grant from the Council for British Archaeology Skye and Loch Broom, undertaken specifically

Transcription

1 EUAN W. MacKIE, BA ENGLISH MIGRANTS AND SCOTTISH BROCHS*1) This paper is published with the aid of a grant from the Council for British Archaeology Introduction There arc several theories current about the origin and purpose of the unique Iron Age drystonc forts of Atlantic Scotland known as the brochs. In 1947 Sir Lindsay Scott argued that the great majority nothing were but exceptionally massive defended farmhouses with relatively low walls: the few known tower brochs were an aberrant development. In this view brochs were the end product of the Iron Age roundhouse tradition'2*. Since i960 Mr J. R. C. Hamilton has several times expounded a different idea, based on largely the results of his excavations at Clickhimin(3>. He envisages the progressive heightening of ranges of storeyed buildings along timber the insides of the walls of early timber-laccd and drystonc forts and that this process favoured the development of the high hollow wall to protect them. The final version of this wall was the round broch tower with its many superimposed intra-mural galleries. He believes that this structural development most probably took place Orkney. in Then in 1965 the present writer offered a scheme which saw the brochs developing in the Scottish Western Isles from earlier hollow-walled forts and being taken from there, with further modifications, to the far north'4'. It is the purpose of this paper to review the evidence for the Hebridean origin in the light of some discoveries made since 1965 and to show also that this hypothesis when it is combined with evidence about the Hebridean Iron Age material cultural sequence which has accumulated since 1962 allows the development ofan explanation ofthe historical context of the origin of the brochs. This paper has three main sections. In the first the evidence for the structural development of the later Atlantic Iron Age forts is summarised. The results of excavations on Skye and Loch Broom, undertaken specifically to test the theory of a Hebridean origin for the brochs, arc outlined and a short section discusses Hamilton's multi-storeyed building hypothesis. wooden The second main section considers the sequence of Hebridean Iron Age pottery and artifacts which has been built up in the 1960's and what light this sheds on contacts with southern England in the first century b.c. It also discusses the possibility of the Hebridean wheelhouse farms being a northern version of settlements like Little Woodbury. The final section attempts to integrate the evidence of structural development with that of the cultural sequence to produce a general explanation ofhow and why the brochs developed. Apart from three appendices at the end this paper does not present the mass of factual data to which the discussions refer. Most of this is published elsewhere and references to the appropriate works will be found. This work is intended as a in the sequel to the writer's exposition Proceedings of the Prehistoric Society for 1965 in which most of the evidence for the views expressed in the next section is set out'4'. The Structural Development of the Forts (a) Earlier Views about the brochs Most previous publications have unanimously held that they developed in the area of their greatest concentration, in Orkney or Caithness, and reasons have been given for suggesting that was Orkney the likeliest place for their first appearance'5'. This view was based primarily on the large number of brochs in those zones (the theory being <*> The terms 'English' and 'Scottish' are used throughout this paper in a purely geographical sense. <2> PPS XIII (1047), <3> In Rivet (ed., 1966), pp : in Wainwright (ed., 1962), pp : Hamilton (1968). For these and other abbreviations at end. used in this paper see Bibliography <4> PPS XXXI (1965), : map of brochs on 99, fig. I. <5> Hamilton in Wainwright (ed., 1962), p

2 that the origin of a specialised building must be sought in the areas where the type evidently had the greatest success(6>) and on the suggestion that, since the laminated Old Red sandstone of the far north was such splendid and obvious building stone, it stimulated great achievements in prehistoric architecture'5*. The reasons for selecting Orkney as the place of origin of the brochs were the occurrence there of a type of pottery which was assumed to belong to intrusive 'broch lords' and an inference that the centre of political power was in those islands in the first century A.D.*7'. Yet such views could only be maintained by evading or treating only in generalised terms the problem of the earlier structures from which the brochs must have developed. For example Childe suggested that the timber framework inside some hillfort walls on the mainland might have inspired the lintelled galleries inside the drystonc walls of the brochs and some allied fortlcts in the Atlantic Yet the distribution of the brochs province*8'. and the vitrified forts is more or less mutually exclusive and there is none anywhere near Orkney (none in fact north of the Dornoch Firth) and none among, and few near, the major concentrations of Inner Hebridean in fact con- brochs in Skye and Mull. Orkney spicuously lacks any other types of fortifications besides brochs and has no plausible prototype structures*9'. The excavator of Jarlshof and Clickhimin, J. R. C. Hamilton, suggested an Orkney origin in 1962 without going into details of the broch's earlier development*3'. In 1966 and 1968 he was more explicit and proposed a typological sequence in which a number of gallery-walled fortlets on the west coast and in the Hebrides were the immediate forerunners of the brochs*3'. A western Scottish origin was implied by this but the Orkney theory was not repudiated. It was quite clear that little effective progress would be made in tracing broch origins until the details of the development of the brochs themselves had been worked out on the lines attempted by A. Graham in I947*10'. Graham had analysed separately a number of broch structural features such as internal diameters, occurrence of basal wall gallery (as opposed to a solid base) and the positioning of guard cells and stairways in an attempt to discover any significant regional variations. The internal diameter proved to be most useful for this and a statistical analysis of its range and geographical occurrence showed and the that the brochs of Shetland, Orkney northern mainland were a homogeneous group while those of the Inner and Outer Hebrides and the central mainland were likewise a separate homogeneous group. The brochs of the northern district as a whole differed significantly from those of the rest of Scotland. Also, and in spite of the homo- of the brochs within the northern geneity region, it appeared that those of Orkney occupied somewhat of an intermediate position between the brochs with small internal diameters mainly in the northern region and those with larger courts, concentrated in the west. A tentative conclusion was that 'we may have to deal with two different races or strains in the species broch'*11'. however the informa- In this type of study tion on the size of the brochs' interiors could not be correlated with other aspects of their architecture and the exercise shows clearly the limitations of the statistical analysis of single isolated features in such cases. Only the comparison of assemblages of characteristics could be expected to yield evidence of actual paths of development in structures as as complex the brochs. Yet Graham's analysis was the first demonstration of the way to a potential breakthrough in broch studies and it was not followed up. Subsequent published views on broch development and origins hardly referred to it. (b) The Structural Evolution of the Brochs The first version of the 'broch chart' was published in 1965<4', and showed the range of four variable broch characteristics together pictorial in form for about 120 sites; a revised (6) A. W. Brogger, Den Norske Bosetningen Pa Shetland- Orknoyene (1930), p ' ') Note 5, p. 65. <8> Childe (1935), p < > RCAM (Scotland), Orkney and Shetland. '10> PSAS LXXXI (1946-7), (") Ibid

3 version is shown here (Fig. i). These traits are the internal diameter (external diameter in the first version), the structural type (groundgallericd, solid-based or transitional), the geographical location and the relative massiveness of the wallbascs (expressed as a percentage of the total diameter). The absolute wall thicknesses arc also shown here. The broch symbols arc spread along a diagonal path across the field; those at the lower right have large internal diameters and walls which are thin both absolutely and in proportion, while those at the top left have small courts and much more massive wallbases, both relatively and absolutely. At the same time it is apparent that the ground-gallcried brochs arc concentrated among the larger, slimmer-walled buildings at the lower end of the 'path', while the solidbased towers, though they overlap to some extent with the others, are the sole occupants of the upper end where small courts arc combined with very massive walls. Also the brochs of three regions the Shctlands, the and the Hebrides with the west Orkneys coast occupy distinct though overlapping positions at the upper, central and lower ends of the 'path' respectively, with the groundgallcried brochs being commonest in the Hebrides and the solid-based towers in the northern islands and mainland. The brochs of Caithness and Sutherland do not fall into a distinct group on the chart but are widely scattered over it. The scatter diagram thus shows some clear patterns and it can be as interpreted demonstrating that a gradual change in broch design occurred which was combined with the introduction of the building to several geographical zones successively. In this case either the massive-walled, small brochs ought to be at the beginning of this typological sequence when the subsequent trend would have been towards larger, slimmer-walled structures or the reverse happened. There is nothing to favour the view that the massive Shetland brochs were the earliest. These islands are geographically the most remote part of the broch area (and of Britain as a whole) and are therefore likely to have been among the last places reached by the tower builders. The Shetland brochs are among the most sophisticated of all and can hardly have sprung into existence in Shetland with no antecedents except the 'blockhouse' at Clickhimin(12). Finally the presence there of the best built and most sophisticated broch of all on the island of a Mousa occupying distinct and extreme position at the upper end of the path on the diagram clearly suggests that the ultimate in broch architecture was developed in Shetland. There exceptional height was probably achieved with small courts and massive wallbases. As a corollary the Hcbridcan brochs with large courts and slim, mainlyr ground-gallcried walls should have been the earliest form. This suggested sequence has many implications. For example the development of the solid-based tower in the north from the Hcbridcan forms primary ground-gallcried could be the result of the lack of naturally defensive knolls (common in the west) in the flat lands of Orkney and Caithness and of the consequent need to build higher brochs there. The solid wallbasc is the obvious solution and the clear evidence from the Orkney groundgallericd brochs of Midhowc and Gurncss that these were unstable and began to collapse early in their histories confirms that the western form was unsuitcd to great height. It also implies that this wall was an early development. Again many minor architectural details, such as the distance of the door-checks down the entrance passage and the height and form of the scarccmcnt ledge on the internal wallface, progressive increase of sophistication show a from the western isles to the far north and combine well with the trend towards more massive, solid-based structures*13*. (c) Broch Prototypes If the earliest brochs were built in the Hebrides their prototypes should also be there. In fact a map of the distribution of analogous structures drystonc fortlcts with various arrangements of galleries and chambers in their walls shows that these arc clearly <12> Hamilton (iy6s), p. 54 ff.: MacKic, note 4, 101-2, 126-7, 140. <13> Note 4, 107 ff. 41

4 concentrated in the Hebrides and Argyll with three outliers in Shetland*4'. On the assumption that the specialised broch wall is most likely to have emerged in an area in which a variety of experiments in Iron Age hollow-wall construction were carried out, there should never have been any doubt that western Atlantic Scotland was its birthplace. All these analogous structures used to be called gallcried duns but among them arc a few which possess the exactly the same high, hollow wall as brochs. The name semibrochs has been suggested for this group, obviously closely related to the towers, to distinguish them from the other, less standardised gallcried duns*14'. In the best preserved of the semibrochs the same form of wall as used in the brochs with superimposed intra-mural galleries can be seen, yet they are designed open-sided as structures on the edges of cliffs and precipices. In the past these fortlets have been dismissed as late, peripheral aberrations of the broch tradition on the basis of the undoubted late and post-broch dates (probably after the fourth century a.d.) of the gallcried duns at Kildonan, Argyll, and Dun Cuier on Barra'15). Bur neither of these duns has the specialised broch hollow wall and they arc therefore irrelevant to the dating of the semibrochs. The structure of the Clickhimin 'blockhouse' in Shetland can be interpreted as favouring an early pre-broch date for the Hebridean semibrochs'14'. There is no doubt that the 'blockhouse' itself dates from before the Clickhimin broch and its architectural features seem to show an unique mixture of the local Shetland fort-building traditions (exemplified in the similar 'blockhouse' forts of Loch of Huxter and Ncss of Burgi) with the high, galleried wall and scarccment ledge of the exotic Hebridean semibroch tradition. In other words the non-shetland features of the Clickhimin 'blockhouse' could only have come, at that early date, from structures elsewhere which had the broch hollow wall but which were not circular towers. Only the Hebridean semibrochs could be the source. The theory that the brochs originated in these western fortlets was tested in 1965 by the first excavation on a semibroch, Dun Ardtrcck on Skyc <16', and in 1968 on another on Loch Broom, Ross and Cromarty. At Dun Ardtreck the archaeological evidence, particularly some Roman fragments, showed clearly that the building was destroyed as a fort probably not later than the second century a.d. and it could therefore have been used for some considerable time before this. This semibroch was therefore at least the contemporary of the Hebridean brochs and it could have been built earlier. In October 1967 a radiocarbon date was received for charcoal scraps from the rubble foundation platform of Dun Ardtreck from the Gcochron Laboratories, Cambridge, Massachusetts. They proved to have an age of 2065 dr 105 years or 115 ± 105 b.c. (GX1120; using the best half life of C-14). This means that there is a 22:1 chance that Ardtreck was built between 325 b.c. and a.d. 95. A date in the second or third century b.c. would be perfectly acceptable in the light of the circumstantial evidence on the development of the broch wall set out above and the radiocarbon date is thus entirely in agreement with the archaeological evidence. In 1968 the drystone fortlet Dun an Ruigh Ruaidh, on the south shore of Loch Broom in Ross and was Cromarty, excavated for five described as a weeks'17'. Although originally broch which had partly fallen over the edge of the adjacent cliff<18>, the structure was diagnosed in 1965 by the writer as a semibroch* 19> and this was confirmed by the excavations. An exceptionally clear strati- was unravelled which graphical sequence showed that the semibroch had originally (in Phases 1 and 2) had an internal raised wooden floor resting on a ring of massive posts; the deep, stone-lined sockets for these were found. Eventually its walls were partly dismantled (Phase 3), the internal raised wooden floor <14> MacKie, note 12. <15> Kildonan PSAS LXXIII (1938-9), : Dun Cuier ibid. LXXX1X (1956-7), <18> Ant. XXXIX (1965), 277: the full report is expected to be in Vol. 4 or 5 of this Journal. W Excavations on Loch Broom, Ross and Cromarty: interim Report 1968 (University of Glasgow), pp <18> PSAS LXXXIII (1948-9), (19) pps XXXI (1965), , 124-7, 139. second 4^

5

6

7 was demolished and the interior was thereafter used as a dwelling (Phase 4). Although plenty of charcoal was found at all levels none of it was so unequivocally associated with the construction of the semibroch as the scraps from the rubble foundation platform of Dun Ardtreck. Nevertheless a series of five radiocarbon dates was obtained <20). Only three of them are relevant to the present discussion, those for three samples of charcoal from deposits laid down at the latest at the start of the secondary occupation (Phase 4). Two of these came from the base of the rested on a Phase 4 floor deposits (which cobbled floor, itself directly on the subsoil) and one from the deliberate fill of a post-hole. The two former were 69 ± 100 b.c. (GaK 2496) and 1058 ±110 b.c. (GaK 2497) while the post-hole date was 656 ± 80 b.c. (GaK 2493), all dates expressed in terms of the best half life of C-14 of 5730 ± 40 years. Clearly old wood was lying about the site and the youngest date here must be the most accurate. A date in the first century b.c. is quite reasonable for charcoal deposited during either the primary occupation of the semibroch as a refuge or, less probably, at the very start of its secondary occupation as a domestic site at the latest. It seems more than probable that the semibroch was built with stone from Dun Lagaidh, the ruined timber-framed hillfort half a mile away to the north-west which we now know was burnt and vitrified in the 6th century b.c.(21> and that wooden timbers from the unburnt sections of its walls were also utilised in the Rhiroy fortlet. The results from the two excavations leave little room for doubt that the dozen Hebridcan scmibrochs were built at least as early as the earliest brochs and probably before them. The circumstantial evidence derived from the study of the structure and distribution of all brochlike forts is now in good agreement with the dating evidence and there is no longer any reason for denying that the likeliest explanation is that the freestanding broch towers were developed in the Western Isles out of the scmibrochs, probably on Skye where the majority of both types of are buildings situated. (d) The Multi-Storeyed Building Hypothesis The final section in this brief review of recent advances in our understanding of the Atlantic stone forts considers a recent alternative explanation by J. R. C. Hamilton of how the broch hollow wall actually deve- here this wall loped<3>. In the scheme presented was invented on Skye and gave rise first to the semibrochs. The conversion of the opensided scmibrochs to the circular tower form occurred later and this is a separate problem, discussed further below. Hamilton believes that the high gallericd wall was originally developed because of the need to protect ranges of storeyed wooden buildings, which were erected against the inside faces of the walls of the earlier Atlantic stone forts, from being set on fire by brands thrown over the wall an during attack. In this view these wooden buildings gradually became higher, with first one and then two raised floors. The need to heighten the stone rampart to protect them resulted in the development first of a high parapet on the wallhcad which was then converted to a completely enclosed upper gallery with a raised parapet on top of that. Thus, the theory goes, was the high hollow wall devised and the ultimate product was of course the broch when the fort builders found that their new device could be raised far higher than was strictly necessary, to build an impressive tower. The implications of this view of the origin of the brochs are similar to those of the hypothesis set out here that the broch wall and the brochs themselves developed in the Hebrides. This is because, except for the Clickhimin 'blockhouse', all the fortlcts cited as immediately pre-broch forms in Hamilton's sequence are Hebridcan. However the earlier theory that the brochs originated in Orkney and that the Hebridcan ones were offshoots from these(5> was reiterated in the more recent articles. The multi-storeyed wooden building theory clearly depends for on proof the existence in both the brochs themselves and in their <20> DES (1969), <2» Ibid

8 progenitors of an upper inside wallfacc not more than a few feet above scarccment on the the lower one. Only this could prove the existence of a building with more than one raised wooden floor, on which the theory depends. Yet none of the assumed pre-broch forts has evidence for such an upper scarccment, except perhaps one; several of them arc too ruined for even the lower ledge to be preserved. However it can be argued that the suggestion (by no means certain) of an upper ledge on the Rhiroy scmibroch on Loch Broom means that the remainder had them too<22). For clear proof we must turn to the brochs as the final stage in the suggested evolution both of the interior, wooden storcyed buildings, and of the hollow, protecting wall. Many brochs arc preserved in parts than 15 feet in height to more and several to more than 20 feet. In none of these is there any trace of an upper scarccment between five and ten feet above the lower one except at Mousa. At Dun Telve the only other known upper scarccment is 29A feet above the ground and some 24 feet above the lower ledge: it can only have been for a roof. It is fair to suppose that this was in most brochs and that originally the case there was nothing between the lower scarccment with its wooden floor and the upper one supporting a light frame roof(23) near the top of the tower. The upper ledge at Mousa is explicable as one of the many ways in which that remarkable structure shows how normal broch architectural features were blended form. It together there in unique and perfect seems a logical ultimate refinement that the roof of the tower should be brought down from the wallhcad to a short distance above the raised wooden floor. This would allow it to be reached more easily for repairs and give better protection from wind and storms. The demonstrable absence of evidence of a support for a second raised wooden floor in all the better preserved brochs except Mousa is a grave objection to the multi-storeyed building hypothesis which is seen to rely for proof entirely on one abnormal structure. An alternative explanation of how the broch high hollow wall can developed be framed and it will appear in detail in the excavation report of Dun Ardtreck scmibroch(1(i). (c) Summary The structural evidence supports the theory of the evolution of the design of the brochs in which the ground-gallericd Hebridcan forms are the earliest. This was design then trans- where the solid-based form ported to Orkney more appropriate to the flat, open, northern terrain probably emerged. In the Hebrides there are suitable prototypes for the brochs in the small group of scmibrochs which have the same specialised wall but which arc more primitive in their design and situation. The only two scmibrochs which have been excavated, specifically to test the theory, proved to be of the required early date'24'. Thus ifthe brochs were invented in the Hebrides, probably Skye, on it is to the contemporary material cultures of the Western Isles and not, as hitherto supposed, of Orkney that we must look for further clues about this remarkable development in Iron Age fort building. The Hebridean Material Cultures (a) Introduction and Early Views Since Childe's The Prehistory of Scotland appeared thirty-six years ago and for the first time drew the Scottish Iron Age cultures into the English and Continental European framework, it has been accepted that certain elements in the material culture of the Hebridean Iron Age were derived directly from the south of or England from even further away in France. Childc himself thought that the long-handled bone combs and other 'weaving equipment' which had been found in some northern brochs and Hebridean whcclhouscs were so similar to those of the Glastonbury 'lake village' in Somerset that they must imply that migrants from the Iron Age B settlements of south-west had come to live on the Western England (22> PSAS LXXXIII (1948-9), "3> Note 10, 70. <2J> Ant. XLIII (1969),

9 Islcs(:i5). pottery was quite different from that of the lake villages and that it had closer similarities with Bronze Age Shetland wares and with A and Continental Hallstatt However he thought that the local English Iron Age vessels. This apparent dichotomy of origin in the Hebridean equipment helped him to suppose that aborigines of Bronze Age and Iron Age A descent had been subjected by intrusive Iron Age B 'castle builders' from south-west England. Further evidence in support of this idea came from the sophisticated diplomatic behaviour of the Orkney chieftains towards the Roman Emperor Claudius in a.d. 43, from which he inferred from the that they were recent immigrants south who had played an important part in the development of the brochs*28'. In 1948 Sir Lindsay Scott attempted to define the south-western origin of the intrusive population of the Hebrides more closely and thought that the pottery did demonstrate clear links with that area*27'. He supported his views with his excavation of the Clcttraval wheelhouse, the first time one of these structures had been subjected to systematic study. He considered that the builders of the wheclhouses from which he thought the brochs had ultimately developed had arrived from the south-west in family units. Further progress in the identification of these supposed immigrants was made in 1952 by C. M. Piggott in her discussion of her excavations on Bonchcstcr Hill in Roxburghshire. This fortified site had produced, in common with a few others, bronze pins and rings of a type found only in south English hillforts, and a migration from the Wessex area was inferred to have taken place at some time between about 25 b.c. and a.d. 50. These dates were based upon the occurrence of similar bronzes in the Maiden Castle hillfort sequence'28'. She thought that these migrants also introduced to southern Scotland the new multi-vallatc hillforts as well as the timber-laced (vitrified) forts north of the Forth, customarily grouped into the Abernethy complex*29'. However by 1961 the dates of the Iron Age levels at Maiden Castle had been revised backwards*30', and the idea had been questioned that it was necessarily the latest occurrence of an artifact type in south England which set the earliest limit for its appearance in the north*31'. As a result it was possible to suppose that south English settlers had arrived early in the second century b.c. The bone weaving equipment and dice of the Atlantic province were thought to be a little later and presumed to show a parallel northward migration. An alternative view was offered in 1962 and 1963 in which it was suggested that the exotic bronzes and bone equipment of southern Scotland and the Western Isles could plausibly have come from the Iron Age B cultures of Yorkshire*32'. Several useful excavations have been done in the Western Isles since Scott studied the problem in 1948 and the most recent of these was on the first broch in the area to have been examined for more than forty years, at Vaul on Tiree. Previously several wheelhouse excavations, and that of Dun Cuicr on Barra, had allowed the construction of a fairly complete sequence of pottery and artifacts from about a.d. 100 to 400 or 500 but the position of the western brochs in this was unknown. There was good evidence that the brochs were earlier than the wheclhouses the latter being in part derived from them'33' so it was clearly important to find a Hebridean broch with a long history of occupation which would precede and link on with the wheelhouses and later phases. Fortunately Vaul had such a long occupation which also extended back into the pre-broch era. (b) The Pottery The excavation of the Vaul broch produced a mass of stratified pottery among which several distinct cultural elements can be seen*34'. <'-5> Childe (1935), pp <2C> First recognised by Childe, Scotland before the Scots (1946), p <"> PPS XX (1948), 60 ff. <28> PSAS LXXIV ( ), 129 ff. <29> Childe (1935), pp (SO) prere (ed., 1959), pp <") Elizabeth Fowler in PPS XXVI (i960), <32> a. C. Thomas in Arch. J. CXVIII (1963), (33) Note 4, , <34> Ant. XXXIX (1965), : the full report is to be a monograph published by the University of Glasgow

10 The remains of two consecutive pre-broch occupations were found and also deposits associated with the actual construction of the tower (as opposed to its primary occupation deposits, also found) which contained fragments of pots which had presumably belonged to its builders. In these primary broch levels were sherds of a new style of Hcbridcan pottery, hitherto absent, which consisted of globular or situlatc jars of dark grey ware with everted rims and decorated with an applied cordon about half-way between rim and waist which had been impressed into a zig-zag pattern (Fig. 2). Sometimes there are concentric channelled arches above the cordon. This curvilinear decoration, which is found in Scotland only in the Hebrides, should be connected with the ornamented Iron Age B pottery of southern or eastern England as Sir Lindsay Scott thought'27'. However this Clettraval ware is already purely a local style and can be matched in the south only by Scott's method of separating out its formal and decorative motifs and treating them individually and in isolation'35'. It seems more likely that the inspiration for the concentric arches in Clettraval pottery was provided by the simple, eyebrow-ornamented, bead-rimmcd bowls of the Iron Age B culture of Wcssex rather than by the much more elaborately decorated Glastonbury bowls'36'. This view is supported by the discovery, in 1966, that a group of sherds from inside the Vaul broch can be reconstructed into a local copy of just such a Wessex bowl. Although this vessel (Fig. 2, 1), measuring only 3.4 inches in height'37', is made of a fabric indistinguishable from hundreds of other Hebridcan evcrtcd-rim sherds found at the site, it was unique in three ways. It had probably been made with an everted rim but this had been carefully trimmed off after so firing that the rim was left plain with the exterior groove resembling beading. Its curvilinear decoration was of single arches, whereas Clettraval sherds have sets of two or more without exception done concentrically. It also lacked an applied cordon below the arches whereas every other everted-rim sherd, with or without arches, which was sufficiently well preserved showed this feature (except one: Fig. 2, 2). The Vaul bowl is the third of what might be called 'memento pots' that have been found on Tirce unique miniature versions ofceramic styles quite alien to the Hebrides. The other two seem to be small copies of English Iron Age A styles at Vaul and Balevullin'38' and all three could be as explained mementos made for new settlers recalling the styles used in the abandoned homeland. Such things arc known only in the Atlantic Zone of Scotland in the Iron Age where there had long been a tradition of fine pottery making. The eyebrow-ornamented bowls are fairly common in Iron Age B contexts in southern England and are usually dated there to the second and first centuries b.c., in late pre- Bclgic times. The latest work on the Glastonbury pottery sequence has suggested that the elaborate curvilinear-ornamented pottery there was mainly used in the first century b.g. <39> but the eyebrow bowls were presumably current earlier in regions further east where 'prc-belgic' now indicates a date not later than cither the early first century b.c. or perhaps the late second century'4"'. The stratigraphical position of the 'Wcssex bowl' in the Vaul sequence is fairly clear. It was found inside the broch, more or less on the published section'41' in Layer 6, the deposits of the earliest users of the tower (Phase 3A). It was probably broken cither at the beginning or end of that phase. From its very uniqueness and its close resemblance to a foreign style one would expect it to have been made very soon after the arrival on Tiree of whatever people had this knowledge of south English pottery and this arrival may of course have been much earlier than the time it was broken. In fact the circumstantial evidence suggests that the bowl was older than the broch because Clettraval (35) Note 34, <3«> Ibid. 27s. <37> Note 24, PI. Ve. <3S) The Vaul pot note 34, 269, fig. I, no. 17: Balevullin PSAS XCVI (1962-3), 165, fig. 2, no. 1. (39) Michael Avery in Proceedings of the Somerset Archaeological and Natural History Society, CXII (1968), (40) prerei note 30, p. 84 ff. <"> Note 34, fig

11 inches Fig. 2. Pottery from the Dun Mor Vaul broch, Tirce, which shows the influence of Wessex Iron Age li ceramic traditions. No. i is the 'Wessex bowl'. sherds appear fully developed in the primary broch levels. We can now assume that this local style developed only after the impact of Wessex B influence and some presumably time must have elapsed between its arrival and its blending with the local wares to form a standard new pottery style. There is in fact some evidence (infra, p. 6i) of a slightly earlier phase in the formation of Clcttraval pottery at Vaul<42>. At this point it may be useful to describe what is known of the earlier history ofcvertedrim pottery in Atlantic Scotland, to place in perspective the incorporation into it ofwessex B decorative motifs. There was hardly any direct evidence about this until very recently, a state of affairs which favoured Sir Lindsay Scott's view that the entire ceramic tradition had arrived from the south at the same time in the first century b.c. Yet it is now clear that evcrtcd-rim pots were in use in Atlantic Scotland long before that. A large one from Vaul with a double cordon was radiocarbondated to the fifth century b.c.(24) and many sherds of evcrted-rim pots with internal fluting (but without cordons) were found with the first pre-broch fort at Clickhimin, Shetland143'. It is not yet possible to analyse in of this ceramic tradition but detail the origins it may go back to the Late Bronze Age Urnficld cultures of France in the seventh century b.c.'44'. The lack of information about it in (4-> Traces of an immediately pre-broch settlement were found with a few cruder Clcttraval sherds. <43> Rivet (cd., iy66), pi. 4, p. 121: Hamilton (iy6s), pp figs (") Note 24, pis. Va and Vb. 47

12 the centuries between the Late Bronze Age and the broch period may be partly due to the fact that the relevant sites have not been discovered. At present it seems that the applied cordon was added to these pots at an early stage, and only in the Hebrides, and that this motif was derived from an even older local pottery style*45'. Thus the Wesscx B decorative clement may into a well established have been incorporated Hcbridcan tradition of cordoned, everted-rim jars and, if future discoveries confirm this, it could mean that there were, in addition to the aborigines who used incised a pottery*34*, number of people of Urnficld origin already in the islands with whom any new arrivals from southern England were apparently able to mix amicably. There are two other sherds from Vaul which show strong influence from Wcssex Iron Age B wares. One is the rim and neck of a storage jar which is well matched in the south*46' while the other seems to be an early form of Clcttraval ware, closer to the supposed Wesscx original, with a shorter everted rim, single joined arches and no apparent cordon (Fig. 2, 5 and 2). Nevertheless it must be remembered that the amount of reliable information about the Late Bronze Age and pre-broch cultures of Atlantic Scotland is still extremely small. No cverted-rimmed pots of the Clickhimin variety were found in the undisturbed Phase 1 levels at Vaul, which probably started in the sixth century B.C.*24', and the major exotic decorative trait found on the sherds from these deposits (bases with interior impressed fingermarks) also has a parallel at Plumpton Plain in Sussex*47'. Also of course no Urnfield cemeteries are known from the Atlantic province. The tenuous evidence of the links with France does depend entirely on a small number of decorated sherds whose significance may be quite different from that suggested earlier. Yet attention must be drawn to them. (c) Glass Beads Other artifacts in Atlantic Scotland were certainly brought somehow from southern England, though the exact time at which they 4«or the idea of them arrived is inevitably obscured by the scarcity of known pre-broch sites. Two main kinds of glass beads appear small rings mostly of yellow paste, and larger ones of coloured glass with a design of inlaid vitreous paste. The small ones can be of other colours red, blue and green while the colour of both the glass and inlay of the larger ones can vary. One of the small yellow ones was dropped when the Vaul broch was constructed, if not slightly before, and a large inlaid bead was also lost fairly early in its history. A whole collection of tiny ring beads not only yellow but of various other colours came from the destruction levels inside Dun Ardtreck, Skye, and can be reasonably by dated Roman fragments to the mid or late second century a.d.(16). This semibroch had been built somewhat earlier, probably in the second or first centuries B.C.*24', so the beads could have been on Skye earlier too. Two yellow beads came from strata associated with the first and second pre-broch forts at Clickhimin (48). Both the tiny ring beads and the larger inlaid ring beads arc found in Iron Age B contexts in southern England, particularly in the Somerset Lake villages*49'. The yellow beads have been found sporadically in the Wcssex area though rarely can any examples be demonstrated to be earlier than about the of them first century b.c.(50). The majority seem to belong to the immediately pre-bclgic and pre-roman Iron Age. There is a possibility that they were made in Scotland after the initial importation*51'. (d) Bronze Rings, The appearance in Scotland during the Iron Pins and Fibulae Age of various types of English bronze pins, <45> At Balevullin note 38, 172 and fig. 3, nos. 26, 27. <46> For example at Little Woodbury, Wilts.; PPS XIV (1048), 6, fig. 2, no. ia; 8, fig. 4, nos. 1 ft. <47> PPS VII (1935), 54, fig. 13, G. <48> Hamilton (1968), p. 80. <49> A. Bulleid and H. St. G. Gray, The Glastonbury Village Lake II (1917), pi. lix (frontispiece): also at Maiden Castle Wheeler (1943), p. 292, fig. 98, nos. 1 and 2. (50) por example at Condcrton Camp hillfort, Gloucester, such a yellow bead was found in an Iron Age B context N. Thomas in Proceedings of the Cotteswold Naturalists' Field Club, XXXIII, Pt. 3 (1959), <«) C. M. Piggott in PSAS LXXXVII (1952-3), 104; John Smith in PSAS LIU ( ), 123 ff., especially

13 WESSEX IRON AGE B' IN SCOTLAND SPIRAL FINGER RINGS ribbon & wire; after Jope z U.J.A.,1957, with some additions for Scotland. EYEBROW-ORNAMENTED BOWL, (Tiree). mm MAIN ZONE OF ENGLISH-INFLUENCED POTTERY (Clettraval) AREA OF WOODBURY FARMING ECONOMY signs of grain storage _(after S.PiqgotQ Map of the distribution in the British Isles of spiral finger rings ot bronze wire and ribbon. Also Fig. 3. marked is the area of the main concentration of the Wessex B-influenced Clettraval ware in Scotland and of the Woodbury farming economy in southern England (from grain storage pits).

14 rings and brooches provides a striking demonstration of links between the north and south of the British Isles. Though this evidence can hardly be used by itself to demonstrate the presence of immigrants it is a valuable addition to the other material mustered here, of the various particularly since the dating bronzes in southern England is fairly well known. The bronze objects arc the earliest penannular fibulae (Type Aa), the La Tenc I fibulae, the ring-hcaded pins and the spiral finger has rings. The significance of these objects been frequently debated and no useful purpose is served here by going over the ground in detail, particularly since R. B. K. Stevenson has recently examined the subject'52'. The situation as understood at present seems broadly to be as follows. Both types of fibulae and the pins are found no further north than the south Scottish mainland one or though two ring-hcaded pins (not to be confused with the North British form with a pro- peculiarly jecting head) have been found in the Hebrides, on Coll and Skye(53>. The spiral rings arc also found in the southern mainland but they occur frequently in the Atlantic province, several times in association with brochs*54'. arc the clearest evidence of the arrival in They Scotland of people from southern England, probably by sea, since the map of their distribution shows them concentrated in these two regions with hardly any between (Fig.3). Thus, of the four important English only bronzes the spiral finger rings are unequivocally associated with the other exotic elements in the Hcbridcan broch culture. No La Tene I fibulae and no ring-headed or crook-headed pins have been found in such contexts. Stevenson has recently argued from other evidence that the fibulae belong to an earlier chronological horizon than the spiral rings and the evidence from south England supports this. Thus the spiral rings now stand alone, their clear evidence of contact with the south no longer blurred by the other material some of which, for example, can be linked with Yorkshire*32'. The Scottish ring-pins with projecting heads are also found both in mainland and in Atlantic Scotland Iron Age now seem to have a much contexts but they earlier origin than has been supposed. They are not so likely to be derived from English pins as to be a peculiar northern development, perhaps originating sunflower pins of the Late Bronze Age*24,55'. (c) in the Swan's Neck Bone Artifacts Bone artifacts provide further evidence of the northward movement of English cultural combs are traditions. Long-handled weaving perhaps the best known and arc among the most useful since they surely indicate the arrival of womenfolk: it seems somewhat could learn this improbable that aborigines weaving technique except by direct example. Though the English combs tend to have longer handles in proportion to the length of the teeth than the Scottish, the late appearance of the latter (they are found on Iron Age A sites in the south) suggests that they were imported. No long combs were found among the well preserved boncwork from on Dunagoil Bute nor have they been recovered in any other vitrified fort. Their absence from Dunagoil is particularly interesting because of the presence at that site of some English ring-headed pins, a spiral finger ring and a La Tenc Ic safety pin of Wesscx type*56'. No long combs have been found in any clear pre-broch context though three sites with suitable stratigraphies are known in Atlantic Scotland Vaul, Jarlshof and Clickhimin. No comb was found anywhere at Vaul, though bone was well preserved there. A comb came from Jarlshof in the late nineteenth century excavations but probably belonged to a late phase. Clickhimin yielded a long comb from the primary broch levels'57' and others have turned up in what (**) Stevenson in Rivet (ed., 1966), pp (53) Coll PSAS XV (1880-1), 152-3, and 81, fig. 3 : Skye D. D. A. and M. S. Simpson in TDGAS XLV (1968), 141 and pi. II. <54> In secondary contexts in the Vaul, Gurness and Crosskirk brochs and in the primary broch stratum at Clickhimin (Appendix A). <i5> J. M. Coles in PSAS XCII (1958-9), i-9- <56> I have to thank Mrs Margaret Peek (nee Lecchman) for letting me sec her Edinburgh M.A. thesis on Dunagoil. The spiral finger ring was assembled from fragments in the Glasgow Art Gallery and Museum. <"> Hamilton (1968), p. 117, fig

15 parallelopiped BONE DICE o ((gj 3 or more ^^--^1/ i^^tsp -1 \l«<f Fig. 4. Distribution in Scotland of parallelopiped bone dice.

16 must be early secondary broch contexts at the Midhowe and Ayre brochs in Orkney- Several have been found in Outer Hcbridcan wheclhouses which certainly belong early post-broch period'33'. to an The Clickhimin long comb is the earliest example known in Atlantic Scotland and shows that these implements were associated with the influx of new cultural traits which seem to accompany the arrival of the brochs. Childc had previously assumed this in 1935 but no well stratified bone combs were known at that time. Long-handled combs are commonly found in Iron Age A and B contexts in southern and in eastern England and were indeed taken to be part of the standard equipment of Hodson's 'Woodbury aboriginal population which, Culture' the he suggests, formed the majority of the people of England throughout the Iron Age (apart from the two known invasions of Marnians in Yorkshire and Belgac in the south-east). Other characteristic type-fossils of the Woodbury culture include roundhouses, ring-headed bronze and iron pins and a general absence of burials, all of which contrasts with the European Iron Age cultures'58'. However it can be argued that some of the long bone combs filtered up from the Yorkshire area'32'. In the Late Woodbury cultures of the south the same type-fossils occur but there is increasing evidence of contact with the late La Tene cultures of Europe in the appearance of new types of fibulae, rotary querns, hillforts with inturned entrances and bone dice. These in the north and so far dice (Fig. 4) also appear only in Atlantic Scotland, though until recently their relative position in the Iron Age cultural sequence there was not clear'59'. Now wc know that a bone die was found in the primary broch levels at Clickhimin so these too should belong to an early stage in the broch period. Fragments of two were found at Vaul after the excavations there were completed'60', so it may be that the lack of typical 'broch' bone equipment at that site was not so significant as it appeared at first'61', and that the long combs and dice appeared with the brochs everywhere (see Postscript, infra). (f) Quemstones Rotary querns provide one of the more vivid illustrations of the dual process by which the Iron Age population of Scotland probably acquired new ideas and techniques first by adopting them from southern neighbours and, second, through their introduction by actual immigrant settlers. Rotary querns came first to southern England, introduced from the continent perhaps in the third century B.C., as one ofthe several innovations which broadly characterise the later phases of the southern Iron Age. Previously, and ever since the introduction of agriculture in the fourth millennium B.C., corn had been ground on the saddle quern. The earliest forms of rotary querns were the tall 'beehives' in the Sussex and Wessex varieties, which were turned by a horizontal handle inserted in a socket in the side of the upper stone'62'. These dome-shaped querns seem to have been spread gradually northwards among the agricultural communities of the midlands, northern England and southern Scotland. Examples occur on palisaded sites like Huckhoe in Northumberland and Harchope in Pecblesshire, settlements which otherwise have little evidence of anything but a firmly local and aboriginal material culture going back to the Late Bronze illustrate well how a new Age'24-63'. They artifact could spread by example without any evidence of population movements. There can be little doubt that this clear and obvious fundamental to a improvement in a technique peasant economy that of grinding grain could have spread quite rapidly among communities otherwise notoriously conservative. These early beehive querns, and their flatter or 'bun-shaped' derivatives, were taken as far as the Forth-Clyde valley in central <58> F. R. Hodson in PPS XXX (1964), <59> For example Stevenson thought that the dice might be post-broch PPS XXI (1955), 293- (») DES (1966), 10. <»i) Note 4, 122. <62> E. C. Curwen in Ant. XI (1937), 140 ff. <«) Huckhoe G. Jobey in Arch. Aei, 4S XXXVII (1959). 271, fig. 14: Harehope R. W. Feachem in PSAS XCII ( ), 185, fig

17 ROTARY QUERNS beehive series: lateral handles 7. tall beehives 2. bun shaped 3. probable " disc series: upright handles f 4. flat disc 5. slightly convex 4 or 5 >1 4 Fig. 5. Distribution in Scotland of two types of rotary querns: the lateral-handled beehive- and bun-shaped series and the disc querns with upright handles. Only those of the latter which have been found on known Iron Asre sites have been included.

18 Scotland but not much beyond (Fig. 5); only one is known from north of the Tay (Appendix were C). They well established when the Romans occupied southern Scotland in the late first and second centuries a.d. and several are known from Roman sites *64'. The clear second century date for these latter stones was one of the main reasons for the persistent dating down- of Scottish Lowland Iron Age sites which was in vogue in the late 1940's and the 1950V65'. But of course the fact that the stones were still being used when the Romans came does not preclude their having arrived two or three centuries earlier; even less does it mean that the native sites concerned were not established long before knowledge of the rotary quern was brought to them. The recent radiocarbon dates have provided a badly needed correction of this late dating(24>. The conservatism of these north British communities is shown by the late survival of the earliest type of rotary quern, with the lateral handle hole. The Huckhoe series demonstrates a very gradual development from the original tall beehive form to a broader, flatter, bun-shaped stone in surface which was also flatter. Huckhoe the grinding indeed had one or two which were practically flat discs yet still with the horizontal handle*63'. The flattening and broadening of the upper stones seems to have been an improvement which took place gradually and independently in several of the areas where the beehives were introduced. However, whereas in the north the horizontal handle persisted even in the near-discoid forms at Huckhoe, in southern England the late prc-bclgic querns developed into thin discoid forms with vertical edges and a large diameter. Those at Maiden Castle must have had vertical handle holes on the upper surface instead of in the edge (they were not preserved on the broken stones illustrated)*66'. It was thought that a similar development took place among the Scottish rotary querns, that the laterally placed handles of the flatter, bun-shaped stones eventually moved up on to the top surface to a vertical position in development parallel a to that in southern England*62'. The many querns found in Atlantic Scotland, on broch and wheelhousc sites, all have vertical handles; many arc slightly bun-shaped and seem to confirm the supposed derivation from the south Scottish forms. The most westerly bun-shaped beehive querns known are from Bute and, though vertically handled disc querns arc known from Iron Age sites on the southern mainland, the majority are from the Atlantic province. As long as the primary material culture of the brochs was unknown there was no reason to suppose that the rotary querns found in broch contexts and in wheclhouscs secondary were other than contemporary with or later than the second century bun-shaped querns of south Scotland; Curwen's unilinear theory of northern quern development could therefore stand. However the earliest rotary quern found at Vaul on Tiree belonged to a slightly prc-broch phase and is most unlikely to be later than the first century B.C.; it probably belongs to the second quarter of that century (infra, p. 59). It is therefore substantially earlier than the bun-shaped querns on the south Scottish sites. Now this earliest Vaul quern was a true disc form with parallel upper and lower faces and an upright handle-socket on the top. The other, later querns from this site had slightly convex upper surfaces but Since no none had handle holes preserved. bun-shaped querns with lateral handles are known from the broch area, and since such bun-shaped querns were definitely in use on the mainland for one or two centuries afterwards, the early Tiree quern can be plausibly as an explained only introduction of the late Iron Age B model directly from southern England. It provides valuable confirmation of the impact of the more advanced southern cultures on the hitherto technologically backward Hebrides and surely explains why all the Atlantic rotary querns were the later form with upright handles. The upright-handled disc quern did eventually filter into southern Scotland, presumably (64> For example at Newstead, Castlccary and Oakwood: see Appendix C. <66> C. M. Piggott in PSAS LXXXII (1047-8), 220 ff, and in ibid. LXXXIV ( ), 129 ff. (66) Wheeler (1943), p. 328, fig. 117, nos. 27 and

19 from the west. Significantly some were found in the Castlchill Wood dun and in the Tappoch broch, both in Stirlingshire: these two sites arc perhaps unlikely to have been occupied before the second century a.d.<67). Duns are a characteristically western Atlantic form of tinystone fort and the presence of a Hcbridcan form of quern in one need not be surprising. Tappoch is structurally a typical Sutherland broch and here an again Atlantic form of rotary quern might be expected. The souterrains of Angus belong perhaps to the second or third centuries a.d. and disc querns were probably becoming more widely adopted in the area by then*68'. A flat disc quern was on already the site broch was built*69'. before the Torwoodlee (g) Triangular Crucibles Several broch and whcclhouse sites have yielded a distinctive type of thin-wallcd, triangular crucible made of fine-grained, hard, light grey clay. Exactly similar crucibles were found at the late prc-roman 'lake villages' at Glastonbury and Mcarc in Somerset*70' and it seems probable that the crucibles were brought to the north with the other Iron Age B artifacts being described here. Few have been found on scientific excavations: one came from the Phase 3 levels at the wheelhouse of A' Cheardach Mhor on South Uist and several from the secondary levels at the Vaul broch on Tiree. Though none has yet been found with primary broch material the many found with Wessex-influenccd sherds at Vaul do suggest that they arrived as part of that influence. There is a little evidence about the type of crucibles in use in the prc-broch era and they were not the triangular ones. All those found at Clickhimin in Shetland from the earliest fort to the wheelhouse periods arc described as of the 'pipe bowl' type, evidently triangular*71'. not A number probably dating to the seventh century B.C.*24' were found on the mainland in the occupation stratum associated with the palisaded enclosure at Wood in Renfrewshire and I Craigmarloch am grateful to Miss Helen Nisbct for allowing mc to examine and comment on the fragments in advance of publication. Although many of the surfaces were of a coky grey texture and thus superficially similar to the triangular crucibles it was clear from several examples that they had in fact been made of the same yellowish clay containing occasional pieces of as gravel had the many potsherds in the same layer. The grey skin had been produced by the intense heat to which they had been subjected. None of them could be seen to have had a triangular shape. Woodbury Farmsteads (h) There may be more evidence for the appearance of new farming techniques in Scotland than that of the rotary querns described earlier. It has been shown that there existed in southern and south-eastern England, from Iron Age A or even late Bronze Age times onwards, an unusually efficient corngrowing economic system based on the single, isolated farmstead which was capable producing of a considerable surplus of grain each season. Basket-lined storage pits in the ground provide the evidence for this surplus and it has been pointed out recently that Dr Bersu's original calculations about the capacity of the pits at Little Woodbury itself were wrong and that he underestimated the annual yield of the farm by a factor of about nine*72'. Clearly such an efficient cereal-growing economy must have been partly a response to the peculiar environment of southern England where there arc many hundreds of square miles of chalk downs and other light, well drained soils which the early Iron Age farmers were able to till with the primitive ards which one assumes were the only ploughs they had. One of the most northerly comparable farms must be Staple Howe in Yorkshire which was a stockaded <"7> Castlchill Wood Feachem in PSAS XC (1956-7), 24-51: the Tappoch quern, ibid. 38. (68) p_ -p. Wainwright, The Souterrains of Southern Pictland (Iy 3). P- 133, fig- 37, and p. 143, fig. 38. <"»> S. Piggott in PSAS LXXXV (1950-1), 109, fig. 10. <7"> A. Bullcid and H. St. G. Gray, The Glastonbury Lake Village I (1911), p. 304, pi. xlix: H. St. G. Gray and A. Bullcid, The Mcarc Lake Village II (1953), pp and fig. 66. (7,l Hamilton (1968), 80, 90, 116, 133, etc., and pi. XXXIb. <72> Ant. XLI (1967),

20 site on a knoll with a subsidiary building interpreted as a raised granary. It too stood on chalk land. The distribution of the Woodbury farming economy was (Fig. 3) plotted primarily from such tangible evidence of surplus corn production, mostly storage pits<73>. Most lynchcts provide parallel evidence for contemporary field systems. Even if it could be shown by other means that Iron Age farmers moved to Scotland from the corn-growing zone of southern England, it is unlikely that they were able to set up precisely similar farms in the new environment. In the highlands and islands the surface is quite different; much of the cultivable land is acid, peaty soil or, in the Hebrides, inachair or turf-covered sand. The only terrain in Scotland comparable to that of southern England is the area of dry grass-moors in the south-cast in the Tync-Forth province where there are indeed a large number of circular stockaded houses with traces of lynchets near a very few <74>. However, none of the few sites excavated there has yielded evidence either of storage pits or or granaries of cultural links with southern England; in any case the Tyne-Forth palisaded sites now seem likely to be at least as old as the sixth or seventh centuries b.c.(24). Two factors have tended to confuse the study of the possible introduction of Wcssex methods into Scotland. The first is the farming existence of these numerous early palisaded sites in the Tync-Forth province together with the lack of artifacts in them which makes it difficult to link them with other culture complexes. Unless evidence of granaries or storage pits is found with them there is no reliable way of linking such sites with south English farms this was also a round wooden house set inside a stockaded Hence a settlement in Scotland similar in like Little Woodbury even though yard. plan to Little Woodbury could scarcely be shown to have been derived from the south other Wcssex B cultural unless itproducedsomc traits. Add to this the differences in environment which must have meant that any migrants could not in any case have reproduced their Wcssex farming system exactly in the north and the difficulties are obvious. The other difficulty is that the term 'Woodbury Culture' has more recently also been applied to the assemblage of native as opposed to immigrant elements of the material culture of the Iron Age population England. of These include the roundhouse, the long weaving comb, the bronze ring-headed pin and the absence of concern with burial all characteristics which contrast with the contemporary continental cultures*58*. The corn-growing economy was obviously part of the Woodbury culture as thus defined in the south but need not have been in other where the environment was parts of England different. The use of the same cultural label to describe an economic system and an assemblage of artifacts is potentially confusing. However there are two Hcbridcan dwelling sites, one with markedly Wesscx-influcnced pottery, which look very like Little Woodbury farms. arc They Clcttraval on N. Uist<27) and Allasdale on Barra(75> and each consists of a stone roundhouse with internal freestanding radial stone piers instead of the roof-posts of the Little Woodbury house. Each stands in a a stone wall which also small farmyard bounded by encloses various working places and a rectangular stone building. Both arc on moorland in contrast with most of the other known whcclhouses which arc on the machair near the sea and are mostly dug-out structures. At Allasdale the subsidiary rectangular steading was divided internally into two parts and was inferred to have been a combined barn and byre. The barn part had a fireplace which could have been used for drying hay to feed the cattle in the winter or for keeping the stored hay dry. The latter seems more probable since there was another 'kiln house' attached to the round farmhouse itself which could have been used for drying purposes. At Clettraval there was only the rectangular byre which has no trace of a dual function yet this farm had the clearest Wesscx-influenced pottery, in the style that takes its name from (73> S. Piggott in I. A. Richmond (cd.), Roman and Native in North Britain (195s), chap. 1. <74> A. Fenton in PSAS XCVI (1962-3), , especially 276. l > Alison Young in PSAS LXX (1952-3),

21 the site. Significant too is the fact that both sites produced neither nor rotary saddle quernstones for corn-grinding though a saddle quern was incorporated in the wall at Clettraval. The absence of corn-grinding equipment is a very strong argument against the inhabitants of these two farms having grown crops, querns are common on other particularly as whcclhouse sites. Both sites are convincingly interpreted as cattle and sheep farms and the subsidiary buildings as byres for the animals and barns for winter feed. It would be hard to find a better illustration of the improbability of southern cultures being reproduced in the north with all their original features. For Clcttraval and Allasdalc arc two farms which seem, from their layout, to be as near to carbon copies in stone of Little Woodbury as one could hope to find and one of them, moreover, has plenty of evidence of Wcssex B influence in its pottery. The wheelhouse structure too seems to have originated in the Jarlshof stone roundhouse which, with its internal wooden posts, was very similar to Little Woodbury <33>. Yet neither Hcbridean site has any clear evidence for crop-growing at all let alone on a sufficient scale to produce a large surplus and both are best interpreted as cattle farms. Their pottery too varies markedly. Allasdalc had much incised ware of a descended from local Neo- type probably lithic pottery whereas Clcttraval had the intrusive, Wessex-influcnced sherds. Another whcclhouse in South Uist, A' Chcardach Mhor, lacked any evidence of the Woodbury farmyard yet had many querns and much Clcttraval pottery. Nevertheless the layout of Allasdalc and Clcttraval is so similar to that of Little Woodbury and allied southern sites that the possibility of the idea of such farmsteads having been introduced from the south at the same time as the other items mentioned must be considered. (i) Discussion of the Exotic Material Culture The implications of the appearance of English artifacts in Scotland, and particularly in the Hebrides, arc important. Do they represent an actual migration of families to the far north or do they simply illustrate the gradual penetration of new ideas and techniques into Scotland and its highland zone, perhaps assisted by trading contacts? There has been a tendency in recent years to avoid the automatic assumption that the appearance of exotic pottery and artifacts in a territory means that colonists had arrived there and to emphasise the continuity of local cultures<7s>. In fact two quite separate phenomena can be suggested to account for the exotic Iron Age artifacts in Scotland. In the first place there can be little doubt that new equipment and ornaments gradually filtered into southern Scotland from northern and north-eastern England without any significant population rotary querns provide of how a movement. The beehive a classic demonstration simple device, whose value must have been immediately obvious to the peasant mind, could spread relatively quickly from the south of England through the midlands and the north as far as the Forth-Clydc valley without necessarily involving population movements. Also new fashions in bronze ornaments and equipment such as penannular fibulae and ring-headed pins could have drifted northwards in the same way from eastern Yorkshire where there was, from the fourth century b.c. onwards, an undoubtedly intrusive group of La Tene settlers from the continent*77*. Elizabeth Fowler showed that the early penannular brooches of Lowland Scotland (Type Aa) closely resembled the Yorkshire ones of the third century b.c.<31). Thomas' map showed how other probably contemporary artifacts such as three-link bridle bits, certain swords and scabbards and long-handled bone combs arc also distributed along a broad path from southern Scotland to north British Yorkshire*32'. The peculiarly ring-headed pin with a projecting head also fits well into this pattern. The fact that these native ring-headed pins and the long-handled combs arc found fairly on frequently broch and whcclhouse sites in Atlantic Scotland led Thomas to suggest that <76> Note 58: also J. G. D. Clark in Ant. XL (1966), 172 ft". <77> LA. Stead, The La Tine Cultures of Eastern Yorkshire (1965). 57

22 the characteristic features of the 'Atlantic Second B' cultures were primarily derived from the Yorkshire area and had fdtered across Lowland Scotland, perhaps brought by actual migrants. The presence of a little pottery on Tiree with resemblances to some Yorkshire Iron Age A wares supports the view that the Hebrides were occasionally not beyond the reach of influence from that area(78). Yet it is perhaps wiser to think in terms of simple culture contact to explain the gradual drift northwards of these Yorkshire Iron Age B that a few bronzes though it is not impossible of the older population moved away when the Continental La Tene settlers began to arrive. That the simple adoption of new ideas by an established population occurred is also suggested by increasing evidence that some of the structures in which the exotic artifacts arc found such as palisaded sites and vitrified forts were being built much earlier than the third century B.C. and that some go back to the Late Bronze Age in the sixth or even the seventh centuries*'-4*. Yet some of the exotic Iron Age material could have reached Scotland in another way and the English artifacts in the Atlantic province illustrate this. As discussed earlier the spiral finger rings which provide an unique and classic demonstration of contact between southern England and Scotland (Fig. 3) arc likely may have originated to be later than the other bronzes which in Yorkshire*52'. More than half the total from the mainland are now known from Atlantic Scotland and they occur with the other southern artifacts such as the bone dice, the long-handled bone combs, the glass beads and discoid rotary querns with vertical handles. The native, northern, projeeting-head bronze ring-pins once thought to have appeared in the broch cultures with these other innovations are now known to have been in the Hebrides much earlier. The impressions of two were found on two sherds in a sixth/fifth and a fourth century B.C. context respectively at Vaul <24-34). Much of this English equipment is not easily explained as having originated in Yorkshire but is more plausibly derived, as an assemblage, from southern, not south-western, England. The copy of the Wessex eyebrow-ornamented bowl on Tircc is surely strong evidence in support of this as is also the apparently simultaneous appearance in the Hebrides ofthe Wcsscx-influcnced Clettraval pottery style. The few known pre-broch sites in the Atlantic province have no trace of these English artifacts except the long-occupied vitrified fort of on Dunagoil Bute which yielded a spiral finger ring <56). They are evidence of the transfer to the far north of a substantial part of the material culture of the later Iron Age inhabitants of some region cast of the Somerset 'lake village' area. No trace of the characteris- is visible in tic Glastonbury decorated pottery the Hebridean styles. The absence of the early La Tene Ic fibulae from Atlantic Iron Age sites (again with the possible exception of Dunagoil) also shows the lateness of these southern contacts: all such fibulae arc on the mainland. There are no demonstrably Bclgic elements among the broch and whcclhousc artifacts. The relative lateness of the exotic Hebridean material within the south English pre- Roraan sequence is also effectively shown by the rotary querns which are all of the flat, disc-like type with vertical handle holes. None of the beehive series with lateral handle holes have been found in the islands though many are known from southern Scotland and from north English sites like Huckhoe (Fig. 5). This suggests that the beehive forms filtered gradually northwards from England whereas, for reasons given earlier, the disc querns were probably brought to Atlantic Scotland directly from the south where they were the immediately pre-roman and the Romano-British type. On the other hand the distribution of the La Tene III fibulae with an south English unperforated catch-plate, derived from the Continental Nauheitn form, provides negative evidence that the English artifacts in the broch province arrived before the first century a.d. These fibulae are usually dated to the first century a.d. and one was found in a grave at Loughey, near Donaghadcc in Northern (78> At Balevullin note

23 Ireland, with a spiral finger ring and other items; Jope thought they might have belonged to a south English migrant avoiding the Roman conquest of a.d. 43 and the years following (79>. The occurrence of only one Nauheim fibula in Scotland (in the south-cast) and their commonness in southern England from Somerset eastwards, in precisely the area from which the other exotic Atlantic Scottish material seems likely to have come, is a strong argument for putting the transfer of this other material to the north at an earlier period. influence have in fact been Traces of Belgic noted in south-cast Scotland'80*. From the limited evidence at present available it seems that the various elements of English Iron Age B material culture arrived in Atlantic Scotland approximately simultaneously at some time in the first century B.c. (Fig. 7). Certainly they form a distinctive stage in the development of the local culture <81). The Wesscx-influcnced pottery, the rotary querns and the long-handled combs (whose distribution is geographically distinct from that of the possibly Yorkshire-derived examples on the mainland) should mean that womenfolk had arrived. This in turn suggests that a fairly comprehensive folk movement from the south took place, though not necessarily involving a large number of people. It is difficult to assume that a sudden burst of regular trade started between two such separate areas, which an intransigent antimigrationist would have to do. (j) The Belgae and Flight from the South It is accepted now that the suggestion of immigrations to explain the appearance, for example, of new pottery styles must not be lightly made. If one could point to a good reason for such supposed migrants leaving their homeland in Wesscx and coming to the Hebrides this would certainly increase the plausibility of this theory. One such reason is available in the known settlement of Belgic tribes in south-cast England even though this settlement is not now thought to be the sudden event that it was once<82>. Such a wholesale colonisation by farming tribesmen and their families, even if it took place in several waves over a few decades, would provide an appropriate stimulus for some of the original inhabitants to look for new homes. The leading families might perhaps be those least likely to tolerate the sharp decline in status which would accompany the arrival of powerful and numerous new neighbours. Yet the start of the Belgic settlement, now plausibly put near the end of the second century b.c.<40), is not the date which can be taken to mark the appearance of Wcssex B refugees in the far north. The Belgae settled first in Kent and south-east England, areas in which the Iron Age B material culture, described earlier, is rare; evidence of is also the Woodbury farming economy lacking (Fig. 3). It would be when the colonists, having established their bridgeheads in Kent, were beginning to extend their influence and settlements westwards into Wesscx that would start to leave that area. For- refugees tunately Allen's study ofthcbritish prc-roman coinage, introduced by the Belgae, has provided clear evidence of the progress of this inland spread. This evidence is doublyvaluable because almost no Belgic sites have been found which can be dated before about 50 b.c.(83). Of the various waves of Gallo-Bclgic coinages entering south-east England the first to affect wide areas inland was Gallo-Bclgic C, a series which may have started as early as rise to a series 100 b.c.<84). This coinage gave of local coinages in inland areas which illustrate well the gradual penetration of the settlers deeper into the country and their progressive loss of contact with the continent. The British derivatives A and B arc found in the Wesscx area and even as far west as the river Severn. The start of British B is put by Allen at about b.c. with the help of the E. M. Jope in Ulster Journal of Archaeology, XX (1957), 73-94: distribution of Nauheim fibulae, 78, fig. 2. <8'» PSAS LXXXVII (1952-3), 18, fig. 3: ibid. XCI (1957-8), 73, 8. (si) Note 4, 114, 120-2, 118, fig. 6. <82> Anne Birchall in PPS XXXI (1965), 241 ff.: C. F. C. Hawkes in Ant. XLII (1968), 6 ff. (83> Birchall, supra, note 82. <8J> D. F. Allen in Frcre (ed., 1959), p. 104 ff. 59

24 Lc Catillon, Jersey, coin hoard so one may assume that the secondary Bclgic settlement of inland areas was beginning about then. Thus 75 b.c. might be a reasonable earliest date for the emigration of natives from Wessex to the Hebrides. Discussion In the previous sections two distinct archaeological phenomena have been clarified. From the first it now seems clear that the Scottish brochs were the latest product in a long history of development and experiment in drystonc hollow-wall building in western Atlantic Scotland and one, moreover, which much of the evidence suggests emerged Skye. on The study of the varied material cultures associated with the brochs suggests that they were introduced to a large number of pre-existing and culturally diverse communities throughout the whole of Atlantic Scotland, from Argyllshire to Shetland, and probably by professional fort buildcrs<85). Either there must have been some loose political unity over this province or, alternatively, no overall authority but rule by numerous independent petty chieftains. The existence of this evidently urgent need for an efficient refuge over such a wide area suggests that the latter conditions prevailed. The second archaeological phenomenon clarified is the appearance of a range of new ornaments and domestic equipment which suggests a migration of families from southern England to the Hebrides in the first half of the first century b.c. This is the most obvious intrusion of a new archaeological culture into Scotland since the Beaker invasions at the start of the second millennium b.c. and the likeliest explanation for it is still that a small number of the Iron Age B peoples of southern England were expelled, or preferred to move out, probably between 80 and 50 b.c. when the Bclgic tribes began to settle there. It is possible that others moved northwards from a.d. 43 onwards to avoid conquest by the Romans but the pre-belgic, native Iron Age B character of most of the new artifacts in Atlantic Scotland suggests that the latter, if any, did not come to the Western Isles. (a) in the Hebridean Culture Sequence artifacts have been found The Arrival of the English Artifacts The south English frequently in brochs and whcelhouscs together with undoubtedly aboriginal material so an obvious question poses itself. Did the arrival of English settlers actually stimulate the development of the broch from the pre-existing Hebridean scmibroch tradition > The fact that intrusive Wessex B influence is strongest in the Western Isles, and that independent evidence suggests that the brochs emerged there, provides good circumstantial evidence in favour of this view but there arc further reasons for supposing it to be correct. One is based on the small amount of reliable data available which links the sequence of artifacts and structures and the other is a fragment of evidence from the fort sequence itself which shows that new influences arrived at a definite point within it. These two separate clues complement one another very well. Most of the stratified Hebridean sites which have contributed to the cultural sequence are whcelhouscs, dwellings which belong without doubt to a peaceful post-broch era<38). Only the Vaul broch supplies information about the earlier phases, nearer the time of the invention of the towers, although there is some useful data from the Chckhmiin broch in Shetland. The latter broch yielded several of the exotic artifacts in its primary occupation strata, including a long-handled bone comb, a bone die and a bronze spiral finger ring(86). No rotary querns were found in these levels but one came from an early post-broch floor at Jarlshof<87>. The Clickhimin broch pottery, which included evertcd-rim jars, lacks any trace of the Hebridean curvilinear motifs. The ring-headed pin from the broch levels was one of the native Scottish forms with a projecting beads were found at Clickhimin in deposits associated with both head<88). Yellow glass ring 85> Note 4, <86> Hamilton (1968), p. 115 ff. <8;l Hamilton (1956), p. 49 (chamber i). <88) Hamilton (1956), p

25 the first and second pre-broch forts, the latter presumably dating to shortly before the broch was built'48'. This last piece of information suggests that the Clickhimin tower was introduced to an established community which had already acquired some of the new ornaments and tools, either by contact with the Western Isles or by the earlier appearance of a few southerners. This evidence shows clearly that some south English influences were present in northern Atlantic Scotland some time before the latest solid-based brochs, as described above, had been developed*26'. The Vaul broch on Tircc is more helpful. This is a typologically early ground-gallericd tower of typical Hcbridcan form and it yielded the Wcssex-influenccd Clettraval pottery in its construction as well as primary occupation levels, together with an almost flat rotary quern (without the handle hole) and a Scottish projecting-head ring-pin. However, the Clettraval sherds in the primary levels were the fully developed local style which must therefore have coalesced from its diverse components at some earlier time. The B bowl was imitation Wesscx Iron Age stratigraphically slightly later in Phase 3a, the primary use of the broch but it is quite likely to have been as an kept heirloom for some years. It shows well how the south English influences were strong on Tircc and, together with the geographical position of the island, suggests that the colonists (or traders) may have found one of their first landing places there. Clettraval ware could easily have been invented on Tircc, but the earlier phases before the standardised form was evolved arc hardly known. Wc have the two prototypes in the Wesscx bowl (for the 'eyebrows') and the much earlier neck-band sherd (for the evcrtcd-rim, cordoned jar) and one sherd from the outer court which looks like a transitional form (Fig. 2, 2). However there is a little evidence that the site was inhabited shortly before the broch with some of the was built and by people south English equipment. The ground had been extensively churned up when the site was cleared for the building of the tower so that most of the traces of this settlement were destroyed. The upper stone of a flat disclike rotary quern with an upright handle hole was found at a very low level and probably belonged to this as shadowy phase (2a) did several severely burnt and blistered sherds. No other pottery anywhere on the site was found which had been heated in this way: the only comparable material the writer has seen is the pottery from the destruction levels of Dun Ardtrcck on Skyc some of which was partially vitrified by the intense heat'16'. The presence of these few incinerated sherds among hundreds of undamaged pieces suggests that a wooden structure had existed on the site before which had been burnt and the shortly debris almost completely removed by the subsequent broch-building operations. The few sherds were a cruder form of Clettraval ware different from that of the broch and levels and look like a suitable post-broch early form before the standardised arrangement of the motifs was evolved. This rather tenuous direct evidence for the presence of some English immigrants at Vaul supported by before the broch was built is strong circumstantial evidence, based first on the fact that sherds of fully developed Clettraval ware were in the primary broch levels indicating that the exotic elements must have come earlier and secondly by the plausible general hypothesis that brochs were built for already existing communities and do not at present appear to represent the arrival of broch-building immigrants. This view holds good for the brochs and their associated artifacts in all parts of Atlantic Scotland and not just in the Western Isles and most of the Vaul evidence is best explained in this way, as will be made clear in the final report'34'. Until more Hcbridcan brochs are stratigraphically excavated the Vaul evidence stands alone but arc that its implications arc important. They the arrival of south English immigrants antedated the construction of one very early western broch. Any such immigrants who landed on Skyc could therefore have stimulated the conversion there of the D-shapcd semibroch into the completely enclosed tower. 1 6i

26 (b) Southern Innovations in the Structural Sequence There is direct evidence that they did do this in the design of the brochs themselves. Large numbers of them possess round, corbelled guard chambers opening off their entrance passages whereas none of the immediately preceding semibrochs does. It is true that the Dun Ardtreck semibroch on Skye has a crude rectangular guard chamber but this is nothing more than the end of the mural gallery where it meets the entrance passage, barred off from the remainder by a thin primary wall. Rhiroy, on Loch Broom, has no guard cells*89* and several other unexcavated semibrochs can be seen to be the same(90). chamber as a The concept of the guard carefully designed, separate feature is only likely to have come from southern England or Wales (Fig. 6) where a number of Iron Age hillforts have such features'91'. Even if there was no other evidence of the arrival of the B culture in the south English Iron Age Hebrides, the addition of separate, round guard cells to the fortlets using the high broch wall just at the point that the circular tower form emerged would suggest that the influence fort-builders reached the north at of English that time. The majority of the English and Welsh forts have double guard chambers, on opposite sides of the entrance, and it might be supposed that this was the primary form in Atlantic Scotland although the majority there (74%) have single chambers cither on the right or left sides of the passage. An analysis of the occurrence of the various arrangements of guard cells in brochs shows that the double cell arrangement is far more frequent among the relatively small number of groundgalleried brochs the primary form according to the theory of their typological development outlined earlier. The number of double cells among brochs of unknown structural type (18%) suggests that this is the low proportion to be expected from a random sample of the broch 'population'; the sudden leap to 50% among the ground-gallcricd towers is the more striking by contrast. The double guard cell could thus have been the arrangement which was first introduced to Atlantic Scotland and which was then rapidly modified in favour of single right or left hand chambers. Savory discussed the occurrence of guard chambers in the Iron Age hillforts of north Wales and Shropshire and concluded that they were most likely to have been a local development, perhaps associated with the tribes of the Cornovii who were in north Wales when the Romans came'911. Most of these 'Cornovian' guard chambers were sub-rectangular and some were wooden constructions within the rampart. However the recent discovery of a pair of circular, stone-lined guard belonging chambers to an early stage in the history of the Rainsborough hillfort, Northants., may require this interpretation to be modified. Avery dates the building of the Rainsborough <* ) Current Archaeology II, No. i (Jan. 1969), 12 ft". (90) For example Dun Grugaig and Rudh an Dunain, Skye: note 4, 139- (si) w. Gardner and H. N. Savory, Dinorben (1964), pp and fig. 15; Avery in PPS XXXIII (1967), , 253-4; S. Piggott, Ancient Europe (1965), pp and refs. FORM OF CELLS BROCH STRUCTURAL TYPE Ground-galleried I Transitional I Solid-based I Unknown type None Right Left Both 3 (30%) 1 (10%) 1 (10%) 5 (50%) 1 (14%) 3 (44%) 2 (29%) I (14%) 14 (38%) 13 (35%) 5 (I3i%) 5 (I3i%) 3 (9%) 19 (59%) 4 (13%) 6 (18%) TOTALS (100%) Table: the distribution ofvarious arrangements of guard chambers among the different structural types of brochs. 62

27 _ /RON AGE FORTS WITH DOUBLE GUARD CELLS * 0 drystone corbelled W cefls with doors \3?TP wood & stone lined /v 4 [/^~^ 00 recesses in sides of j ^/ entrance passage tf"^ 4 -<fy J Fig. 6. The distribution in the British Isles of Iron Age forts with double guard cells opening offthe entrance passage.

28 chambers approximately to the fifth century b.c. and the destruction which put an end to their use to the early fourth century: in any case they belong to a clear Iron Age A context <91>. The effect of this discovery in the northern part of the Wessex area may be to push the date of the north Welsh guard chambers backwards by a few centuries so it is at present not entirely clear which area subscribed the cells to Atlantic Scotland in the idea of guard first century b.c., when the brochs appeared. The derivation of all the exotic traits there from the Wessex B cultures is made easier by the Rainsborough discovery even though, as noted, the guard Age cells there are from an Iron A context. So far we do not know whether they were in use in Iron Age B times except perhaps at near Leckhampton, Gloucester*92). It may be of course that the idea of guard cells was taken to north Wales as part of the same refugee movement which brought them to Scotland. The discovery of spiral bronze finger rings at Dinorben, Denbighshire, and Dyserth, Flintshire, might support this idea although some of these rings could have moved northwards in the mid-first century a.d. as well. Not enough information is yet available about the origin of the English guard cells for any firm conclusions about this to be drawn. The transfer of the idea to Scotland ought to have occurred in the mid-first century b.c. if the rest of the evidence for southern contacts at that time is a reliable guide. The design of the Scottish guard chambers is different too. In the English and Welsh forts the chambers are in effect recesses in the sides of the entrance passage without distinct doorways but in the Scottish duns and brochs they occur as complete, corbelled cells with a narrow, low doorway to the passage. The difference in design is no doubt explicable by the different building material and by a preexisting tradition of corbelled drystone cells in the north which was adapted to the new function. (c) Summary and Conclusions (Figs. 7, 8) The general situation at the onset of the era broch-building thus seems to be as follows. We have reasonably good evidence that a number of families, rather than traders, bearing the late Iron Age B (orlatewoodbury) culture of the Wessex area settled in the Hebrides just before the earliest ground-gallcricd brochs appeared, probably on Skye. The earliest of the successive waves of Bclgic colonists arriving in Kent and south-east England probably started shortly before 100 b.c. according to the coin evidence. The same evidence suggests that they were penetrating the Wessex area from about 80 b.c. reason for the onwards and this provides a good retreat of some of the presumably speaking P-Ccltic native inhabitants of that area to north Wales and Scotland. A date of between 80 and 70 b.c. thus seems a suitable one at present for the first appearance of Wessex refugees in southern Scotland and the Hebrides. evidence for the evolution The independent of the Atlantic Scottish hollow-walled forts indicates that the Hebridean semibrochs were the immediately pre-broch form. The development from them of the totally enclosed, freestanding round tower the broch coincides with the appearance of south English or Welsh influences in the form of round guard to the numerous chambers, a structural parallel exotic southern artifacts. The conclusion seems inevitable that the broch towers developed from the local semibrochs at about the same time that the Wessex B settlers arrived. This is the most that can be strictly inferred to follow from the evidence but it is tempting the chain of deductions a little further and to conclude that it was the immigrant southerners who saw the possibilities of further development in the indigenous semibrochs already in use on Skye. Remembering perhaps their abandoned round wooden farmhouses with internal raised lofts, they could have realised that if the tall hollow rampart was built as a compact circle an immensely strong, practically impregnable tower would result which could be built almost anywhere, away from (92> Transactions of the Bristol and Gloucestershire Archaeological Society, xlvii (1925), 98 ff. 64

29 ATLANTIC IRON AGE west I north STAGES OF MATERIAL CULTURE IV II A.D B.C. 70 Fig. 7. A suggested scheme for the stages in the evolution of the Iron Age material culture in Atlantic Scotland, a revised version of that in PPS XXXI (1965), p. 118, fig. 6. The black arrows represent the arrival traits and the white arrows the continuation of existing ones. The numbered stages correlate with of new those on fig

30 cliffs and precipices and close to the fields and settlements. The emergence of a class of professional fort builders and the rapid spread of the new design among the Hebridean settlements seem a natural development in such circumstances. Thus probably was the broch born in the west, as an accidental and immi- combination of local ingenuity grant inspiration, into an unsettled social environment whose needs it exactly fitted. of the brochs has the This view of the origin merit of fitting the latest archaeological evidence closely and also of adopting aspects of most of the traditional theories and to some them. Those who would extent reconciling have had the towers created in response to special conditions, or by immigrants with special knowledge and skills, can look to the English settlers whose P-Celtic flair and initiative wrought wonders with local building traditions. Those who prefer to see the brochs as the final result of an ancient Scottish fort-building tradition will have no quarrel with the concept that they emerged from the locally developed semibrochs, themselves presumably the end products of several centuries of Hebridean building skill. As for the social context in which the towers were invented and used, the hypothesis offered here which views their origin and spread primarily as the responsibility of a specialist class of fort designers and builders con- their architectural prob- veniently separates lems from their social background. The old argument about whether the brochs were the castles of an intrusive and dominant minority group or the refuges of the aboriginal population keeps emerging in different forms and has never been satisfactorily settled. It is clear now that it will only be settled by careful excavation of the deposits inside and around individual towers and not bycontemplation, however prolonged, of the great drystone keeps themselves and their marvellous design. Only if we can reconstruct how and by whom the brochs were used from the evidence contained within them are we likely to be able to understand for what precise purposes any particular one was built. Whatever the conditions were in which the were in existence before brochs flourished, they the towers appeared: the earlier stone forts and semibrochs are clear evidence of this. Remarkable and are unique though they the brochs are no more than specialised versions of Iron Age communal refuges*93' which emerged, almost by a fluke, to satisfy an existing demand. Such useful constructions may well have been used differently in different regions, as communal refuges in the Hebrides, as perhaps chieftains' castles in some parts of the north or even occasionally for reasons of personal prestige, as perhaps at Mousa. All that this paper has tried to show is that the stages through which the design of the hollow-walled fortlets developed to reach the perfection of the broch of Mousa can be inferred quite easily and that the crucial step by which the older, cruder semibrochs were converted to formidable, freestanding refuges tower- was probably the responsibility of immigrants from southern England in the first century b.c. How the new invention was then used throughout Atlantic Scotland is a different problem. Postscript After this paper had been completed, and when it was about to be sent to press, there appeared an article on bone dice by D. V. Clarke which contained an apparently detailed attack on the whole concept that there was a period of strong cultural links between Scotland and southern England in the late is to be Iron Age(94). If a complex hypothesis demolished at least one of two things needs to be done. Firstly if errors or omissions of fact can be pointed out in sufficient number then the reliability of the hypothesis is seen to be in doubt. Secondly if no substantial errors are better explanation of the relevant found, a data may be proposed. I suggest that Clarke's criticism fails on both counts (he does not even <93) The exact primary function of an excavated broch is difficult to infer unless the strata are particularly informative, as they were at Vaul. This broch was almost certainly first used as a sporadically inhabited, communal refuge (note 34). <94> D. V. Clarke, 'Bone dice and the Scottish iron Age', PPS XXXVI (1970),

31 PERIOD HEBRIDES I material culture 1 structures N. ISLES PHASE iron 300- age iii ar -200' f i.brochs cease to be wheelhouses leeiikiubt t used as forts post- broch iron age ii l0of ico I -^Roman fragments i Clettraval and Vaul ware t t Wresex Iron Aiie B material _ ground-galleried guard cells -^ brochs solid-based brochs broch building ag e i b.c. semibrochs Clickhimin II I gatehouse forts fort building late some English Iron *A'jre A influence i t aboriginal Vaul and Balevullin ware early forts (none excavated) -Clickhimin I Iron Age A influence courtyard houses bronze age sooh> some Urnfield everted rim ware T prefort Fig. 8. Chronological chart summarising the suggested cultural development in Atlantic Scotland from 6oo b.c. to a.d The dating is based on C-14 and, from the latter part of Iron Age II onwards, on Roman fragments. attempt an alternative explanation) and is moreover rendered null and void by itself containing several important factual errors. This being so I suggest the 'English migrants' hypothesis should still stand as the likeliest explanation of the known data that we have at present. There is space here to present only an outline of the reasons for this conclusion. The first and major fault in Clarke's argument is that he has knowingly used a published summary of the views set out in this paper which was clearly labelled as such <95> and which contained only a small proportion of the facts used here and without references. Moreover it was plainly stated in that summary that the full version was to appear in this journal. Yet nowhere was this mentioned and the clear impression given that the was items of evidence questioned the bone dice, spiral finger rings, rotary querns, guard cells and the 'Wessex bowl' from Tiree were the only ones in my scheme, Clarke thus ignored the whole crucial problem <»5) -y/ MacKie, 'The Historical Context of the Origin of the Brochs,' in Scottish Archaeological Forum, I (1969), pp. 53-9, especially note I. (>1

32 of the structural development of the brochs and their prototypes, the evolution of the Hebridean Iron Age material culture as a whole, the development of the Hebridean pottery styles now well documented and in which the impact of southern influences is plain the glass beads, triangular crucibles, long-handled bone combs and farmsteads like Little Woodbury together with the negative but important evidence provided by the English ring-headed pins and the La Tene III Nauheim fibulae. Yet one must still admit that if the items that were mentioned could be shown to be against the 'English migrants' theory then some doubt could be cast on it. Unfortunately there are several important errors of fact in Clarke's argument which suggest that more is to be care is needed if this hypothesis effectively challenged. For example Clarke claims*96' that no dice in Scotland can be dated to earlier than the second century a.d. and that therefore they do not overlap at all in time with, and hence cannot be connected with, the dice in use in southern England (which, he suggests, probably went out of fashion with the Roman conquest from a.d. 43 onwards). The die concerned is that from Clickhimin in Shetland which was recovered from an occupation layer outside the broch*97'; the excavator clearly states his view that this on deposit the floor of 'temporary huts' accumulated during the construction of the broch<98'. The date of c. a.d quoted by Clarke(99> does not refer to this horizon or to the die but to a later layer of rubble associated with the abandonment of the broch and the building of a wheelhouse*100' ; the dating evidence was a piece of Roman glass of late first or early second century rubble layer a.d. date from a deposit above the mentioned'101'. The die is thus earlier, probably substantially earlier, and its date depends on that estimated for the construction of the broch. Such a solid-based tower as that at Clickhimin is likely, in my view, to have been built early in the first century a.d., or perhaps towards the end of the first century b.c., an estimate shared by Hamilton*102'. Therefore the die could easily have been deposited before a.d. i and at a time when dice were in use in southern England. This mistake over a crucial piece of evidence causes a large part of the counterargument to collapse; it is the more surprising because the main purpose of the article was to of bone discuss the precise context and dating dice, and it included a comprehensive list of the Scottish examples. The same reasoning applies to the bronze spiral finger ring of simple form which came from the same 'temporary hut' floor as the die*103' and is of similar age. Contrary to what Clarke states <104> these rings are not rare A has a list of in Atlantic Scotland; Appendix 12 (excluding silver spiral rings) from the province compared with 22, or one or two more, from the rest of the mainland. The statement that 'their distribution (in Scotland) is hardly what would be expected if these objects were introduced as part of Wessex B equipment to Atlantic Scotland'*104' is startling. A glance at Fig. 3 here will show how the reverse is true; it is exactly what one would expect if people from the south came up the west coast route and penetrated Lowland and Atlantic Scotland. Incidentally the spiral finger ring and the small yellow glass bead from the pre-broch fort levels at Clickhimin <io5) snow clearly that some artifacts of south English type arrived in Shetland even earlier, before the Clickhimin broch was built. This is what one would expect if the cultural contact with the south occurred slightly before the earliest brochs were developed in the Hebrides, as argued above (p. 60 ff). Another error occurs in the discussion of rotary querns. Clarke states that at Jarlshof 9- <96> Note 94, 218. <9?> Hamilton (1968), p. 113 and layer 3 on section G'G, fig. <98> Ibid., p. 97. (»») Note 94, 216. (too) Hamilton (1968), p. 107 and layer 6 on section J'J, fig. 10. (i i) Ibid., pp. 133 and 137 and layer 10 on section H'H, fig. 10. <102> Ibid., table on fig. 3. <">3> Ibid., p (ioi) Note 94, 219. <105> Hamilton (1968), pp. 80 and 90 and pi. XXXIa.i for the ring: the same, with fig. 41.1, for the bead. 68

33 rotary querns are unknown before the wheelhouse (later post-broch) phase*106', the implication being that the arrival of these querns in Shetland was much later than that of the brochs. But in fact a fragment of a rotary quern was found in a much earlier post-broch level, on the floor of the aisled roundhouse*107'. Moreover the evidence from Jarlshof cannot show that querns did not arrive earlier still, with the broch, because hardly any artifacts were found in the primary broch levels*108'. The evidence from Vaul of course shows that the flat disc-querns arrived on Tiree slightly before the broch there (p. 54 as above) Clarke had to admit. Space allows one more only comment on Clarke's criticism. It must be suggested that the present boundary of Northamptonshire is not relevant to the question of whether the Rainsborough hillfort, with its double guard cells, was in the Wessex B culture province in the Iron Age, which Clarke denies*109'. In fact the distribution of the spiral finger rings (Fig. 3) shows clearly that it was. Differences of opinion such as this need to be justified, simply not set out as briefly established fact and used as such. (loo) Notc Q4i 219. <10,> Hamilton (1956), p. 49, (in chamber i). <I08> Ibid., pp <ioo) Note 94j 220 Bibliography V. G. Childe, The Prehistory of Scotland (1935); S. Frere (ed.), Problems of the Iron Age in Southern Biitain (1959);. R. C. Hamilton, Excavations at Jarlshof, Shetland (1956);. R. C. Hamilton, Excavations at Clickhimiu, Shetland (1968); A. L. F. Rivet (ed.), The Iron Age in North Britain (1966); F. T. Wainwright (ed.), The Northern Isles (1962); R. E. M. Wheeler, Maiden Castle, Dorset (1943). APPENDIX A: BRONZE SPIRAL FINGER RINGS IN SCOTLAND (Fig. 3) Unless otherwise stated each site has produced ring, made either one of round bronze wire or, rarely, of flat ribbon; all are undecorated. Clickhimiu broch, Shetland (HU ): two rings, one from the pre-broch fort levels and one from a primary broch deposit; Hamilton (1968), pp. 80, 90 and pi. xxxia.i, and pp. 114, 116 and fig. 50.3, respectively. Gurness broch, Orkney (HY ): at least three, allegedly found on the fingers of a severed hand in a midden. Unpublished and with the Department of the Environment in Edinburgh. Crosskirk broch, Caithness (ND ): from the interior floor level: DES (1966), Watten, Caithness (c. ND 2354): one found 'in a mound at Watten': PSAS XXVIII (1893-4) Eriboll, Sutherland (c. NC ;): two found in a broch Archaeologia Scotica V (1890), 109 and 106, and pi. xvi, u and v. This may be the broch at Camas an in an earth- Duin. The later reference to their discovery house at Eriboll is presumably wrong RCAM (Scotland), Sutherland, no. 159, p. 55. Vaul broch, Tiree (NM ): from the Phase 4b midden in the mural gallery: report to be monograph published by a the University of Glasgow. Glenramskill dun, Kintyre (NR ): DES (1964), 18-19: the ring was identified as such in the Glasgow Art Gallery and Museum. Dunagoil vitrified fort, Bute (NS ): one ring reconstructed from fragments: Margaret Lecchman, The Vitrified Fort of Dunagoil (Edinburgh M.A. thesis, 1962), p. 45 and pi. v, no. 18. Cleaves Cove cave, Ayrshire (NS ): two rings, unstratified: J. Smith, Prehistoric Man in Ayrshire (1895), p. 62. Gourock Bum, lower fort, Ayrshire (NS ): fragments of one ring, DES (1968), 13 ; I have to thank Mr Alastair Hendry for letting me see it. Hyndford crannog, Lanarkshire (NS ): three rings, PSAS XXXIII (1898-9), 383, fig. Lochlee 10. crannog, Ayrshire (c. NS 4327): of thin ribbon, PSAS XII (1878-9), 234, fig. 104a. Castle Haven galleried dun, Kirkcudbright (NX ): two rings; PSAS XLI (1906-7), 79, fig. 9. ha-.nberton Moor, Berwickshire (c. NT 9559): two rings, one with a serpent head terminal, found with a hoard of Roman mctalwork and a few native objects, PSAS XXXIX (1904-5), 374, fig. 4. Granton pier cist grave, E. Lothian (NT ): in a grave with a penannular fibula and two skeletons, D. Wilson, Prehistoric Annals of Scotland (1863), Vol. 2, P Black Rock Cairn graves, E. Lothian (c. NT 4683): of one flat bronze ribbon, PSAS XLII (1907-8), 334, fig- 2. Traprain haw hillfort, E. Lothian (NT 5875): two rings, PSAS LIV ( ), 67, fig. 4, no. 14, and ibid., XLIX ( ), 175, fig. 26, no

34 Newstead Roman Fort, Roxburghshire (NT 5734): one of flat bronze ribbon, J. Curie, A Roman Frontier Post and its People (1911), pi. 84, fig. 5. The Camps, fortified site, Roxburghshire (NT ): several rings from an unpublished excavation, RCAM (Scotland), Roxburghshire, no. 457, pp Moniefieth hill fort, Angus (NO ): PSAS XVII (1882-3), 302, fig. 4. Hurley Hawkin, promontory fort and broch, Angus (NO ): I am grateful to Mr David Taylor for information about the ring. Abernethy timber-laced hillfort, Perthshire (NO ): PSAS XXXIII (1898-9), 31, fig. 16. Culhin sands, Morayshire (c. NJ 0263): at least one, PSAS XXV (1890-1), 507, fig. 32. APPENDIX B: PARALLELOPIPED BONE DICE IN SCOTLAND (Fig. 4) A more detailed description of each die, with drawings, will be found in the paper by D. V. Clarke in PPS XXXVI (1970), and The numbers in Clarke's appendix are given here. NMAS stands for National Museum of Antiquities of Scotland, Edinburgh. Unknown site in Scotland: Hunterian Museum, University of Glasgow, no. B (Clarke no. 21, fig- 3)- Bute, from a cave: in Glasgow Art Gallery and Museum, reg. no. 'i8-24g. See Childe (1935), p. 240, footnote 4 (Clarke no. 1, fig. 3). Clickhimin broch, Shetland (HU ); one die from the primary broch deposits: Hamilton (1968), pp. 116, 118 and fig. 49, 5 (Clarke no. 18, fig. 5). Sanday I., Orkney (Hermicsgarth), c. HY : one die in the Hunterian Museum, no. B (Clarke no. 16, fig. 3). Sanday I., Orkney (Stackwick Stywick or Bay, c. HY 6839): one die in NMAS, no. GA 694 (Clarke no. 17, fig. 2). Sanday L, Orkney (Peter Kirk, Newark), HY : an abnormally short, almost cubical copy of a bone die with inconspicuous single dots; Hunterian Museum, no. B (Clarke no. 14, fig. 5). Papa Westray L, Orkney (St. Tredwell's chapel), HY : one die apparently from the broch mound next to the chapel; in NMAS, no. HR 194 (Clarke no. 15, fig. 4). Burrian broch, Orkney (N. Ronaldsay I.), HY : three dice from the secondary occupation levels, in NMAS, nos. GB 229, GB 230 and GB 231 (Clarke nos. 8-10, figs. 4 and 5): PSAS X (1872-4), 11 and fig. Ayrc broch, Orkney (HY ): two dice, one from the primary floor level in presumably secondary outbuildings; PSAS XLVIII ( ), 43, fig. 10, and 50 ff.: NMAS nos. L and L (Clarke nos. 6 and 7, figs. 2 and 4). Lingrow broch, Orkney (HY ): one die found with two Roman coins of Crispina (A.D ); J. Anderson, Scotland in Pagan Times: the Iron Age (1883), p. 244 and footnote (Clarke no. 13, lost). Gurncss broch, Orkney (HY ): two dice in NMAS, unregistered (Clarke nos. 11 and 12, fig. 2). Hillhcad broch, Caithness (ND ): two dice, PSAS XLII (1908-9), 17-18; NMAS, nos. GA 846 and GA 847 (Clarke nos. 1 and 2, fig. 4). Dun Mor Vaul broch, Tiree, Argyllshire (NM ): two broken dice, unstratified, DES (1966), 10; Hunterian Museum, nos. A and A (Clarke nos. 19 and 20. fig. 5). Foshigarry wheelhouses, N. Uist, Inverness-shire (c. NF ): one die, PSAS LXV (1930-1), 313, fig. 6; NMAS, no. GA 149 (Clarke no. 5, fig. 3). Bac Mhic Connain wheelhouse, N. Uist, Invernessshire (c. NF ): one die, PSAS LXVI (1931-2), 59, fig, 15, no. 9; NMAS, no. GNB 65 (Clarke no. 4, fig 4). Dun Cuier galleried dun, Barra, Inverness-shire (NF ): six dice, three of which are markedly cruder in design than the rest; all from the dun floor level, PSAS LXXXIX (1955-6), 317, fig. 13; NMAS, nos. GU (Clarke nos , fig- 6). Clarke calls all these objects 'gaming pieces' yet three are no different from the other dice while the three cruder ones are easily explained as unskilled copies. A skilled dice maker must have some specialised metal tool for making the neat circles round the dots and, if such was no longer available, the roughly incised squares and rectangles round the dots on three of the Dun Cuier dice are probably the best anyone could do. These six dice are far more similar to the rest than the abnormally short one from Peter Kirk, Sanday, Orkney, which Clarke includes among the dice without comment. APPENDIX C: BEEHIVE AND BUN-SHAPED QUERNS IN SCOTLAND (Fig. 5) This is a preliminary list of sites which have yielded rotary querns with lateral handle holes. It is based on search of the literature, an examination of the collection in the National Museum of Antiquities, Edinburgh, the Glasgow Art Gallery and Museum and the Hunterian Museum and also on some personal information from the staff of several north British museums. It is not likely to be complete as the compilation of such a list is hampered by the fact that descriptions of rotary querns in early excavation reports arc almost invariably too loose to permit precise identification and they were rarely, if ever, illustrated. Only comprehensive museum visits will yield the required information. NMAS stands for National Museum of Antiquities of Scotland, Edinburgh. Cardean Roman fort, Angus (NO ): one bunshaped quern in NMAS, no. BB 124. Creich, Fife (c. NO 3221): one bun-shaped quern, NMAS, no. BB