¹ The following photographs in the Cambridge University Collection are relevant for a study of the camp: BVD 85-102, BVE 1-13, K17-AJ 5-15 taken 26 July 1975; CAF 28-49, 52-53 taken 21 July 1976; CDE 19-30 taken 20 July, and CDV 51-61, K17-AS 50-51 taken 31 July 1977.

² I am grateful to Mr A. A. Smith, of Pitbee, Pitcaple, who kindly gave permission for work in the fields of Logie Home farm; to Mr John Stuart of Hillhead of Lethendy who permitted digging in the fields of Westerton, and to Mr H. Ohldag of Easterton, for work near Easterton.

³ In both 1975 and 1976, G. Maxwell and N. B. Clayton helped with the digging, being joined in 1976 by R. M. Ogilvie and by D. Willington with boys from Trinity College, Glenalmond; in 1977 a few trenches were dug by G. Maxwell and C. D. St. Joseph. Without all this help the results could not have been attained so quickly.

⁴ Roy's MS map of Scotland, of about 1750, shows, in the neighbourhood of Dumo, a patchwork of cultivated fields spreading out from the stream valleys. The area of the camp seems still to have been unenclosed moorland.

⁵ Photograph No. K 29002, from Skyviews and General Ltd., 26 Stooks Hill, Leeds 11. Mr Ohldag showed me this photograph.

⁶ JRS lxiii (1973), 231–32.Google Scholar

⁷ For ease of comparison with previous lists of these camps (JRS lxiii (1973), 231–3Google Scholar; lxvii (1977), 143-4) the areas are given in imperial measure.

⁸ At Stonehaven there is a shallow embayment of the coast where two small streams, the Cowie Water and the Carron Water, reach the sea c. 700 m apart. At the S end of the Bay a rocky point cuts off a small cove, the site of the medieval and modern harbour. A mass of rocks at the harbour-entrance was only removed in 1825. How usuable this cove would have been in its natural state is uncertain. Perhaps Roman transports would have been beached on the strand of Stonehaven Bay, though this is wide open to the E. Mr Angus Graham has kindly supplied me with notes on the harbour.

⁹ See Table, JRS lxiii (1973), 231.Google Scholar

¹⁰ It overlies a camp of 10 acres, a camp of 63 acres, a signal-station and the defences of an annexe attached to the complex of earthworks that remain from superimposed Flavian and Antonine forts.

¹¹ See Table in JRS lxiii (1973), 230.Google Scholar

¹² JRS lxiii (1973), 231.Google Scholar

¹³ Agricola, ch. 25, 1: ac saepe isdem castris pedes quisque et nauticus miles mixti copiis el laetitia sua quisque facta, …

¹⁴ Discovery and Excavation, Scotland, 1977, 53.

¹⁵ JRS lxiii (1973), 223–4.Google Scholar

¹⁶ JRS lxiii (1973), 233.Google Scholar Dio, indeed (Ixxvi, 13, 3), reports that Severus ‘did not desist until he had approached the extremity of the island’. The phrase is too vague and in any case too unrelated to the true shape of Scotland to be pressed closely in dating the most northerly camps, neither series of which approaches the actual extremity of the island.

¹⁷ The site at Logie (c. NO 699629) in Angus, provisionally identified on the evidence both of air-photographs and of ditch-sections as a Roman camp (JRS lxiii (1973), 226Google Scholar; lxvii (1977), 140) was examined by further digging in 1977. This established lengths of a V-shaped ditch, running for 400 m or more, in two directions at right-angles. However, no gates could be identified, and the angle between the two lengths is sharp. These results show that the remains cannot be accepted as Roman. They are best explained as seventeenth or eighteenth-century boundary-ditches of vanished plantation-belts. After the ditches had largely silted up, field-drains were inserted in some lengths in the first half of the nineteenth century.

¹⁸ JRS lxvii (1977), 144Google Scholar, where this argument was first developed.

¹⁹ Many suggestions as to the site of the battle that was the climax to Agricola's final campaign have been made over the last two centuries. They range between the neighbourhood of Ardoch and the Moray Firth. For recent discussions see O. G. S. Crawford, The Topography of Roman Scotland (1949), 130-3, who on the evidence then available placed the battle at Raedykes; Burn, A. R., P.S.A. Scot. lxxxvii (1955), 127–33Google Scholar (between Knock Hill and the Pass of Grange); Henderson-Stewart, D., Trans. Anc. Mon. Soc. 8 (1960), 75–85Google Scholar (the Pass of Grange). In their recent edition of the Agricola, R. M. Ogilvie and I. A. Richmond, De vita Agricolae (1967), 252, consider ‘that it may have been in the vicinity of the Pass of Grange’.

²⁰ Britannia ix (1978), 397.Google Scholar

²¹ Agricola, ch. 25, 3 and 26, 1, referring to the sixth campaign.

²² Other possible positions are near Stonehaven, where an advancing army would have to skirt the edge of the Highlands as this approaches the sea, and the neighbourhood of the Pass of Grange. Raedykes, the only Roman camp known near Stonehaven, is the smallest in the table on p. 275, and thus unlikely to have served as Agricola's base-camp before the battle. Moreover it lies N of the point where an army's manoeuvres would be most restricted between the hills and the sea. As to Grange, the Pass that forms the lowest part of Stratha Isla, now followed both by a main road and a railway, affords the easiest route to Moray. Knock Hill (NJ 538551) which has sometimes been linked with the battle, is a distinctive conical hill summit 430 m) some 5·5 km NE of the Pass and visible over a wide area. The question may be asked whether it would have afforded enough ground for the large tribal forces that Calgacus had assembled? Moreover, the nearest known Roman camp, 109 acres in size, is that at Muiryfold (NJ 489521), in a position that could only have been reached after the narrows of the pass between Little Balloch Hill and Sillyearn Ridge had been traversed. (For references to Raedykes and to the Pass of Grange, see note 19).

²³ The Mither Tap, NJ 683224; Dunnideer, NJ 613281; Pittodrie, NJ 694244; the Barmkyn of North Keig, NJ 599200; Brace's Camp, NJ 768190; and perhaps the lightly defended enclosure at Tillymuick, NJ 649245. The land in Lower Garioch is amongst the most fertile in NE Scotland.

²⁴ Antiquity xxv (1961), 270–1 and fig. IGoogle Scholar.

²⁵ That some exhortation both on the Roman and on the Caledonian side took place before the battle may be likely enough, but the two speeches, supposedly by Calgacus and by Agricola, which seem to be so appropriate to the situation, have long been accepted as rhetorical compositions of Tacitus.

²⁶ A gently sinuous crop-mark was observed from the air in 1977, near the edge of a small terrace on the N side of the Urie, by Strathorn farm. Two trenches dug in September 1977 showed this to represent a shallow feature 1·85 m wide, with steep sides and a flat bottom 0·70 m below the surface. An early boundary of arable land, or an open leet for water are possible explanations. The feature extends between NJ 6853 2725 and NJ 6895 2721.

²⁷ A recent reconstruction of the battle of Vetera in A.D. 70, when Cerialis had five legions and appropriate auxiliary infantry and cavalry at his disposal, shows a battle-line of about 4 km. The auxiliaries were in front, the legions disposed behind. However, this battlefield was constricted between the Rhine and a small tributary, the Poll, while marshy ground hampered movement. Petrikovits, H. von, Die römischen Streitkräfte am Niederrhein (Dusseldorf, 1967)Google Scholar (Führer des Rheinischen Landesmuseums in Bonn), Bild 31 and pp. 59-61.

²⁸ Agricola, ch. 35, 4.

²⁹ Air-photographs taken in 1977 (K17-AS 50-1) show at about the point NJ 6887 2721 a number of small rectangular pits, just visible on the surface as shallow depressions. They are probably gravel-diggings. Other irregularly-shaped pits occur beside the main road at NJ 6868 2685 (? shallow quarries), and at NJ 6817 2703, NJ 6830 2705, and NJ 6848 2710. None of these may be ancient.

³⁰ If the equation Bennachie = Mons Graupius be accepted, this very striking mountain would join the four major tribes (Caledonii, Silures, Ordovices, Brigantes), the three large rivers (Clota, Bodotria, Taus) and the two islands (Mona, Thule), that comprise, with one unknown tribe (Boresti) and an unidentified harbour (Trucculensis), the only geographical names used by Tacitus in his account of Britain.