^{page 46 note 1} P.P.S., 1947, 1 ff.

^{page 47 note 1} For the earlier excavations see P.S.A.S., LXIX, 480 ff. For facilities for the excavations here described I am indebted to H.G. the Duke of Hamilton and to his factor, Mr D. Macmillan, B.L., of Lochmaddy. The principal share in the work has been taken by my son, Mr N. L. Scott, and to him the results obtained are in large part due. During part of the second season I had the great advantage of the co-operation of Prof, and Mrs Stuart Piggott. Mrs Piggott has also most generously lent me drawings made in the National Museum at Edinburgh of pottery from other sites of the culture, and has thus most valuably supplemented my own visits to the museum and greatly facilitated the writing of this paper. It is to be hoped that her drawings may be published as a corpus of the pottery of the culture.

^{page 48 note 1} It may be useful to record that road-menders in the Hebrides find it worth while to make quarries at intervals of something like a quarter of a mile along the road, and thus to cart stones for distances only up to about a furlong. Where no road exists and a cart is not used the distance of convenient haulage of stone may be expected to be less. This provides some criterion for judging whether a building in the vicinity of a prehistoric site is near enough to have been built of stones robbed from it. Allowance must be made for special factors, such as the availability of water transport by raft.

^{page 49 note 1} I am indebted to Sir Alfred Clapham for this information.

^{page 51 note 1} The natural drainage of the hillside has been worsened by the subsequent growth of peat and, in recent times by deep peat cuttings immediately north of the house and the tomb. These are perennially rilled with water, which tended to flood along the rock surface into the lowest level of the excavations.

^{page 51 note 2} This is known as a “pot oven” and is illustrated in Estyn Evans' Irish Heritage, fig. 34.

^{page 51 note 3} I am very much indebted to Dr K. C. Dunham of the Petrographical Department of the Geological Survey for examining and advising on mineral specimens. I quote passim from his report.

^{page 52 note 1} I quote from the report of Mr M. Y. Orr of the Royal Botanic Garden, Edinburgh, to whom I am greatly indebted for identifying plant remains.

^{page 52 note 2} Mr Orr reports: “Small pieces of a fern stem or rachis, recognisable only by presence of scalariform tracheides.”

^{page 52 note 3} It is a difficulty in working in structures built of gneiss that elements in the rock are liable to disintegration into coarse grit. A large orthostat in the facade of the chamber tomb was undisturbed but totally reduced to a mass of grit.

^{page 53 note 1} It is convenient not to restrict the term to its modern usage of a cattle byre, but to use it to include a shed for any farm animal.

^{paeg 57 note 1} P.S.A.S., LXIX, 480 ff., figs. 32 and 40 (top row); LXXXII, forthcoming.

^{page 57 note 2} P.S.A.S., LXVIII, 283 ff.; LXIX, 206; LXX, 249 ff. I have given reasons for the omission of Wiltrow in P.P.S., 1947, 19 n.1.

^{page 58 note 1} Wheeler, R. E. M., Maiden Castle, 186 Google Scholar; Hawkes, C. F. C., S.A.C., LXXX, 258, 281 and 288 Google Scholar.

^{page 58 note 2} Dr Raftery's views as summarised by Prof.O'Riordain, , P.P.S.. 1946, 163 Google Scholar.

^{page 58 note 3} For the Little Asta cist see P.S.A.S., LXVI, 72 ff., and Childe, V. G., Scotland before the Scots, 16, 17 Google Scholar. It is worth noting that with one exception all the pottery urns from burials in Shetland are vessels of apparent Iron Age type—see P.S.A.S., LXVII, 345 ff. and figs. 1–9.

^{page 58 note 4} Arch. LXXVII, 195, pl. xxi, 6 and fig. 55Google Scholar.

^{page 58 note 5} Wheeler, R. E. M., Maiden Castle, 187 Google Scholar.

^{page 58 note 6} Something like it appears, though abnormally, in a Class II broch in the Northern Orkneys at a date not likely to be earlier than A.D. 100 ( Midhowe, , P.S.A.S., LXVIII, 505, fig. 50, 10Google Scholar).

^{page 59 note 1} P.S.A.S., LXVIII, 288, 290, figs. 54, 7 and 56, 3.

^{page 60 note 1} I am greatly indebted for facilities for the study of this pottery to the Institute of Archaeology (Maiden Castle and Breton sites); the British Museum (Woodbury); Lewes Museum (Sussex sites); Glastonbury and Taunton Museums (Glastonbury, Meare and Ham Hill); Wells Museum (Wookey Hole).

^{page 60 note 2} Excluding here Brittany, where the wheel was in partial use.

^{page 61 note 1} I am much indebted for the opportunity to search the store house at Glastonbury Museum, where the immense quantity of unpublished sherds is kept. A physical examination of the Meare sherds was not possible and the test was made on drawings, for access to which in advance of publication I am indebted to the generosity of MrGray, St. George. (Meare Lake Village, vol. I, has now appearedGoogle Scholar).

^{page 61 note 2} The pottery is in the Institute of Archaeology. The straight-sided pots at the Lake Villages include the Marnian ‘saucepan,’ which Prof. Hawkes regards as part of a native AB tradition in contrast to the Breton tradition in the village wares (S.A.C., LXXX, 256). At Maiden Castle the ‘saucepan’ is rare (Maiden Castle, 228 and fig. 70, no. 156).

^{page 61 note 3} Déchelette, , Manuel, II, iii, fig. 649Google Scholar. These pots were carried into Britain in the succeeding century and are well seen at Newstead (J. Curle, A Roman Frontier Post, pl. LIII).

^{page 62 note 1} Maiden Castle, 216, and distribution map at pl. XXVI.

^{page 62 note 2} Cf. Glastonbury, 11, pls. LXX, iv, and LXXVIII, P. 186 (the former from Ham Hill); Maiden Castle, figs. 65 and 68; S.A.C., LXXX, 284 fig. 4a (from Newhaven).

^{page 62 note 3} Also (more distinctively) P.S.A.S., LXIX, 517, fig. 33.

^{page 62 note 4} H. E. Balch, Wookey Hole, pl. XIV, 6 and unpublished examples in Wells Museum. For Mont Beuvray see Déchelette, , Manuel, II, iii, fig. 679Google Scholar, and Bulliot, Fouilles de Mont Beuvray, Atlas, passim. It occurs also, but exceptionally, at All Cannings Cross.

^{page 62 note 5} Glastonbury, II, pl. LXXXII, P. 229. For the metal bowls, ibid. I, frontispiece and the examples quoted on pp. 181–2. Twenty-two huts at Glastonbury produced rivet heads indicating the former presence of metal bowls (Meare, 1, 21).

^{page 63 note 1} For a discussion of the stamped rosette (or ‘circlet’) and its Breton origin see Hawkes, C. F. C., S.A.C., LXXX, 255 ff.Google Scholar

^{page 63 note 2} Déchelette, , Manuel, II, iii, figs. 489, 2Google Scholar; 644–5; 663–6. The Breton pottery is best seen in du Chatellier, La Poterie aux Epoques Gauloise en Armorique, plates, passim.

^{page 63 note 3} See Glastonbury, II, fig. 164 for the development of the designs at that site.

^{page 63 note 4} Wookey Hole, unpublished pot in Wells Museum; Porthmeor, Radford, C. A. R., Journ. Roy. Inst. Cornwall, XXIV, App. II, fig. 6Google Scholar.

^{page 63 note 5} Meare, P. 50, pl. XVIII; cf. also P. 225, pl. IV, where an applied cordon follows a more angular undulating line.

^{page 63 note 6} Hawkes, C. F. C., S.A.C., LXXX, 258 and 288 Google Scholar; Perkins, J. B. Ward, ‘The Pottery of Gergovia,’ Arch. Journ., 1940, fig. 8, 2Google Scholar.

^{page 63 note 7} J. R. Ward Perkins, op. cit., fig. 10, nos. 4–6. The device does not seem to occur in England till the 2nd century A.D. (cf. Verulamium Report, fig. 32, no. 46—a tazza from the ‘triangular’ temple).

^{page 64 note 1} It is not suggested that the distinction between pottery classed as B and that classed as A is absolute, or that all elements in the Clettraval pottery are incapable of interpretation in terms of an A derivation; but the elements of the ware are typical of the B wares and not typical of the A wares.

^{page 64 note 2} Meare, 16; and Maiden Castle, 216–8. It may be appended to Dr Wheeler's interesting discussion that no need appears to the present writer to regard the lake villages apologetically as refugee settlements, or to invoke, however tentatively, traditions of lake dwelling surviving among Alpine stocks to explain their siting. The rivers of the Somerset marshes provided excellent harbours, which were used as such till the Middle Ages (when, for example, even Pilton much further inland was a port), and that on a coast which along its whole length from Bristol to Mount's Bay is otherwise acutely lacking in sheltered landing places. Seafaring people like the Veneti would regard the lake village sites all the more with favour in that their island character gave them the protection from interference from landward which sea traders establishing themselves on an alien coast particularly sought. While Dr Wheeler's chronology is followed in this paper, it would seem on general grounds not improbable that the establishment of such trading settlements anteceded the refugee movement which resulted from Caesar's campaign in 56 B.C., as was argued by Prof. Childe in Prehistoric Communities, p. 244.

^{page 64 note 3} Ibid., 204–10.

^{page 64 note 4} Ibid., 216, 7.

^{page 64 note 5} Ibid., 215.

^{page 65 note 1} There is a general case for seeking the early development of fine relief patterns in Gaul, where Arretine examples were available for copying throughout the 1st century B.C. Roman pottery with raised patterns reached the Hebrides in the late 1st or 2nd century A.D. in the shape of rare pieces of Sigillata, but too late to have influenced the earlier development of the relief technique, which moreover is based on patterns not typical of Sigillata. Similarly the ‘frilled’ decoration of such vessels as the tazzas mentioned on p. 63, n. 7 above could hardly have influenced technique (9) in the North, even if it were not highly improbable that such vessels ever reached the Hebrides.

^{page 65 note 2} The time was one of extensive tribal movement and early Irish traditions, such as those recorded by Bede and Nennius, though not datable, show the wide range over which raiding and settlement could occur.

^{page 66 note 1} Cf. the pots from the 9th-10th century rath of Lissue ( Bersu, G., U.J.A., 1947, 51 Google Scholar) and the so-called ‘souterrain ware’ (H. C. Lawlor, Nendrum, pls. xvi, et seq.).

^{page 66 note 2} Munro, R., Scottish Lake Dwellings, 112 and fig. 86Google Scholar.

^{page 66 note 3} Risary, Buaile. Beveridge, Erksine, North Uist, 209 and plateGoogle Scholar. The All Cannings Cross specimen is hardly a parallel as its shallow bowl is 3½ inches long (All Cannings Cross, 176 and pl. 37).

^{page 66 note 4} Glastonbury, II, Frontispiece, G19. It appears to have been fitted with an iron shank run in with lead for use as a pendant since traces of these metals remain and the base is filed down.

^{page 67 note 1} Beveridge, Erskine, North Uist, p. 247 and plateGoogle Scholar.

^{page 67 note 2} Rubber, from VI, 1, of olivine dolerite or basalt, 1¾ inches broad and ½ inch thick broken transversely. Disc, from XI, 1, of biotite gneiss, 3 inch diameter and 1¼ inches thick. A pumice specimen from I, 1 analysed by Dr Dunham was ‘composed of brown glass, clear in thin section, having refractive index 1.525 (thus probably a basaltic glass); ideomorphic crystals of labradorite and pale green augite up to 0.4 mm.’ The late Dr H. H. Thomas showed pumice from Rudh' an Dunain Chamber Tomb, Skye, to be of West Indian origin, carried by the Gulf Stream (P.S.A.S., LXVI, 212). Pumice was extensively used in the Hebrides from this period onwards and is still picked up on the shore for domestic use. It has recently been found on an East Ireland coastal site ( Dundrum, E. E. Evans, U.J.A., 1942, 12 Google Scholar).

^{page 67 note 3} A small piece of bog iron ore (limonite) might have been a stalactitic growth on the site. Clinkers, from III, O, III, I and V, 3, proved not to be iron slag products but probably chance products of combustion from the hearths.

^{page 68 note 1} Mr E. M. Jope most kindly undertook this analysis.

^{page 68 note 2} For the fabric of the Glastonbury triangular crucibles see Glastonbury, I, 301. I am greatly indebted to Mr St. George Gray for confirming the character of the Clettraval specimen by comparison with these, and to Mr R. B. K. Stevenson for comparing it with the triangular crucibles in the National Museum of Scotland.

^{page 68 note 3} Dr F. C. Fraser was good enough to examine the slight bone fragments surviving. Traces of bone were recorded from I, I, II, 2, V, 2, V, 3 and the Stage 1 hearth but were unidentifiable. The molar of a sheep from VI, 3 is disregarded as possibly recent.

^{page 69 note 1} P.S.A.S., VII, 165 ff. R.C.A.M., Outer Hebrides, no. 395, gives a much later and consequently less complete plan; it corrects the orientation of the souterrain and shows it to be set tangentially to the house wall.

^{page 70 note 1} This has been doubted, but Capt. Thomas, a naval surveying officer, is unlikely to have been mistaken on this point and the complete South Uist house described in P.S.A.S., III, 124 ff. (site now unknown) was corbelled over its central area.

^{paeg 71 note 1} Personal observation. Based on this and on the data in Admiralty Tide Tables, 1948, the elevation required would be 7½ feet, but the tidal data are slightly discrepant with those recorded on Admiralty Chart 2474 and in the West Coast of Scotland Pilot, Part II, 208, and the more conservative figure of 6 feet has been adopted. This accords with a figure given in North Uist, 213.

^{page 71 note 2} Much of the land sinking has occurred in historic times: see the documents quoted in North Uist, vi ff. and the facts given about Baleshare on p. 48. In the rest of this paper the name Uist, where used without qualification, will indicate the whole group of islands, as in early usage.

^{page 71 note 3} Even in the 17th century, when the great part of this land must already have been lost, North Uist was well known for its produce of barley, which, with the charitable character of the people, drew many strangers to the island to obtain gifts of grain. For this fact, and for the mediaeval practice of travelling round the country to obtain gifts and hospitality (‘thigging’), see Martin, Martin, A Description of the Western Islands, 2nd ed., 1716, 78, 79 Google Scholar.

^{page 71 note 4} For a map showing the modern distribution of machair land, see Darling, F. Fraser, Natural History of the Highlands and Islands, 1947, 158 Google Scholar.

^{page 71 note 5} North Uist, 121 ff; plan reproduced in R.C.A.M., Outer Hebrides, fig. 11.

^{page 72 note 1} North Uist, 207 ff.; R.C.A.M., Outer Hebrides, no. 270.

^{page 73 note 1} North Uist, 200 ff.; plan reproduced in P.P.S., 1947, p. 22, fig. 8.

^{page 73 note 2} P.S.A.S., LXVI, 32 ff.

^{page 74 note 1} P.S.A.S., LXV, 300 ff.

^{page 75 note 1} P.S.A.S., LXVI, 42 ff. with plan. In R.C.A.M., Outer Hebrides, no. 271, there is a partial plan which differs in some details.

^{page 76 note 1} The report was published by Dr Callander from Dr Beveridge's draft in P.S.A.S., LXV, 317 ff. See also North Uist, 212 ff.

^{page 77 note 1} P.S.A.S., LXV, 354. The finds from this site are not preserved in the National Museum.

^{page 77 note 2} A valuable comparative discussion of the material by Dr Callander is in P.S.A.S., LXV, 349 ff., which is supplemented by LXVI, 64 ff. For the wider connections of the material types see Childe, V. G., Prehistory of Scotland, 237 ff.Google Scholar

^{page 77 note 3} I am greatly indebted to Miss Margery Platt of the Royal Scottish Museum for examining, in co-operation with Mr R. B. K. Stevenson, a selection of the bone implements from Vallay preserved in the National Museum. As it may be hoped that this valuable report may be published, I need not anticipate it further than to say that deer and whale bones were much the most commonly used for implements, followed at a long interval by sheep and ox, and to stress the inevitably tentative character of identifications based on worked bones. Dr Beveridge did not send to the Museum the quantities of unworked bones which he must have found.

^{page 78 note 1} The use of this implement has been much discussed: for references see Maiden Castle, 303 ff. It occurs at a number of Scottish sites, all coastal, and in England at Glastonbury, but also at sites such as All Cannings Cross and Maiden Castle where fishing is doubtful. There are variations between examples: the points are sometimes sharp, sometimes blunt; the butts are sometimes rough, sometimes trimmed square; the implements are sometimes socketed, and occasionally pinned as well, and sometimes not socketed. The socket where it occurs is ⅝ inch in diameter or less, and it has been argued that this would make the shaft too weak for use as a dart or spear against land animals. Use as a shuttle has been proposed but seems excluded unless the point is blunt and the butt trimmed; moreover, the modern analogy quoted is questionable. Use for fishing has not before been suggested but is supported by its association with harpoons at the two Vallay sites and its absence from the other sites. All the examples are socketed and would seem effective as fish spears where sharp and, mounted in pairs, as eel prongs where blunt. Flounders are still speared on Vallay Strand with a single iron spike stuck into the cleft end of a stick; when made for the purpose the spike is barbed, but the unbarbed point of a hay fork is used when the specialised instrument is not to hand (Beveridge, op. cit. 323 and personal observation).

^{page 78 note 2} Antiquity, June, 1947, 87 and 95.

^{page 78 note 3} The industry now flourishes only in the Faroes. For a useful account of it see Williamson, K., The Atlantic Islands, 1948, ch. 4Google Scholar.

^{page 78 note 4} Martin, Martin. A Description of the Western Islands, 2nd ed., 1716, 66 Google Scholar.

^{page 79 note 1} A rather larger skull disc, pierced with three holes, from Hillhead Broch, Keiss (infra), is in the National Museum.

^{page 79 note 2} From Bac Mhic Connain (P.S.A.S., LXVI, p. 58, fig. 14). It is obviously closely analogous to the bronze trace-loops from Stanwick, Yorks. (Leeds, Celtic Ornament, fig. 31) which can be fairly well dated to mid 1st century A.D.

^{page 79 note 3} Foshigarry is the only site from which the pottery is fully published. Callander illustrated all the decorated sherds and a good selection of rim sections (P.S.A.S., LXV, 342 ff. and figs. 23–25). For other sites there are descriptions and sometimes a few photographs, but the main source of information is the material deposited in the National Museum and this is by no means complete.

^{page 79 note 4} Including the triskele, which, it may be noted, occurs four times at Meare but not apparently elsewhere in the south. (Meare, I, 19).

^{page 80 note 1} Relying on Dr Callander's description of the pottery, which is not preserved.

^{page 80 note 2} Relying on the selection of the pottery preserved in the National Museum.

^{page 81 note 1} P.S.A.S., IX, 361 ff. R.C.A.M., Orkney, no. 818. Little beyond Petrie's plan was published at the time but the Commission had access to his notebook.

^{page 82 note 1} P.S.A.S., LXXI, 129 ff. and LXXIII, 167 ff. R.C.A.M., Orkney, no. 245(3).

^{page 82 note 2} Mr Calder assumes the finds to belong to the secondary use of the house as a pottery workshop, but many of them are inappropriate to such use and the supposition does not seem inherently likely. A bead of blue coloured patinated glass found on the under side of the turf is ignored as improbably connected with the house. There were other unrelated buildings, shown on the right in fig. II.

^{page 82 note 3} P.S.A.S., XIX, 23 ff. and XXIV, 451 ff. R.C.A.M., Orkney, no. 195. The pottery is unpublished.

^{page 85 note 1} P.S.A.S., LXVIII, 224 ff., and for the hut circle and Dr Curle's final summary, LXX, 237 ff. Dr Curle's ‘stage I’ of house III refers to the underlying courtyard house.

^{page 86 note 1} The interpretation of the structural data which is here tentatively offered is a simplification of that of Dr Curle, which inferred extensive reconstruction, including the complete rebuilding of the northern arc of the wall. The space interpreted above as a wall-cavity was devoid of occupation debris and its floor, in part paved, was at a higher level than that of the interior of the house. Similar wall cavities will be described below. The souterrain was built with the house and its opening is hardly intelligible if the house wall was subsequently moved inwards.

^{page 87 note 1} P.S.A.S., XLI, 11 ff. R.C.A.M., Shetland, no. 1149.

^{page 88 note 1} P.S.A.S., XLI, 25. A photograph published by the Royal Commission (fig. 493) shows the surface of the house wall to be unfaced towards the cavity.

^{page 89 note 1} P.S.A.S., LV, 110 ff. R.C.A.M., Skye, no. 479.

^{page 90 note 1} The calculation is set out in P.P.S., 1947, p. 13. The term ‘scarcement’ indicates an insetting of the inner face of the parapet (like the ‘shelf’ in bay VI at Clettraval) which is a substitute for a corbel course.

^{page 91 note 1} Dated by Mr Kilbride-Jones to A.D. 50–80 (P.S.A.S., LXXII, 367, 390).

^{page 91 note 2} P.S.A.S., XLIX, 57 ff. R.C.A.M., Skye, no. 508.

^{page 92 note 1} By DrCurle, James, P.S.A.S., LXVI, 289 Google Scholar.

^{page 92 note 2} There are also two double-conical beads of Late Saxon type.

^{page 93 note 1} Curle, A. O., P.S.A.S., LV, 83 ff.Google Scholar

^{page 93 note 2} R.C.A.M., Outer Hebrides, no. 393.

^{page 95 note 1} Omitting Ireland in our present uncertainty regarding the character and dating of the Iron Age raths For souterrains elsewhere see: A. Blanchet, Les Souterrains Refuges de la France; Hencken, H.O'N, Cornwall and Scilly, 132 ff.Google Scholar; Anderson, J., The Iron Age, 281 ff.Google Scholar; Childe, V. G., Prehistory of Scotland, 212 ff.Google Scholar; the volumes of the Scottish Royal Commission; and, for Jutland, Hatt, Antiquity, XI, 172 ff.Google Scholar

^{page 95 note 2} Trail, W., ‘The Broch of Burrian,’ Arch. Scot., V, 342 Google Scholar.

^{page 95 note 3} Broch, Okstrow, Arch. Scot., V. 78 Google Scholar. It is claimed that the Aikerness Broch chamber was fed by a spring (R.C.A.M., Orkney, 77) but there was only a foot of water in a cranny of its rock cut bottom (fig. 131).

^{page 95 note 4} East Broch, Burray, Arch. Scot., V, 72 Google Scholar. The souterrain was entirely external to the broch. In one case (Burrian, near Garth, Harray) the chamber was as much as 10 feet 6 inches by 6 feet by 13 feet high, corbelled over (R.C.A.M., Orkney, no. 21).

^{page 96 note 1} A distribution map is to be desired. The northern examples are recorded in the Royal Commission's volumes. For summaries of the eastern ones and references to some of the many scattered papers see Anderson, J., The Iron Age, 281 ff.Google Scholar, Childe, V. G., Prehistory of Scotland, 213 ff.Google Scholar and Simpson, W. D., The Province of Mar, 71 ff.Google Scholar

^{page 96 note 2} A. Blanchet, op. cit., 37, 53 and 84.

^{page 96 note 3} At Castlelaw, Midlothian, iron working took place in the chamber of a souterrain, which must then have been unroofed, since smelting would otherwise have been impracticable (P.S.A.S., LXVII, 362 ff.).

^{page 96 note 4} This is not to say that, though seemingly liable to be a death trap, a souterrain may not have served as a temporary hiding place in an emergency; but, even in primitive life, emergencies are rare events. Some souterrains had secondary exists, which have been pointed to as minimising the risk entailed in using them as refuges; these ‘exits’ are commonly minute, and sometimes impassable, and were probably ventilation shafts. (At Foshigarry the aperture was 16 inches by 22 inches.)

^{page 96 note 5} Curle, A. O., P.S.A.S., XLV, 18 ff.Google Scholar

^{page 96 note 6} Abercromby, Lord, P.S.A.S., XXXVIII, 102 ff.Google Scholar

^{page 97 note 1} P.S.A.S., VI, 13, 14.

^{page 97 note 2} Curle, A. O., P.S.A.S., XLV, 23 ff.Google Scholar

^{page 97 note 3} Thorneycroft, W., P.S.A.S., LXVII, 187 ff.Google Scholar

^{page 97 note 4} W. D. Simpson, The Province of Mar, fig. 22.

^{page 97 note 5} P.S.A.S., IX, 147 (s.v. ‘no. 1’) and VI, 402 ff. The latter paper records that another hut circle excavated in the immediate neighbourhood—one of a group of solid-walled houses, 21 feet to 27 feet in diameter—had within it a ring of seven post-holes and a central post-hole. For the distribution of weaving combs see the paper first quoted and Childe, V. G., Prehistory of Scotland, 239 Google Scholar.

^{page 97 note 6} P.S.A.S., VI, 210 ff. and III, 440 ff.

^{page 97 note 7} For the distribution of stone cups see Callander, J. G., P.S.A.S., V, 145 Google Scholar, and V. G. Childe, op. cit., 246. The associations recorded with cairns and stone circles can best be disregarded; they derive from very early records at a time when hut circles and their associated mounds were not recognised as habitation sites. Outside Scotland stone cups have been found in the Isle of Man, Ireland and Cornwall.

^{page 97 note 8} One of the souterrains of the Tay group contained Roman pottery of Ist century type and the ist century coin from the Don group has been mentioned above.

^{page 98 note 1} P.S.A.S., LXXV, 23 ff. and LXXX, 11 ff.

^{page 98 note 2} R.C.A.M., Caithness, no. 250.

^{page 99 note 1} This is also Mr Angus Graham's belief, as he has been good enough to confirm, but he suggests that further examination of the door and the external scarcement is desirable. Anderson's statements in his original excavation report in Arch. Scot., V, 131 ff.Google Scholar, do not fully accord with his much later account in The Iron Age, 229 when he had come to believe that all outbuildings of brochs were secondary. Seeing that other brochs are known which had a second door opening through the broch wall, it may be desirable to await a re-examination of the structure of Yarrows before concluding that the broch originally had no byre.

^{page 99 note 2} Beveridge, Erskine, Coll and Tiree, 73 ff.Google Scholar

^{page 99 note 3} For the pottery see idem, 174 ff., and the related plates. The well-known stag sherd from Coll (Dun Borbaidh) may or may not belong to the aisled round-house culture.

^{page 100 note 1} See, for example, R.C.A.M., Sutherland, figs. 40–42, 45–47, and Caithness, fig. 20; P.S.A.S., LIX, 221 ff.

^{page 100 note 2} P.P.S., 1947, 26–8. For Dr Bersu's dating of the Isle of Man houses see U.J.A., 1947, 45, and, of the Scotstarvit house, Report of the Scottish Regional Group of the Council for British Archaeology, May, 1947. I am indebted to Mr Basil Megaw for the information that the Ballacagan house in Man produced a perforated bone bobbin typical of the aisled roundhouse culture, and also for showing me the poor pottery from these houses. This has nothing in common with the wares of South-west Britain and makes it unlikely that the Man settlement proceeded direct from thence.

^{page 100 note 3} There were foundations of two rectangular structures at Meare, but there is nothing to indicate that these were dwellings; they may have been byres or storehouses (Meare, 1, 10).

^{page 100 note 4} R. Munro, Scottish Lake Dwellings, For Buston see pl. IV, figs. 189, 190 and summary on p, 204 ff.

^{page 100 note 5} P.P.S., 1946, 164.

^{page 101 note 1} Unexcavated buildings which can fairly confidently be compared with Clettraval are: R.C.A.M., Outer Hebrides, nos. 199, 354, 393 and P.S.A.S. III, 124. Sites which have yielded characteristic pottery with structural remains less or more denned are: Kilpheder, Geirisclett, Buaile Risary, Udal and, in Vallay, Rudh' an Duin, Tota Dunaig and the Old Cattlefold (material in the National Museum, Edinburgh). Also Voiskinish, Benbecula, and Sitheana Phiobaire, Daliburgh, S. Uist (information from Mr T. C. Lethbridge. There are numerous other buildings, including a number on islets in lochs, which on superficial evidence are likely to be of the type of Clettraval.

^{page 102 note 1} The Royal Commission has named these ‘galleried duns’ and argued that they are prototypes of the brochs (Outer Hebrides, xxxvi; Orkney and Shetland, 31 ff.). They are, however, of very varied type, with little in common except their use of cavity walls. None has been excavated. They would seem more comparable with such cavity-walled structures as Ardifuar and Druim an Duin on Loch Crinan and Kildonan in Kintyre, which are tentatively attributed to the later Roman period or early Dark Ages (see the discussion of Kildonan, excavated by MrFairhurst, H., in P.S.A.S., LXXIII, 220 ff.Google Scholar). Tirefuar in Lismore, which has the same wide entrance, broad scarcement and internal batter as Ardifuar, may belong to this class.

^{page 102 note 2} P.P.S., 1947, 29–32.

^{page 102 note 3} Idem, 4, n. 1.

^{page 102 note 4} R.C.A.M., Outer Hebrides, no. 124. The house is some 15 feet in diameter within a wall, probably hollow, some 10 feet thick and standing to 6 feet above the loch.

^{page 102 note 5} One souterrain, at Nisibost, had a corbelled chamber of some size and produced a rotary quern (P.S.A.S., III, 140).

^{page 102 note 6} P.S.A.S., LVIII, 185 ff.

^{page 103 note 1} At Bragar: R.C.A.M., Outer Hebrides, no. 23 and P.S.A.S., LVIII, 198.

^{page 103 note 2} At Valtos: R.C.A.M., Outer Hebrides, nos. 84 and 98.

^{page 103 note 3} P.S.A.S., LXXIII, 55 ff. The pottery, however, has some resemblance to North Irish material which is supposedly of the Dark Ages.

^{page 103 note 4} The mattock was from the nearby souterrain of Saverock by Kirkwall (P.S.A.S., LXV, 352).

^{page 103 note 5} The six coins were from Lingrow, which is on the Bay of Scapa, where Agricola's fleet, and subsequent merchants, presumably landed. They were found scattered through the outbuildings ( Anderson, J., The Iron Age, 244 n.Google Scholar).

^{page 103 note 6} At Deerness, east of Scapa Flow. A quantity of high-class pottery has been recovered as a result of sea erosion of the site. (In the National Museum unpublished, as is most of the pottery from the two brochs).

^{page 104 note 1} After describing Claudius' annexation of Britain, Orosius says: ‘Orcades etiam insulas ultra Britanniam in Oceano positas Romano adiecit imperio.’ For Prof. Childe's acute deduction that this means that Orcadian chiefs sent envoys to make formal submission to Claudius see Scotland before the Scots, App. IX, p. 129.

^{page 105 note 1} The objects were found 5 feet below the surface of the links of Stacwick Bay on the site of a structure now undiscoverable (P.S.A.S., XLVII, 9 and R.C.A.M., Orkney, no. 475).

^{page 105 note 2} P.S.A.S., IX, 120.

^{page 105 note 3} Mr Angus Graham records ‘particularly strong’ external defences to 15 broehs in Shetland, 7 in Orkney, 7 on the Northern Mainland and none in the Hebrides (P.S.A.S., LXXX, forthcoming).

^{page 105 note 4} Arch. Scot., V, ii, 349 ff.Google Scholar and 364.

^{page 105 note 5} P.S.A.S., LXVIII, 512.

^{page 106 note 1} Prehistoric Communities, 249.

^{page 106 note 2} For confirmatory dating evidence for the northern brochs see P.P.S., 1947, 29–33.

^{paeg 107 note 1} P.S.A.S., LXVI, 341, 393–4. The silver brooch found near Carn Liath in South-east Sutherland was sub-Roman and at earliest of the 4th century (P.S.A.S., LXVI, 338).

^{page 107 note 2} P.S.A.S., XXXV, 112 ff. and XLIII, 11 ff.

^{page 107 note 3} From personal examination of the pottery, which is in the National Museum but has not been published. Stray references make it appear that there were two other decorated examples which I did not find, the one showing a relief decorated band (Everley Broch) and the other ‘an impressed chevrony pattern’ (Keiss Broch)—see P.S.A.S., LVIII, 198 and XXXV, 125. A tenth house belonging to this group (Hillhead) has recently been excavated by Mr Calder; it produced only undecorated pottery, which, it may be noted, included a bead rim.

^{page 107 note 4} Arch. V (1777), 216 Google Scholar, with plan and elevation.

^{page 108 note 1} An exception may be Dun Lagaidh, which has been examined by Mr Calder and Dr Steer in a recent survey of Loch Broom in Wester Ross. A grass-covered mound covered an apparently broch-like structure lying within the enclosure of an earlier murus gallicus fort. (Report of the Scottish Group of the Council for British Archaeology, May, 1948.) Loch Broom is far beyond the limits of the murus gallicus tribes and the exception, if established, emphasises rather than modifies the statement above.

^{page 108 note 2} The area is, however, a wooded one and substantial houses built in timber are a possibility; the excavations, which were usually satisfied with clearing out the souterrain, would not have found these. The habitation floor overlying the souterrain at Castle Newe (supra) produced a late ist century coin (a denarius of Nerva) and a pair of 2nd century enamelled bronze armlets which, unless loot, suggest a house of some wealth.

^{page 108 note 3} P.P.S., 1947, 17.

^{page 109 note 1} For the dating of the introducton of the rotary quern, see P.P.S., 1947, 29, n. 2, and the references there given. The earlier records of broch excavations seldom distinguish between saddle and rotary querns, but some data are recorded from excavations in the present century. Thus it appears that all querns were rotary at Dun Beag in Skye, which was built at the beginning of the ist century A.D.; that six out of seven querns were rotary at Ayre broch on Scapa Flow, which was built before the middle of the century; that all querns were saddle querns in the lower (late ist century) level at Aikerness broch in the Central Orkneys; that a majority of all querns recovered from the long occupation of Midhowe broch in the Central Orkneys were saddle querns; and that saddle querns were still in common use in the North-east Caithness settlement, which was prpbably founded before the middle of the 1st century. They had, however, disappeared before a Class II broch in South-east Caithness (Ousedale Burn) was built, that is perhaps by A.D. 100.

^{page 109 note 2} For the English aisled roundhouses see pp. 26–27 of the former paper. For Gaul we know generally that the houses outside the oppida were round and walled with wood and wattle and we have fair information about the construction of the souterrains, though not about the houses to which they belonged (A. Blanchet, op. cit. supra). For the Cornish houses and souterrains see H. O'N. Hencken, Cornwall and Scilly; for Chun Castle see Leeds, E. T., Arch., LXXVI, 205 ff.Google Scholar and LXXXI, 33 ff.

^{page 112 note 1} Quoted by Fittis, , Sports and Pastimes of Scotland, 39 Google Scholar.

^{page 112 note 2} P.S.A.S., LXXV, 35. Enquiries confirm Dr Curle's view of the height of wall required, but it is hardly possible to confirm the assumption that wolves were so numerous, and accordingly so hungry, as to attack the farmyard. Enquiries as to the modern use of so high a farmyard wall in countries infested by wolves have had negative results.

^{page 112 note 3} In historical times islets were regarded as adequate refuges from the attack of wolves. ( Millais, J. G., Mammals of Great Britain and Ireland, I, 193.Google Scholar)

^{page 113 note 1} Though experience shows these brochs not to be unclimbable.

^{page 113 note 2} P.P.S., 1947, 30–32. There is satisfactory evidence of five broch towers: Mousa, on a small island lying off Lerwick in Shetland; Dun Dornadilla, among the mountains of North-west Sutherland; Dun Carloway, on Loch Roag in North-west Lewis; and the two Glenelg brochs lying close together in a narrow glen hidden from the sea among the mountains of the mainland opposite Skye. There may, of course, have been others, and the strongest Claimants are in Shetland (Culswick, Burraness and Clickhemin); but such massive structures are not difficult to identify from their ruins and not easy to remove.

^{page 116 note 1} According to Kendrick, T. D., History of the Vikings, 339 Google Scholar.

^{page 116 note 2} According to Powell, F. York (Ency. Brit., 11th Google Scholar edn., s.v. Iceland) and based on the census of A.D. 1097 which showed 4,560 bondir (land owners) as taxable for the expenses of those attending the Althing (Landnamabok, trans. Ellwood, III, XX, and p. 162 n).

^{page 121 note 1} Reference may be made to evidence from houses described in Part II of the paper in general support of this conclusion, particularly Machair Leathann and Usinish (monolıthic architrave) and Jarlshof (corbelled construction). The alternative that one of the three slabs A, B, C, was used as a pillar, like the pillar in Bay VIII, is incompatible with the positions in which the slabs were found, but could be supported if considerable reconstruction, including the rebuilding of the kerb of Bay IX, were assumed to have taken place in Stage III. The assumption is unlikely.

^{page 122 note 1} The line of slabs mentioned as having apparently fallen from the wall shelf in Bay VI presumably represents part of the corbelling in that bay.

^{page 122 note 2} A fragment of evidence confirming that the outward slope of the bay roofs was slight comes from a sectional drawing recently published from the notebook of Sir Henry Dryden (R.C.A.M., Orkney, no. 438). It relates to the broch of Wasso in Orkney and the heights of a slab pier and of the scarcement indicate an outward fall of 1 foot in the 6 foot span from the pier to the parapet of the main wall. There is ample evidence that the aisle roofs of wag byres (supra) hada similarly slight outward slope. That the bay roofs in brochs drained outward into the wall core I have shown from the evidence of Dun Troddan in Glenelg (P.P.S., 1947, p. 23).

^{page 123 note 1} Based on Capt. Thomas, F. W. L., R.N., P.S.A.S., VII, 153 ff., 1869 Google Scholar; DrMitchell, A., The Past in the Present, 1880, 49 ff.Google ScholarPubMed; and personal study of surviving houses.

^{page 123 note 2} Norse Building Customs in the Scottish Isles, 27 ff.

^{page 123 note 2} Op. cit., p. 47.

^{page 123 note 3} At Jarlshof; cf. also Freswick, Caithness ( DrCurle, A., P.S.A.S., LXIX, 265 ff.Google Scholar. and LXXIII, 71 ff., and Prof. V. G. Childe, ibid., LXXVII, 5 ff.).

^{page 123 note 5} Actually the great majority of blackhouses are no longer than most English cottages. The idea of their great length arises from the often quoted and illustrated examples reported on by Capt. Thomas from Western Lewis which, as he makes clear, are abnormal.

^{page 124 note 1} A. Mitchell, op. cit.; Curwen, E. C., Antiquity, Sept., 1938 Google Scholar.

^{page 124 note 2} By his excavations at the Car Dyke near Cambridge (verbal information in advance of publication).

^{page 124 note 3} In a book just published, Merlin's Island, Mr T. C. Lethbridge has re-argued the case for continuity of culture from Roman times, adducing additional data, inter alia clothing and the lamp (loc. cit. 42). I am glad also to note his support in the argument developed above that the blackhouse was not of Scandinavian origin (loc. cit. 80 ff.).

^{page 124 note 4} The impact of the modern world on the Hebrides began only after 1609, when the treaty known as the Band and Statutes of Icolmkill was made with the chiefs. Thereafter the islands were gradually absorbed into the kingdom of Scotland through religious, educational and economic contacts. We have detailed knowledge of the Hebrides and St. Kilda in the later 17th century from the works of Martin Martin of Kilmartin in Skye, who was trained in the native traditions as a physician and emigrated to London, where he was associated with the Royal Society group and was encouraged to write A Late Voyage to St. Kilda, 1698, and A Description of the Western Islands, 1703 (quotations from the 2nd edn. 1716). For St. Kilda and the Shetlands we have the records of the naval surveying officer Capt.Thomas, F. W. L., R.N., in P.S.A.S., VII, 153 ff.Google Scholar; these relate to the mid 19th century. For the Faroes in the late 18th century we have Landt, G., A Description of the Feroe Islands, trans, from Danish, , 1810 Google Scholar; and in current times, Williamson, K., The Atlantic Islands, 1948 Google Scholar.

^{page 125 note 1} In the Gaulish souterrains the finding of charcoal is reported to be general (A. Blanchet, op. cit., 84).