^{page 75 note 1} It seems convenient to use this term for the massif, bisected by the river Mühlbach, which is bounded on the east by the Red Tower Pass, on the west by Petroseny and the Vulcan Pass, and on the north by Mühlbach (Szász-Sebes). It is usually divided into Sebeshely Mts. (west), Mühlbach Mts. (centre), and Cibin Mts. (east); but this demarcation into definite ranges is neither practically precise nor scientifically justifiable. Its chief heights lie about the source of the river Mühlbach—Cindrelul (2243 m.), Steflesci (2244), Piatra Alba (2180); Verfu lui Petru (2133) and Surian (2061) are farther west.

^{page 75 note 2} The only exception known to me is Tundervár near Torda, which Finály (Arch. Jahrbuch, 1910, 390) compares with the Muncel Cetate. (In the maps of the Austro-Hungarian Military-Geographical Institute (Zone 22, Col. xxx) a Cetate is marked near Orlát, about 15 km. west of Nagyszeben (Hermannstadt) i.e. in the eastern half of this triangle.)

^{page 76 note 1} This district was visited some years ago by Professors Finály, Lange, and Kuzsinszky of the University of Budapest, commissioned by the Hungarian Government to report on the desirability of reserving these remains from the operations of the Forestry Department. I understand that they visited the Muncel and Piatra Rosie, which sites are marked on Finály's wall-map of Roman Hungary, but I am not aware that any formal report on the antiquities as such was published or contemplated,

^{page 76 note 2} i.e. in Transylvania.

^{page 77 note 1} C. iii, 1415, 1416, Arch-epig. Mitt. xiii, 194.

^{page 77 note 2} Téglas, Ung. Revue, 1893. p. 438.

^{page 77 note 3} For this and generally Gooss, , Chronik der archäologischen Funde Siebenbürgens (= Archiv. für siebenbürg. Landeskunde xiii, Heft 2, pp. 203–338Google Scholar), which is by no means superseded by Martian's Archäologisch-prähistorisches Repertorium für Siebenbürgen (Mittheilungen der anthropologischen Gesellschaft in Wien, 1909, p. 321 ff.

^{page 77 note 4} Eph. Epig. iv. 188.

^{page 77 note 5} Alsó-Városviz was an important point, for it was here that the old Dacian road which ran from near Mühlbach across the mountains (parallel to Maros) reached the valley of the Városviz.

^{page 78 note 1} Theb. i, 20, conjurato dejectos vertice Dacos. Silvae, i, 1, 7, domus ardua Daci; 1, 1, 80, tu tardum in foedera montem Longo Marte domas iii, 3, 169, quaeque suum Dacis donat clementia montem. Vollmer refers in his note to Dio's ὄρη ἐντετειχισμένα.

^{page 78 note 2} Redrawn for the J.R.S. by Mr. J. Addison. The illustrations in plates II-v are from Cichorius.

^{page 79 note 1} A fragment of Trajan's own Commentaries, as is well known, indicates that the advance was made by this route. I am glad to find myself, in the localisation adopted, in agreement with Mr. Stuart Jones, and, with the exception of one detail, with von Domaszewski; but this is in no way essential to the main subject of this paper, which is concerned with the three fortified places shewn at XIV, XVIII and XXII, irrespective of their identification, highly probable as that may be, with Arcidava, Berzovia and Tibiscum. It tends, however, to come into conflict with the ‘Double Advance’ theory, which has forced Petersen, its chief exponent, into difficulties in the later halting-places. He whisks away the ‘marching camp’ and permanent fortress of XXI and XXII (Caput Bubali and Tibiscum) to the neighbourhood of the Key of Teregova, and reintroduces here his ‘Lower Moesian army.’ One result of this (the only one which here concerns us) is that he is forced to telescope Centum Putea, Berzovia, Azizis, Caput Bubali and Tibiscum into the three stations shewn, xv-xx, and is condemned (practically) to forego identification after Arcidava, surely a confession of failure.

^{page 80 note 1} von Domaszewski thinks this the wall of a permanent castellum, but the trees shown within the enceinte, and still more the marching camp, make this impossible. Nor are we to see here, with Cichorius, lines to block and hold the Apus valley. It is simply an extraordinary reinforcement of the camp defences, the significance of which later.

^{page 80 note 2} Cichorius' investigation of Varadia showed that the hill fortress lacked a water supply, and this he finds indicated here in the legionary in the foreground of XVI fetching water from the stream. The soldier, however, certainly belongs to the marching camp, not to the fort. C. employs this curious method of identification again at L in connexion with another fort which stands on an eminence at the head of a singular zigzag path. (This path points ahead to XCII and XCVII, where it reappears—the intention is to indicate the identity of the Long Walls, with which it is there connected, with those in L, and thus to signify that they lie north of the Danube, with the important corollary that the fighting in XCVIII takes place north, not south of the Danube.) The artist who handles freely the detail of sites so familiar as Ancona and Salonae was not likely to deviate into this Baedekerism in depicting these stations in the Western Carpathians. Topographical indication is a primary concern of the designer, but this is a thing of a very different order from topographical minutiae, although Cichorius only too often ignores this distinction.

^{page 80 note 3} A second bridge is shewn in xv but smaller, i.e. in the distance. We may suppose this scouting, party to cross the Czernovecz, tributary of the Karasu, and proceed along its left bank in the direction of Centum Putea (Nagy-Szurduk). The second bridge hints that this force later recrosses the Czernovecz, or merely foreshadows its return to its starting-point.

^{page 80 note 4} Cichorius' sketch-maps (11, pp. 71 and 85) and Kiepert's map (at end of 11) present some not unimportant discrepancies: (1) C's map at p. 71 (rightly) places Arcidava on right, Kiepert's on left bank of the Apus; (2) Caput Bubali. I would retain here the figure of Tab. Peut. (Azizis III, Caput Bubali) which von Domaszewski proposes to alter to XII, with the intention, I imagine, of making a further equidistant stage. In any case it is hard to see why Cichorius should place Caput Bubali so far west as between Ohabica and Ruzs. Kiepert goes less wide of the mark, it would seem, in putting it at Prebul, with the result, however, that C's route, drawn on Kiepert's map, exhibits a (now) unintelligible detour via Ruzs. We may perhaps draw the line of advance (provisionally) direct from Valemare to Zsuppa—Caput Bubali can hardly lie far outside it.

^{page 81 note 1} This circumstance does not pass unremarked by Petersen (i, 21), whose interest, however, in this series of pictures lies elsewhere, as already observed.

^{page 81 note 2} Geschichte der röm. Kaiser, ii, 174.

^{page 83 note 1} So, still more intently, at Azizis XX in the direction of Tibiscum. This, according to ‘Double Advance’ reasoning, there signifies that he momently expects the emergence before his eyes of the ‘Eastern army’ from the Teregova defile. Yet Trajan's attitude is the same in XII as in XX and he is (probably) flanked by the same two comites (Sura and Livianus). At XVIII the first captured Dacian is brought before him (again with his comites) to be examined, i.e. the artist this time, by a neat variation, gives a different expression to the preoccupation which throughout engages the emperor.

^{page 84 note 1} Some of these are certainly odd, e.g. the immobility of the mule. The barbarian clearly holds a club in right hand. The sculptor seems to have missed his ‘points’ here somehow, and by inverting right and left legs to have presented us with one of the (surely few) physically impossible ways of falling off an animal.

^{page 84 note 2} Beiträge zu einer kritischen Geschichte Traians, 83 Anm. 3.

^{page 85 note 1} Co-operation is not necessarily implied in Tac. Hist. iv, 54Google Scholar, ‘vulgato rumore a Sarmatis Dacisque Moesica ac Pannonica hiberna obsideri.’

^{page 86 note 1} A. von Premerstein, Das Attentat der Konsulare auf Hadrian, p. 7, accepts the statement of Jordanes: ‘Diese beiden stammverwandten Völkerschaften waren ehedem mit ihren äussersten Ausläufern im Süden Daciens am Flusse Aluta zusammengetroffen.’ I cannot suppress my doubts that the Iazyges at least at any time and in any manner extended to the Aluta.

^{page 86 note 2} Perhaps Stradonič, the well-known Celtic stronghold on the Beraun south-east of Prague (Pic, Die Urnengräber Böhmens, p. 16).

^{page 87 note 1} Romanisation of Roman Britain, 3rd ed. p. 75.

^{page 87 note 2} Cumont, Comment la Belgique fut romamsée, p. 40; also some, remarks of Haverfield in a review of Déchelette's Manuei d'Archéologie, etc. J.R.S: 1914, P. 232.

^{page 88 note 1} This refusal is probably shown on the Column at C, so that μετὰ ταῦτα must mean not ‘after the (first) war’ but simply ‘subsequently.’

^{page 88 note 2} Petersen takes τὰ ὲρύματα to refer to the Sarmizegethusa. See below.

^{page 89 note 1} Petersen i, 84 ff.

^{page 90 note 1} Roman Sculpture, p. 185.

^{page 91 note 1} This is hardly too strong a term for the abrupt division here. Its significance has commonly been missed by reason of its falling between the plates of Cichorius. It is worth while to inspect it in Fröhner.

^{page 91 note 2} So Groag, , O.J. v (1902), Beiblatt, p. 39Google Scholar.

^{page 96 note 1} See also that in Neigebaur, , Dacien aus den Uberresten des klassischen Alterthums (Kronstadt, 1851), pp. 91–93Google Scholar.

^{page 97 note 1} The discovery or at least the transportation of these is probably shown on the Column at CXXXVIII, but in a region remote from Sarmizegethusa.