Recent Roman Iron Age metalwork finds from and Taside

Fraser Hunter

In the last few years several metal artefacts of , (DES 1991, 22), Roman Iron Age date have been discovered in - Fife and , largely through the activities of I A . T metal-detectorists. This paper aims to put these on , record, with a discussion ranging through other Roman finds from Fife in an attempt to tease out . D E F some wider conclusions. Hopefully, this will also M S (EFMS) D D. indicate the value of co-operation with metal- detectorists and the benefits of s Treasure Tankard handles are an Iron Age type which Trove law, which enables such material to be safe- continued into the Roman period. This is only the guarded in museums (Sheridan 1995). fifth from Scotland (MacGregor 1976, nos 287-288, 290-291; no 289 from Culbin is suspect). It belongs The Artefacts to Corcorari s group V (1952), although his typo- logy has flaws (cf Jackson 1990, 44). There are Although three of the artefacts discussed below are close parallels from Okstrow broch, in Iron Age types in origin (the penannular brooch, (MacGregor 1976, no 291) and Caerleon Roman button-and-loop fastener and tankard handle), all fort, in southern Wales (Evans and Metcalf 1992, are Roman period in date: the presence of zinc in 152, no 328), although the Ballinbreich example the alloys shows they were made from remelted lacks their strut joining the terminals. The Caerleon Roman metal, as zinc is only found at trace levels example is dated to the 1st - early 2nd century AD, in the pre-Roman period (Bayley 1990,13). Find while MacGregor suggests Okstrow is 2nd century spots are mapped in Illus 1. Alloy types are based AD. From this, the Ballinbreich fragment is prob- on qualitative X-ray fluorescence analysis, mainly ably late 1st - 2nd century AD. by Dr Katherine Eremin of the NMS Analytical Research section. Surface corrosion will have B-- , B C ( 2.2) affected the results, and hence the alloy types are L 40 ; L 22 , W 16 , H 8 ; quoted as broad groups only (see Bayley 1990, 8 L 25 , W 13 , H 9 . A: for terminology). , M- ; NGR NO 272 204. S T , B C ( 2.1) . D EFMS L 20 ; W 18 ; H 22 . A: D D. , T C III (-) -- . T - (W 1970). O ; - , , . A . : Wild (1970) provides the most thorough , 5 discussion of the type, arguing for a late 1st - 2nd . century AD date. Their function was to join straps, particularly in horse harness. Class III is the M- ; NGR NO 272 204. A commonest type, found predominantly in south- N M R ern Scotland and northern England: the bossed decoration is typical of Celtic-style metalwork

T F A , 2 (19%),113-125 114 Roman Iron Age metalwork

Illus 1. Find spots of the recent discoveries (drawn by Marion O'Neil).

in the area (Leeds 1933, 110). They occur in both Fowler (1960, 174-5) dates type A3 to 1st - 3rd native and Roman contexts, and some at least were century AD, with some residual ones in Anglo- made on Roman sites: there is an unfinished Saxon graves (see also Mackreth 1989, 98). example with an unperforated loop from New- stead, not previously noted as such (Curle 1911, H , P ( 2.4) plate LXXV, 7; NMS FRA 668). As with much L 95 ; W () 30 . A: - - Roman Iron Age or Romano-British material, it is ; - not clear who is making what for whom. T - , P , L A ( 2.3) . T E 26 28.5 ; : 2.5 ; 3 . A: , H A3() (F 1960), -- . T . T - ; . S , , . T , . C -, - . F . .

F M K MK-S F J S : L A, NGR NO 4405 2095. T NO 2434 1849. T -M L . A EFMS , I A . A T T. 250 E (NMRS NO 42 SW 25), 500 Penannular brooches were a pre-Roman Iron N - Age type which became popular during the (S J 1967). A EFMS Roman period on both native and Roman sites. T T. Fraser Hunter 117

be dated any closer than late 1st - late 2nd century , AD (Snape 1993,14-15). . T C- . E S- F -- , C M ( 3.1) L 28 ; W 16 ; T 6.5 . A: , , . W F , , - , , C ( . T MG 1976, -). T-. A , F R B - : , T: NGR NO 1825 2025. A P M , - T T, PMAG 1992.600. . T . N , Such brooches are known both with a variety of : Celtic designs, as here, and with more geometric , motifs (Bateson 1981, 26-8; MacGregor 1976, 123-4; (S Hattatt 1987,124-5; Bohme 1970, 10, Abb 4), 1993, 6 4.3; . H 1989, 315, although to see a degeneration from Celtic to 329-30). I 1 - 2 geometric is probably over-simple (cf MacGregor, AD. op cit). A broad date range of late 1st - 2nd century AD can be suggested. F J OD -: NGR NO 365 134. N Discussion . A EFMS T T. Having described these objects in some detail, what can they tell us? Are they just arid dots on F , B ( 3.2) the map to be labelled yet more brooches and L 28 ; W 24 . A: forgotten? This would be an injustice: they are H . T interesting objects in their own right, and the , Ballinbreich tankard handle and Pusk headstud . T brooch in particular are significant additions to . their types. Yet these stray finds can tell us much T more, if only they are asked the right questions. , The first question is what is the nature of their . find spots. There are three main possibilities: A rubbish from settlements; votive offerings; or casual losses. It is reasonable to see fragmentary . T 3.1 C (S 1993, objects as derived from settlement. Hence the finds 15); 1 - 2 AD. from Ballinbreich, Balmerino, Cupar Muir and Pusk probably derive from unrecognised settle- M- : NGR NO 3590 2460, ments. For those brooches lacking only the pin, the SE A. N -M most economical hypothesis would be casual loss. . D EFMS D However, the brooch perhaps warns D. against this pragmatic stance. Its context suggests it was a votive offering: the site, a liminal location E , I ( 3.3) near water, is typical, and personal ornaments L 68 ; W 23 ; H 27 . A: were favoured sacrifices at the time (Hunter forth- coming; cf Coventinas Well, Northumberland, E R (C Allason-Jones and McKay 1985). Such use of R 1969, 296-7) brooches is well paralleled: in Scotland, Roman . O brooches were incorporated in the native votive ; hoard from Lamberton Moor (Anderson 1905), and , . other stray finds of intact brooches could T ; plausibly be votive gifts (eg from and Polmaise; Curle 1932, fig 32, fig 36.2-3); more . T generally, brooches were common offerings in Roman-Celtic temples (eg Harlow; France and 118 Roman Iron Age metalwork

Illus 4. Roman finds from Fife. Note that coin hoards are also plotted from and Inchyra to give a better idea of their distribution: the Leven hoard is of uncertain date, but the remainder are late 2nd - early 3rd century AD (drawn by Marion O'Neil).

Gobel 1985, 70, 137). The earlier votive use of the There is not the evidence to attempt a compre- Tay is seen in the Late Bronze Age swords thrown hensive portrayal of Roman-native relations in the into it (Coles 1960, 85), and it is no surprise that area. Instead, discussion will focus more closely on such a major river had continuing ritual signific- selected aspects of the Fife evidence, which hope- ance; such long-term votive use is attested else- fully will throw some light on the general picture. where in Scotland (Hunter forthcoming). The small The scatter of Roman artefacts in Fife (Illus 4) is hoard of late-2nd-century also from the markedly augmented by these new finds, and here Inchyra foreshore is likely to be another votive the significance of metal-detecting discoveries deposit, presumably by native hand given its date within a professionally monitored programme is (DES 1994, 90). apparent. Given the generally small number of All the find spots considered here merit further Roman artefacts from Iron Age excavations, and work, particularly a field-walking programme. the models of restricted access to Roman material With the postulated settlement sites, the question which are developed from this, such an increase must be, are they Roman or native? The artefacts carries implications that Roman goods may have would be equally at home in either. The context, been reaching Scotland in greater numbers than however, points to a native origin. There is normally realised. Excavation may not be the ideal currently no evidence of a permanent Roman strategy to recover this evidence, and a broader presence in Fife, although the Severan legionary approach including metal-detecting stray finds is base at Carpow is close by - the series of marching needed to develop our views of Roman-native con- camps running west-east through the peninsula tact. An initial attempt at this is made below for represents campaigning, not permanent occupa- brooch finds. tion (Keppie 1986,153-5). Forts may yet be found, Interpretation of the Roman material is and Keppie (1990, 3-5) has noted the unrealised hindered by our poor understanding of Iron Age potential of stray finds as site indicators. In our settlement in the area. The general picture shows present state of knowledge, however, we shall take both open settlement and hillforts, the latter a more sceptical view than earlier antiquarians, perhaps declining by the Roman period (Macinnes who saw Romans in every hillock (Small 1823; 1982), but excavations have been few and far Miller 1857), and assume instead that these between. However, the range of objects matches brooches are most likely to be Roman finds from that from elsewhere: Samian and other pottery, non-Roman sites. brooches, glass, a bronze vessel and a scatter of Fraser Hunter 119

coins and coin hoards (Robertson 1970). debris from Constantines Cave (ibid, 254). The What the Fife material does elegantly extensive faunal material included a wide re- demonstrate is the shift in Roman-native presentation of skeletal parts, implying whole relationships in the late 2nd - 3rd century, carcasses were brought to the sites, primarily the when there was a concentration of silver coin normal domesticates with some wild species and hoards from up the east coast to shellfish. Crosses were incised on the walls at both , with several from Fife and caves, part of a widespread devotional use of such Kinross (illus 4; Robertson 1975). Unrest on the sites in early historic and medieval Fife (Stuart frontier at this time is attested in the written 1867, lxxxvii-xciv). The walls of both also bore sources, and the series of putatively Severan simple animal carvings (Wace and Jehu 1915, 236, marching camps indicate that Fife was part of the 242, fig 7; Stuart 1867, p129): such undetailed problem. The hoards point to the new Roman depictions are best seen as part of a scattered, policy of appeasement or bribery - buying peace- broadly Iron Age tradition in southern Scotland ful neighbours (Todd 1985, 230-1) - which is (Van Hoek and Smith 1988, 33-4) rather than as hinted at again in the fragmentary Late Roman Pictish carvings, which are generally more silver spoon in the Norries Law hoard (Stevenson diagnostic (contra Wace and Jehu 1915, 242). 1956). We may draw certain interpretative strands out Subsequent discussion will look at three areas: of this. There is evidence of craft processes: iron- the context of certain Roman finds; burials with smelting and bone- and antler-working. Both caves Roman grave goods; and the broader interpreta- saw later Christian ritual use, and both have tion of brooch finds. probable Iron Age animal carvings. It seems on balance unlikely that we are dealing with normal domestic occupation: rather, the caves were the Roman finds from Constantines Cave foci for more specialised activities, perhaps and Kinkell Cave regarded as too unpleasant or inappropriate for There are few site finds of Roman material in Fife: the settlement itself (cf Shepherd 1983, 335 for the hillfort of Clatchard Craig, a kitchen midden caves and Pictish metal-working). at Lower Largo, two burial sites (discussed below), There is a further interpretative level we can and Constantines Cave and Kinkell Cave (Robert- proceed to, albeit tentatively. While modern son 1970; Close-Brooks 1986). Here the somewhat excavations of caves are rare, work at the enigmatic cave finds will be considered. From the Sculptors Cave, Covesea, identified a published account, both were relatively rich in Roman Iron Age / Pictish phase of ritual use, Roman finds, with Samian, coarse ware and a involving the deposition of artefacts (including a bronze jug handle from Kinkell cave and coarse rich Roman assemblage) and perhaps bones, ware, substantial portions of several amphorae and and the carving of Pictish symbols on the walls part of a cylindrical glass bottle from Constantines (Shepherd 1993, 80-81). This gives us a model to Cave (Wace and Jehu 1915). Unfortunately, all test against the Fife caves. The animal figures here appear to be lost apart from two sherds of Dressel may relate to ritual use of the cave, as may the 20 Spanish olive oil amphorae from Constantines wider range of carvings, some clearly Pictish, in Cave, now held by East Fife Museum Service. This other Fife caves at Wemyss and Caiplie (Ritchie quantity is unusual - generally only high-status and Stevenson 1993; Murray 1961). The Christian sites boast such assemblages (cf Robertson 1970), use of the caves may then be following an older but few would see caves as high status. tradition, again as at Covesea (Shepherd 1993, It is a hard task to attempt any reinterpretation 80-81). This in no way contradicts the role of when the finds are lost and the records few. In the caves in craft activities, as these themselves addition, the quantity of material recovered and were probably bound up with ritual - the magical the presence of later incised crosses indicates the associations of metal-working, for instance, are sites saw use over some considerable time, further commonplace in pre-industrial societies (Budd complicating the analysis. Yet it is worth speculat- and Taylor 1995; Hingley forthcoming), and caves ing a little, if only to open up some new possibil- are common locations for such activity (Ritchie ities. and Stevenson 1993, 205). This may then give us There are a number of clues to the caves use. a context for the rich Roman material - as at The depth of cultural deposits suggests more than Covesea, it could be offerings, with exotic and casual occupation. Constantines Cave preserved powerful material being used in local rituals. what was identified as an iron-smelting furnace, The occurrence of a few caves elsewhere with while the fragmentary nature of much of the similarly rich Roman assemblages (eg Bomess faunal material at Kinkell Cave suggested bone- Cave, ; Curle 1932, 372-3) working debris rather than butchery (Wace and suggests this model may have wider applicabil- Jehu 1915, 245-6); there was also antler-working ity. 120 Roman Iron Age metalwork

Illus 5. Brooch and spear- or javelin-head from Merlsford, Fife (drawn by Marion O'Neil).

after 60 AD (Wheeler and Wheeler 1932, 71-4; A probable 'warrior burial' with Roman brooch from Merlsford (Illus 5) Stead and Rigby 1989, 91-3, 101; Feug6re 1985, Another interesting phenomenon is the presence in 265-6; Mackreth 1989, 97). This example cor- Fife of two burial sites with Roman grave-goods. responds to Augst type 4.4.4 (Riha 1979, 100, Taf The Hallow Hill burials (Proudfoot 1976) will 19), but the typology cannot discern between pre- shortly be published. Less well known is the and post-conquest arrivals in Britain. Of interest probable burial from Merlsford, which produced a here is the British distribution: markedly south and small iron spearhead and an early Roman brooch. east English, with only a scatter of outliers, and no The context is far from perfect, as Stevenson has others from Scotland or the Tyne-Solway frontier noted: the original account is vague and the early (Mackreth 1989, 97; Snape 1993). This brooch museum records confused (Stevenson 1966, 25, 40). clearly reached Scotland before the actual Roman However, a plausible interpretation is that during invasion, which is very unusual (Stevenson 1966, the excavation of a cairn an iron spearhead and 25). Roman brooch were found: it is inferred that they were associated (which is supported by similar F (NMS FG 2) adhering soil traces), most probably in a burial. L 76 ; W () 26 ; 12 This is of some significance, as Stevenson recognised, but has subsequently been little I - , commented on. The opportunity is taken here to , publish the finds fully and to discuss the discovery . T , a little further. -, . R L D (NMS FG 1) L 57 ; W 17 ; H 11 . A: - There is no meaningful typology for Iron Age ; - spearheads (Stead 1991, 74-5), although both T , , angular blades and midribs can be matched in the , series from Traprain (Burley 1956, nos. 384-395). ; This example is important as a dated example , . T , to add to the sparse Scottish corpus. The small . S diameter of its socket suggests it is a javelin rather than a spear. . A ; . It is assumed here that these finds do indeed represent a burial of the early first century AD. Langton Down brooches are a Gallic type, Interpretation is far from simple: burials are rare in starting in the late first century BC and running Iron Age Scotland (Whimster 1981, 410-16), and through the Claudian period, with few surviving the unusual nature of this one, with its exotic early

Fraser Hunter 121

I 6. R N, T, -R . P H D C D. T L T 3 R . U ( M ON).

brooch, makes it particularly hard to understand. is from Camelon, probably of late first century AD However, it is significant on two counts: as a date (Breeze et al 1976). This may augment the case warrior burial, and as a burial with Roman grave for Camelon as a native burial, although the goods. On the first count, the only Scottish parallel unusual nature of both makes it foolhardy to be 122 Roman Iron Age metalwork

dogmatic. It is, however, a valuable addition to the their assemblage, to facilitate comparison. The data record. set totals 207 brooches - while further discoveries On the second count, burials with Roman grave may modify patterns, there are sufficient brooches goods are rare in Scotland, running from Fife to to put some faith in them. with outliers in and It is assumed here that most strays relate to Orkney (Hunter forthcoming, fig 12.3). The date native rather than Roman activity. There will of range is broad, making a single explanation course be exceptions, but as many are far from any unlikely, although the role of Roman artefacts as known Roman site, and are apparently debris from prestige goods in native society suggests these settlements or votive deposits rather than casual should be burials of some status. The broochs date losses, it seems in general valid. If a Roman origin would equally allow the Merlsford example to were to be argued, a wider range of everyday be further evidence of pre-Roman contacts with Roman copper alloy objects would be expected: southern tribes, as Fitzpatrick (1989) has argued: this is rarely the case. the use of exotic material was a feature of the In interpreting the patterns, there are several southern Scottish Iron Age (Hunter forthcoming). variables to consider. The most obvious is chrono- It could even be argued that the burial is not logy; there are pre-Agricolan and post-Severan indigenous, the personal nature of jewellery brooches (the P-shaped type) from non-Roman suggesting this was an incomer. In our current sites which are unknown on Scottish Roman sites state of knowledge, these interpretations can only for obvious reasons. The other two key variables be interim. What matters is to put the find (and its are availability and choice. For the first, there is a ambiguities) on record: discussion can now broad congruence between what was used on proceed on a more informed basis. Roman sites and what appears in native hands, as would be expected. However, choice was clearly Roman brooches in Scotland being exercised: there are differences in brooch type ratios which appear significant. Trumpet and The third strand to be considered here is how the (less markedly) headstud brooches are proportion- new brooch finds fit into the wider Scottish ally over-represented in native contexts, while picture. It has become increasingly clear that native knee brooches are under-represented, except at societies were not slavishly taking whatever Traprain. There is also a distinct native preference Roman goods trickled down to them, but that a for dragonesques. In addition, many of the clear selection process led to high-quality material brooches in native hands are of notably high reaching native hands (Robertson 1970, 200), quality, such as the crossbows from Erickstanebrae where it was then used as prestige goods in local and the Moray Firth (gold and gilt bronze society (Macinnes 1984, 243-4). What story do the respectively) and the silver trumpet brooch from brooches tell? Ayrshire (Robertson 1970). This shows that, as Some general issues of Roman brooches in non- with other artefact types, native preferences were Roman contexts have been considered by biassing the acquisition of certain brooch types. Hedeager (1978), who debates whether they The point is confirmed more clearly in Table 1, represent exchange goods in their own right or where wider comparanda are drawn to the Tyne- simply the minor personal effects of those operat- Solway frontier (Snape 1993, 29-31). Here, to allow ing the Roman-native contact (Hedeager 1978, 204, direct comparison, types are considered relative 208). She concludes that in Free Germany the only to bow brooches of 1st - 2nd century AD date brooches reflect the latter process, seeing them as and dragonesques, thus allowing for the distorting a secondary product of the Romano-Germanic effect of later brooches and ambiguous types (such trade, whose primary goal was the acquisition of as penannulars) on the figures. luxury goods by local chiefs (Hedeager 1978, 209). From this, the question is what made some While the primary goal of chiefs in Scotland was brooches desirable and others not? The very idea apparently similar, assuming a similar role for of wearing a brooch was somewhat alien to native brooches requires more detailed analysis which societies, which showed a strong preference for cannot be attempted here. However, the material pins in the pre-Roman Iron Age with only a scatter can be approached from a different angle, by of early penannulars and a marked sparsity of comparing the frequency of different brooch types. imported La Tune bow brooches (Stevenson 1955, Illus 6 presents this information for Newstead (the 1966; Fowler 1960, fig 8). By contrast, Roman largest Scottish Roman assemblage), Traprain Law brooches clearly enjoyed considerable popularity. (the largest native assemblage), and the rest of the This in itself suggests they had a role beyond that non-Roman finds (subdivided into definite site of Hedeagers everyday necessities (1978, 208); finds and strays), using data from McNaught while perhaps not carrying the social cachet of a (1993), Burley (1956) and Robertson (1970) with Samian bowl or a patera, their apparent popularity additions. Types are given as the percentage of suggests they found a social niche as status

Fraser Hunter 123

Table 1. Relative proportions of major brooch types in northern Britain, expressed as a percentage of the total 1st - 2nd century bow brooches and dragonesques in each assemblage. Raw data from Snape (1993), McNaught (1993), Burley (1956) and Robertson (1970) with additions.

Tyne-Solway Newstead Traprain Scottish non-Roman

Headstud and variants 11% 16% 22% 20% Trumpet and variants 37% 32% 44% 41% Knee 28% 27% 18% 0% Dragonesque 5% 6% 12% 12% n = 326 n=68 n=50 n=59

symbols or identifiers. Seen in this way, the Acknowledgements preferences are interesting, as they show a bias I am grateful to the finders, Ricky Blake, David towards brooches with clear echoes of previous Drummond, K McKenzie-Smith, John ODonnell native decorative traditions: the dragonesque and John , for exemplary reporting of with its highly Celtic shape and ornament; the their discoveries. Thanks are also due to Marion trumpet, perhaps because of similarities between ONeil for the illustrations, to Rob McNaught for its head shape and the trumpet motifs popular in access to his unpublished dissertation, and to Mike local metalwork; and the headstud, which was Bishop, David Clarke, Katherine Eremin, Mark often enamelled, an indigenous tradition (Bateson Hall and Mike King for assistance in various ways. 1981, 7). In contrast, the knee brooch, found only at Traprain, is a Germanic military type with British variants (Snape 1993, 20). The Traprain material in general is harder to understand, with a marked References bias to a restricted range, no early brooches and only one late one. The presence of knee brooches is Allason-Jones, L and McKay, B 1985 C perhaps symptomatic of the high degree of contact W, (Gloucester). with and assimilation of Roman culture on Anderson, J 1905 Notes on a Romano-British Traprain. hoard of bronze vessels and personal Here, in conclusion, may be an interesting ornaments found in a moss on Lamberton insight into native uses of the Roman presence. Moor, , P S A S 39 While alien objects such as Samian and glass (1904-5), 367-376. vessels were willingly adopted as status symbols Bateson, J D 1981 E- I A, for certain display purposes, dress and appearance R S-R B, (Oxford). (key factors in displaying identity) show a Bayley, J 1990 The production of brass in greater conservatism, and it was the brooches antiquity with special reference to Roman with familiar native overtones, not the exotic, Britain, in Craddock, P (ed), 2000 Y which were favoured. Whether this reflects the Z B, 7-27, (London). preferences of a different social stratum for the Bohme, A 1970 Englische Fibeln aus den familiar, or whether it is symptomatic of a pick- Kastellen Saalburg and Zugmantel, and-mix attitude to Roman culture, remains to be S J 27, (1970), 5-20. discussed. Breeze, D J, Close-Brooks, J and Ritchie, J N G It is readily recognised that these conclusions 1976 Soldiers burials at Camelon, are speculative, and further finds may alter the , 1922 and 1975, B 7 figures and the interpretation. A deeper study, (1976), 73-95. encompassing all brooches from Scottish Roman Budd, P and Taylor, T 1995 The faerie smith sites, is needed, along with reappraisal of the meets the bronze industry: magic versus whole corpus of Roman finds on non-Roman sites. science in the interpretation of prehistoric However, this preliminary work does stress again metal-making, W A 27/1 the complexity of Roman / native interactions, and (1995),133-143. the potential of artefactual material to illuminate Burley, E 1956 A catalogue and survey of the them. It is into such a context that stray finds, as metal-work from Traprain Law, P S reported here, must be set. A S 89 (1955-6),118-226. 124 Roman Iron Age metalwork

Close-Brooks, J 1986 Excavations at Clatchard Keppie, L 1990 The Romans in southern Craig, Fife, P S A S 116 (1986), Scotland: future discoveries, G 117-184. A J 16 (1989-90), 1-27. Coles, J M 1960 Scottish Late Bronze Age metal- Leeds, E T 1933 C O B I work: typology, distributions and chrono- A.D. 700, (Oxford). , PSA S 93 (1959-60),16-134. MacGregor, M 1976 E C A N Collingwood, R G 1930 Roman-Celtic art in B, (Leicester). Northumbria, A 80 (1930), 37-58. Macinnes, L 1982 Pattern and purpose: the Collingwood, R G and Richmond, I 1969 T settlement evidence, in Harding, D W (ed), A R B, (London). L P S S-E Corcoran, J X W P 1952 Tankards and tankard S, 57-73, (Edinburgh). handles of the British Early Iron Age, P Macinnes, L 1984 Brochs and the Roman P S 18 (1952), 85-102. occupation of lowland Scotland, P S Curie, J 1911 A R F P I P, A S 114 (1984), 235-249. (). Mackreth, D 1973 R B, (Salisbury). Curle, J 1932 An inventory of objects of Roman Mackreth, D 1989 Brooches, in Brockley, K, and provincial Roman origin found on sites P 1984-5, 87-99, (Oxford). in Scotland not definitely associated with Mackreth, D 1994 Copper alloy and iron Roman constructions, P S A S 66 brooches, in Cracknell, S and Mahany, C (1931-32), 277-397. (ed), R A: S E DES D E S, (Edin- A, 162-177, (York). ). McNaught, R B 1993 A catalogue and study of Evans, D R and Metcalf, V M 1992 R G the Roman copper alloy brooches from the C, (Oxford). fort of Newstead, unpublished BSc Feugere, M 1985 L F G M, dissertation, Dept of Archaeological (Paris). Sciences, Bradford University. Fitzpatrick, A P 1989 The submission of the Miller, G 1857 An inquiry respecting the site of Orkney islands to Claudius: new evidence?, the battle of Mons Grampius, A S A R 8 (1989), 123- S 4 (1857), 19-52. 129. Murray, J E L 1961 Rock-cut symbols in Caiplie Fowler, E 1960 The origins and development of Caves, P S A S 94 (1960-61), 324- the penannular brooch in Europe, P 325. P S 26 (1960), 149-177. Painter, K and Sax, M 1970 The British Museum France, N E and Gobel, B M 1985 T R- collection of Roman head-stud brooches, B T H, (Gloucester). B M Q 34 (1970), 153-174. Hattatt, R 1987 B A, (Oxford). Proudfoot, E V W 1976 Hallowhill, DES 1976, Hattatt, R 1989 A B O A- 33-34. , (Oxford). Riha, E 1979 D F A Hedeager, L 1978 A quantitative analysis of K, (Augst). Roman imports in Europe north of the limes Ritchie, J N G, and Stevenson, J N 1993 Pictish (0-400 AD), and the question of Roman- cave art at East Wemyss, Fife, in Spearman, Germanic exchange, in Kristiansen, K and R M and Higgitt, J (eds), T A M Paludan-Miiller, C (eds), N D I, 203-208, (Edinburgh). S A, 191-216, (Copen- Robertson, A 1970 Roman finds from non-Roman hagen). sites in Scotland, B 1 (1970), 198-226. Hingley, R forthcoming Iron, iron-working and Robertson, A S 1975 The Romans in North regeneration - a study of the symbolic Britain: the coin evidence, in Temporini, H meaning of metal-working in Iron Age (ed), A N R Britain, in Gwilt, A and Haselgrove, C C W 11.3, 364-426, (Berlin). (eds), R I A S, St Joseph, J K 1967 Air reconnaissance: recent (Oxford). results, 10, A 41 (1967), 148-149. Hunter, F forthcoming Iron Age hoarding in Shepherd, I A G 1983 Pictish settlement Scotland and northern England, in Gwilt, A problems in N.E. Scotland in Chapman, J C and Haselgrove, C C (eds), R and Mytum, H C (eds), S N I A S, (Oxford). B 1000 BC - 1000 AD, 327-356, Jackson, R 1990 C: T L I A (Oxford). E R M, (London). Shepherd, I A G 1993 The in Moray, in Keppie, L 1986 S R R, Sellar, W D H (ed), M: P (Edinburgh). P, 75-90, (Edinburgh). Fraser Hunter 125

Sheridan, A 1995 Portable antiquities legislation N B, 17-44, (Edinburgh). in Scotland: what is it, and how well does it Stuart, J 1867 S S S, 2, work?, in Tubb, K W (ed), A: T (Edinburgh). B?, 193-204, (London). Todd, M 1985 The hoard of : trade Small, A 1823 I R A or subsidy?, P S A S 115 (1985), R D F, (Edinburgh). 229-232. Snape, M 1987 Museum note, 1987, A Van Hoek, M A M, and Smith, C 1988 Rock A (5th ser.) 15 (1987), 309-312. carvings at Goatscrag rock shelters, Snape, M E 1993 R B N Northumberland, A A (5th B, (Oxford). series) 16 (1988), 29-35. Stead, I M 1991 I A C E Wace, A J B and Jehu, Prof 1915 Cave Y, (London). excavations in East Fife, P S A S Stead, I M, and Rigby, V 1989 V: T 49 (1914-15), 233-255. K H L S, (London). Wheeler, R E M and Wheeler, T V 1932 R Stevenson, R B K 1955 Pins and the chronology E P, R of brochs, P P S 21 (1955), 282- P-R S L P, G- 294. , (Oxford). Stevenson, R B K 1956 Pictish chain, Roman Whimster, R 1981 B P I A silver and bauxite beads, P S A S B, (Oxford). 88 (1954-56), 228-230. Wild, J P 1970 Button-and-loop fasteners in the Stevenson, R B K 1966 Metal-work and some Roman provinces, B 1 (1970), 137- other objects in Scotland and their cultural 155. affinities, in Rivet, A L F (ed), T I A

A

N - R I A F T . T , R F , C C K C M. A S R . Keywords: Roman brooches, button-and-loop fastener, tankards, Roman-native contact, Iron Age burial, cave use