, Archaeologia, 101, 1–62
Stead, I. 1968. An Iron-Age hill-fort at Grimthorpe, Yorkshire, England, Proc. Prehist. Soc., 34, 148–90
Stead, I. 1982. The Cerrigydrudion ‘hanging bowl’, Antiq. J., 62, 221–34
Stead, I. 1984. Some notes on imported metalwork in Iron Age Britain, in F. Macready and F. H.
Thompson (eds) Cross-Channel Trade between Britain and Gaul in the pre-Roman Iron Age, 43–66,
London: Soc. Antiq. Occas. Pap. New Ser., 4
Stead, I. 1985a. Celtic Art in Britain before the Roman Conquest, London: British Museum Press
Stead, I. 1985b. The Battersea Shield, London: British Museum Press
Stead, I. 1991a. The Snettisham treasure: excavations in 1990, Antiquity, 65, 447–64
Stead, I. 1991b. Iron Age Cemeteries in East Yorkshire, London: English Heritage
Stead, I. 1995. The metalwork, in K. Parfitt (ed.) Iron Age Burials from Mill Hill, Deal, 58–111, London:
British Museum Press
Stead, I. 2000. The Salisbury Hoard, Stroud: Tempus
Stead, I. 2006. British Iron Age Swords and Scabbards, London: British Museum Press
Trachsel, M. 2004. Untersuchungen zur relativen and absoluten Chronologie der Hallstattzeit, Bonn:
Universitätsforschungen zur prähistorischen Archäologie, 104
Wainwright, G. 1979. Gussage All Saints: An Iron Age Settlement in Dorset. London: Dept of the
Environment Archaeol. Rep., 10
Wheeler, M. 1954. The Stanwick Fortifications, North Riding of Yorkshire, London: Rep. Res. Comm. Soc.
Antiq. London, 17
D. Garrow, School of Archaeology, Classics and Egyptology, University of Liverpool
Email: Duncan.Garrow@liverpool.ac.uk
C. Gosden, School of Archaeology, University of Oxford
Email: chris.gosden@arch.ox.ac.uk
J. D. Hill, British Museum, London
C. Bronk Ramsey, School of Archaeology, University of Oxford
123dating celtic art
A3 Garrow:Layout 1 01/09/2010 15:03 Page 123