² In Appian the main operations end before the burning of the Capitol, 6 July, 83 (B.C., i, 391). For the hard winter, ibid., 394.

³ Appian, , B.C., i, 397–400Google Scholar; Plut., , Sull., 28, 7 f.Google Scholar; Veil., ii, 26, 1; Liv., Per., 87; Oros., v. 20, 6; Floras, ii, 9, 24; Eutrop., v. 8, 1; Diod. xxxviii/ix, 15; Fenestella, Fr. 16 P; Vir. Ill. 68, 3. I adopt the general view on the site of this battle of Sacriportus—not too far from Segni (Signia), as in Plutarch, rejecting Appian's notion of Sullan advance through Setia.

⁴ App., , B.C., i, 294Google Scholar (help to Cinna, 87 B.C.); Marius' preparations there in Veil., ii, 26, 1; Plin., , N.H., xxxiii, 16Google Scholar; Val., Max., vii, 6, 4Google Scholar.

⁵ App., B.C., i, 402Google Scholar—‘Sulla, having shut Marius in Praeneste, set about cutting it off with ditches and palisades sited a long way back from the town, and put Lucretius Ofella in charge of the task, relying no longer on open battle to bring Marius to terms, but on starving him out.’ On Ofella, , R.E., xiii, 2Google Scholar, col. 1686 f., no. 25.

⁶ App., , B.C., i, 405–408Google Scholar. The ‘massacre’ by L. Damasippus, on Marius' orders, was presumably a last desperate attempt to prevent losing Rome by treachery. Sulla left the city held by a modest defence force—probably veterans due for discharge (App., , B.C., i, 408Google Scholar; 430; cf. 427 and Gabba ad locc; Plut., , Sull., 29, 4–5Google Scholar; Sall., , Hist., i, 40MGoogle Scholar).

⁷ App., , B.C., i, 395f.Google Scholar; 401; 410 (with Gabba's commentary); Veil., ii, 28, 1; Oros., v, 20, 5, 7; Plut., , Sull., 28, 8Google Scholar; Pomp., 8, 4–6; Crass., 6.

⁸ App., , B.C., i, 414 f.Google Scholar, with Gabba.

⁹ Ibid., 416—‘from Lucania’; ‘from Samnium’—though there were Samnites and Lucanians already serving with the government forces further north, recruited, it seems in 84/3 (App., , B.C., i, 400; 420Google Scholar). The figure of 70,000 is certainly exaggerated.

¹⁰ App., , B.C., i, 416Google Scholar—‘M. Lamponius from Lucania, Pontius Telesinus from Samnium and Gutta the Campanian pressed on with 70,000 men to extricate Marius from the siege but in τὰ στϵνὰ, the only route whereby penetration was possible, Sulla maintained a position which shut them off from their way past.’ For the losses, ibid., 425—‘despite the Samnites’ spirited willingness to sustain heavy losses on his behalf (Carbo's) in fighting en masse around τὰ στϵνὰ.’

¹¹ Ibid., 417. Note that Appian clearly distinguishes between operations against Sulla and those against Ofella.

¹² Ibid., 423—‘Carbo sent Damasippus with two legions … but not even these could pass Sulla's guard over τὰ στϵνὰ’.

¹³ Ibid., 427—‘Carrinas, Marcius and Damasippus with all forces at their disposal advanced on τὰ στϵνὰ intending to join the Samnites in an all-out effort at forcing through. Even so they failed, but then made for Rome, hoping that together they would catch the city empty of men and supplies—and set about making camp near the ager Albanus,’—that is, as commentators agree, in the triangle Bovillae—Marino—Castelgandolfo.

¹⁴ Sources in M.R.R., ii, 69.

¹⁵ In Journal of Philology, xxxv (1920), 1–18Google Scholar (repeated in C.A.H., x, 275 f.Google Scholar), to which I am most indebted.

¹⁶ Besides, even if a force could have been evacuated in some sort of operational cohesion, Crassus' activities among the Marsi (Plut., Cr., 6) may already have secured the Sullan cause, and the failure of any breakout, in that direction—even if Crassus was no longer there (App., , B.C., i, 413Google Scholar).

¹⁷ On cuniculi around Praeneste, H. Fernique, Étude sur Préneste, 123, and see now Forma Italiae: Praeneste, 37 f.; compare the story of Marius' death while trying to escape from the city through one of them—App., , B.C., i, 434Google Scholar; Strabo, v, 3, 11; Liv., Per., 88; Plut., , Mar., 46, 9Google Scholar; Sull., 32. 1; Vell., ii, 27, 4–5; Oros., v, 21, 9; Val., Max., vi, 8, 2Google Scholar; Vir. Ill., 68, 4.

¹⁸ Gardner, op. cit., 14, n. 2, suggesting Livy as the source. Livy was notoriously inaccurate in matters of topography and military science, but not to the extent, I submit, of using incorrect language to describe the facts he found in his sources.

¹⁹ App., , B.C., i, 414Google Scholar.

²⁰ Gardner does consider the features which I think to be those more probably meant, but rejects them—on insufficient grounds, in my view.

²¹ Pompeius (and perhaps Crassus too) moved across from Umbria to prevent Carbo's pursuit (App., , B.C., i, 426Google Scholar; cf. 413–414; Vell., ii, 28, 1), but thus left open a path for the forces of Damasippus, and later Carrinas and Censorinus too, down the Via Flaminia from Ariminum, or thereabouts.

²² When there is no reason to suppose that he feared attack on the besieging lines from either north or south. The Samnite insurrection seems to start later, unexpectedly.

²³ As opposed to δίοδος, ‘way through’ (L.S.J., s.v. πάροδος).

²⁴ And see above, p. 34 and n. 16 for the hills behind the town.

²⁵ Alternatively, the Via Latina could be closed by a small independent force at the Algidus Pass itself Pozzi, E., Att. dell' Acc. d. Scienze d. Torino, xlix (1913–1914), 675 f.Google Scholar, abandoning literal interpretation of τὰ στϵνὰ as an inexact translation of a Latin source, places Sulla's lines between Artena and the Colle Pratarolo, centred on Valmontone—too far east, for in this case the Oscans could not have reached the ager Albanus through Artena and Velitrae. However, that they did so by night shows that Sulla's lines (and so. a possible flank attack) were not too far away, possibly near Macere (cf. n. 28 below). Pozzi leaves unexplained Sulla's success at the στϵνὰ against attacks from the north (on which see below).

²⁶ Gardner, op. cit., 13–15, where he will allow tactical importance to this area only if it was more thickly wooded than it is today. I disagree, for reasons which follow. In fact, on parts of the ridges today the amount of wood and undergrowth is considerable, but unfortunately no safe index of conditions in 82 B.C., for which I can find no evidence. If (as cannot be proven) at that time the Italian native scrub or macchia persisted, not having yet given way to cultivation, routes by which it could be penetrated in military formation would be few and difficult, easily controlled by comparatively few defenders. Cf. Nissen, H., Landeskunde, i, 431Google Scholar and references there. Mommsen, Th., R.G. iv, 327 f.Google Scholar, was a little closer in placing Sulla across the Valmontone—Palestrina road. Similarly Strachan-Davidson, J. L.'s Commentary on Appian B.C., i, 94Google Scholar, favours a site either on the heights controlling ground between Gave and Valmontone or (it is not clear) at Valmontone itself. Both assume an attempt to break through the blockade directly into Praeneste, not past it, neither can explain the στϵνὰ in which relief from the north was stopped, and Mommsen adds some unacceptable suggestions of his own.

²⁷ App., , B.C., i, 425Google Scholar.

²⁸ Ibid., 427; Plut., Sull., 29, 3. It is this circumstance, rather than distance from Praeneste, which precludes the ravine at S. Germano (between Ferentinum and Anagnia, mentioned by Gardner and Strachan-Davidson), from which an alternative route to the ager Albanus through the Monti Lepini is out of the question.

²⁹ ‘Rivulets’ in Gardner (p. 8). In his view ‘they are easily turned, and could not seriously interfere with the advance of an army from the N.W.’ (p. 13, n. 1). The Marians, however, could not risk getting too close to Rome, for that would expose them (for all they knew) to flank attack (and perhaps indeed the veterans and young nobility guarding the city might be fairly formidable with such an advantage). Thus they would be forced into these defiles, the existence of which Gardner admits (p. 14). By the time of the final attack, Marian forces were lower in numbers—probably four legions or less in the northern force, two under Damasippus (App., , B.C., i, 423Google Scholar) (less losses sustained) and one each under Carrinas and Censorinus (cf. ibid., 425—‘two legions belonging to Damasippus, and others under Carrinas and Marcius’).

³⁰ At what point in the story M. Crassus joined Sulla for the battle of the Porta Collina is unknown.