² Agricola, 44, 1, ‘natus erat … Gaio Caesare tertium consule, idibus Iuniis.’

³ Ibid., 4, 1, ‘pater illi Iulius Graecinus … Marcum Silanum accusare iussus est et, quia abnuerat, interfectus est.’

⁴ Little is known of Graecinus, but that little is seldom fully recalled. He is mentioned by Seneca (de ben. ii, 21, 5) as refusing contributions towards his praetorian games from the wealthy but ill-famed patrician, Paullus Fabius Persicus, who was later the friend of Claudius (CIL xiii, 1668 = ILS 212, col. 2, 6, ‘nobilissimum virum amicum meum’), and from Caninius Rebilus, the learned and rich jurisconsult, whose immorality was notorious (Tac., Ann. xiii, 30Google Scholar, 3): and it is stated that the refusal was so couched as to leave no doubt concerning the reason: ‘utrum hoc munera accipere est an senatum legere ?’, comments Seneca (loc. cit.). The same blunt candour emerges in his comment on the peripatetic philosopher Ariston, recorded by Seneca (epist. iii, 8 (29), 6), ‘nescio quid de gradu faciat, tanquam de essedario interrogaretur.’ Finally, Columella mentions his two books on viticulture in the following terms (i, 1, 14): ‘cuius (Cornelii Celsi) velut discipulus duo volumina similium praeceptorum de vineis Iulius Graecinus composita facetius et eruditius posteritati tradenda curavit’: in other words, he improved on his model both in elegance and knowledge of the subject. It is noteworthy that Seneca does not give the same reason for his death as that offered by the family. He remarks (de ben. ii, 21, 5) merely ‘si exemplo magni animi opus est, utamur Graecini Iuli, viri egregii, quem C. Caesar occidit ob hoc unum, quod melior vir erat quam esse quemquam tyranno expedit’. The picture is that of an upright and knowledgeable man of candour, devoted to viticulture and with a predilection for chariot racing: both interests, in good wine and good horses, were probably popular in the Fréjus district, his homeland.

⁵ The point is due to Hirschfeld, Wiener Studien v, 120.

⁶ Agricola himself is described (Agr. 4, 1) as ‘vetere et inlustri Foroiuliensium colonia ortus’; while his mother's estates are mentioned as situated in the district of Ventimiglia (Agr. 7, 1).

⁷ See above, note 4.

⁸ Agr. 4, 2.

⁹ Ibid. 7, 2: ‘matrem Agricolae in praediis suis interfecit, praediaque ipsa et magnam patrimonii partem diripuit quae causa caedis fuerat.’ It may also be presumed that there was, or had been, property at Forum Iulii, as part of the paternal estate.

¹⁰ Ibid. 4, 4.

¹¹ Ibid. 6, 1. For her father, Domitius Decidius, see CIL vi, 403 = ILS 966, which shows him to have been a quaestor aerarii chosen by Claudius. His assignation to the Voltinian tribe suggests that he was probably a Narbonese, as Dessau (PIR ii, 21, no. 124) observes. For Claudius and his Gallic policy see Momigliano, , L'opera dell' imperatore Claudio, Firenze, 1932, 89–90Google Scholar, with n. 1 (Claudius: the Emperor and his Achievement, Oxford, 1934, 44–5Google Scholar, with n. 10).

¹² Agr. 9, 7.

¹³ Ibid. 6, 1.

¹⁴ Ibid 5 2

¹⁵ Ibid. 5, 4.

¹⁶ Suet., Nero 18.

¹⁷ Cf. Agr. 17, 1: ‘ubi cum cetero orbe Vespasianus et Britanniam recuperavit, magni duces, egregii exercitus, minuta hostium spes.’

¹⁸ For the principle, cf. Tac., Ann. i, 49Google Scholar, 5 f.: for the soldier's reaction after civil war, note the passage ‘truces etiam turn animos cupido involat eundi in hostem, piaculum furoris; nee aliter posse placari commilitonum manes, quam si pectoribus impiis honesta vulnera accepissent’—an acute psychological analysis.

¹⁹ Suet., , Div. Vesp. 9, 20Google Scholar: ‘amplissimos ordines … exhaustos caede varia.’

²⁰ Parker, , The Roman Legions, Oxford, 1928, 145–7Google Scholar: cf. Braithwaite, , C. Suetoni Tranquilli Divus Vespasianus, Oxford, 1927, 42–3Google Scholar.

²¹ Miller, , JRS xv, 183–5Google Scholar, defining the occupation as one of two stages, semi-permanent and permanent.

²² According to Ptolemy, , Geogr. ii, 3Google Scholar, 16–17, the Brigantes lay ‘below’ (i.e. south of) ‘the Selgovae and the Otadini and extend between both seas’. Five of the places mentioned by Ptolemy are certainly identified, namely, Vinovium (Binchester), Cataractonium (Catterick), Isurium (Aldborough), Olicana (Ilkley) and Eburacum (York). There remain Epeiacum, Calatum, Rigodunum and Camulodunum. Of these Camulodunum, a native fortress (dunum) in south-west Yorkshire, is presumably Almondbury, the outstanding native fortress in that area, just as Rigodunum in north Lancashire is presumably Ingleborough. Calatum and Epeiacum are probably Roman forts, and it is tempting to connect Calatum with Galacum of the Antonine Itinerary, either Overborough or Lancaster (Haverfield, , Arch. Journ. lxxii, 81Google Scholar). Epeiacum has been identified with Chesterle-Street, but the name of that fort, anciently Cuncaceaster, is more probably Concangium. Since the Selgovae were the south Scottish tribe extending to the Solway, while the Otadini, or Votadini, included Northumberland and presumably extended to its natural southern boundary, the Tyne, or perhaps later to its artificial boundary, the Wall of Hadrian, the inference is that the Brigantes included Durham, Cumberland, Westmorland, Lancashire, and the West and North Ridings of Yorkshire. It is not surprising that they could be styled (Agr. 17, 2) ‘civitatem, quae numerosissima provinciae totius perhibetur’. Subdivisions of the tribe at present known are the Setantii (Ptol., , Geogr. ii, 3Google Scholar; 2), of the Fylde, and the Textoverdi, of South Tynedale (Arch. Ael. Ser. 4, xi, 138 = EE ix, p. 593). The East Riding of Yorkshire belonged to the Parisi.

²³ Agr. 17, 2: ‘multa proelia et aliquando non incruenta.’

²⁴ Ibid. 8, 2: ‘primo Cerialis labores modo et discrimina, mox et gloriam communicabat.’

²⁵ CIL vii, 1204; EE vii, 1121: cf. Whittick, , Trans. Newcomen Soc. 1932, xii, 13–14Google Scholar, and Atkinson, D. in Flints. Hist. Soc. Trans. x, 1924, 11 fGoogle Scholar. It should be noted that the pigs of A.D. 74 were found at Chester, a point not without significance for the date of foundation for that fortress: see Chester Arch. Journ. vii, 85–6, nos. 196–7.

²⁶ The indirect evidence is said to come from tombstones without cognomina, but in fact there is only one of these (EE vii, 903 = Haverfield, Catalogue of Roman Inscribed Stones in the Grosvenor Museum 42, no. 54), whose reliability Ritterling questions (P-W, s.v. ‘legio’, 1772–3) on the analogy of such stones as CIL viii, 210, where the cognomen precedes the tribal name. For pottery, see Pryce, Davies, Ant. Journ. xviii, 41 fGoogle Scholar. On the other hand it is difficult to see how the attack on Anglesey by Suetonius Paulinus in A.D. 61 (Tac., Ann. xiv, 29Google Scholar, 30), when infantry was transported in ships, could have been carried out without a base at Chester; it must be remembered that any occupation must have been slight and easily removed from the rocky surface, and the base may even have been on a site other than that of the later fortress, or possibly, though improbably, on the Mersey.

²⁷ EE ix, 1039: as illustrated by Haverfield, Catalogue of Roman Inscribed Stones in the Grosvenor Museum 127, no. 199.

²⁸ Vegetius, , de re mil. iii, 6Google Scholar, refers to maps as follows ‘ut sollertiores duces itineraria provinciarum in quibus necessitas gerebatur non tantum adnotata sed etiam picta habuisse firmentur, ut non solum consilio mentis verum adspectu oculorum viam profecturus eligeret.’

²⁹ The Roman road-system in the military districts of Britain is wholly based upon the two conceptions of penetration and of insulation, as a glance at any map of the system will show.

³⁰ PBSR xiii, 30–1.

³¹ See Morris, The Welsh Wars of King Edward the First 294, on Edward's tactics and dependence upon the sea, together with some very suggestive remarks upon the difference between the Welsh and Scottish problems.

³² While one tombstone of a veteran (EE ix, 1051) is known from Chester, ten certain examples of serving soldiers of Leg. II Adiutrix occur (EE ix, 1047–1050, 1052–7), and there are six more which are rendered probable examples by the birthplace of the soldiers mentioned upon them (EE vii, 885, 892, 908; ix, 1087, 1089, 1095). On the other hand, only two come from Lincoln (CIL vii, 185, 186).

³³ Filow's date of A.D. 88–9 (Klio vi, 40 sqq.) is considered too late by Ritterling, P-W xii, 2, 1443, in view of the requirements of the military situation. He would put the recall in A.D. 85–6 or 86 at latest, that is, immediately after Agricola's recall, as is administratively not illogical.

³⁴ Agr. 20, 2: ‘loca castris ipse capere, aestuaria ac silvas ipse praetemptare.’ For the character of the country see Ekwall, Place-names of Lancashire p. 225, on West Derby and Newton hundreds; also p. 16, on the elements rum, rod and -ley. The Forests of Rossendale, Pendle, and Bowland are well known.

³⁵ Ibid. ‘nihil interim apud hostes quietum pati, quo minus subitis excursibus popularetur; atque ubi satis terruerat, parcendo rursus invitamenta pacis ostentare.’

³⁶ ‘Cichorius, Reliefs der Traianssäule, Taff. 18–19, sc. xxiv (heads in hands and mouth); Taf. 40, sc. lvi (heads on spears); Taf. 51, sc. lxxii (heads in hands).

³⁷ Germania Romana edn. 2, i, Taf. ix, 3.

³⁸ Agr. 20, 3: ‘ut nulla ante Britanniae nova pars pariter inlacessita transient.’ This refers not to the harried districts, but to ‘multae civitates quae in ilium diem ex aequo egerant’. There is thus no need to assume, with Anderson, that it means so little harried by their hostile British neighbours. It means that the Romans took over the districts in working order, without previous campaigning in them, and it may refer to the pagi of the north-east, mostly in county Durham. Certainly the civitates can hardly here be whole tribes, for otherwise the action must have extended too far north to fit the events of the following year.

³⁹ Collingwood, , CW² xxxiii, 192Google Scholar, 324–5, and RCHM Westmorland p. xxxvii, map: also Raistrick, , YAJ xxxiv, 146Google Scholar, fig. viii, for a distribution map of the Craven district.

⁴⁰ Cichorius, op. cit. Taf. 104–7, sc. cxlii–cxlvi, for an extended illustration of such rounding-up operations by cavalry: Taf. 50, 51, sc. lxxi (legionaries storming a native fortress).

⁴¹ The very existence of these hill-forts is little known, but reference may be made for Ingleborough to Phillips, Rivers, Mountains and Seacoast of Yorkshire pl. iii, and for Carrock Fell to Collingwood, CW² xxxviii, 32–41Google Scholar.

⁴² The control of movement was the most vital question, especially among semi-nomadic pastoral tribes, as anyone knowing Highland history will appreciate.

⁴³ Agr. 22, 1: ‘tertius expeditionum annus novas gentes aperuit.’

⁴⁴ Ibid.: ‘conflictatum saevis tempestatibus exercitum.’ Although Tacitus describes the British climate as dirty (ibid. 12, 3: ‘caelum crebris imbribus ac nebulis foedum’), this is the only passage in which he blames the weather for difficulties in operations.

⁴⁵ With reference to the parenthetic clause, aestuario nomen est, it should be noted that Fletcher, , CQ xxxvii, 90Google Scholar, commenting upon Ann. xi, 30, 1, adduces two further instances of the use of a dative in a parenthetic clause of the kind (Ann. xii, 51, 5, and Hist. ii, 4, 2). The objections of Gaheis and Anderson thus fall to the ground, and the clause may be accepted as original.

⁴⁶ Anderson, in Furneaux-Anderson, Agricola p. 106, states categorically that ‘Agricola cannot possibly have got so far thus early in his campaigns’. But no grounds are adduced for the assertion, and the later Saxon or medieval experiences with large armies do not suggest that the distance was in any way difficult to attain.

⁴⁷ The best known large site is Mumrills: see Macdonald, Roman Wall in Scotland edn. 2, 196, fig. 10. There was, however, also an early site of large extent at Cadder (op. cit. 304, figs. 42 and 311) covering some twenty acres, but this is not well understood. At Castlecary the amount of pottery recovered is sufficient to suggest an extensive site (op. cit. 250–2).

⁴⁸ Croy Hill, op. cit. 268, fig. 34. This site gives the clue to the interpretation of the ditch system at Bar Hill, where the ditches extending westwards but not traced beyond the Antonine west ditches, must belong to an annexe like the south annexe at Croy: see op. cit. 274, fig. 36 (relation of the two forts), and fig. 35, p. 272.

⁴⁹ Cichorius, op. cit. Taf. 14, sc. xvi, for a palisaded annexe with beacons inside it; Taf. 36, sc. xlviii, xlix, for much more elaborate structures.

⁵⁰ The gradients on the Annandale route from Solway to Inveresk involve a minor climb from the Solway to Birrenswark (665 feet), Erickstane Moss (1,286) and Melbourne (1,080): on the east route there is Gamelspath (1,674), Black Halls (1,464), Harker's Hill (1,090), and Soutra (1,350).

⁵¹ According to Tacitus, Agr. 24, 1, the more was made ‘in spem magis quam ob fortnidinem’, that is to say, Agricola himself did not fear hostilities from the Irish, but hoped for an opportunity to conquer the island.

⁵² Agr. 24, 3: ‘saepe ex eo audivi legione una et modicis auxiliis debellari obtinerique Hiberniam posse.’ This was probably a minimum estimate, curiously reminiscent of Strabo's estimate for the conquest of Britain (Geogr. iv, 5, 3, 200 C) τοὐλάχιστον μέν γάρ ένὁς τάγματος χρῂӡοι ᾶν καἱ ίππικού τινος. Actually, as in the case of Britain, the force required would probably have been larger.

⁵³ The earldom of Strathmore has always been the key to the possession of Scotland: hence the early concentration of power in Fife, which was not easily raided, but within reach of Lowlands and Highlands alike.

⁵⁴ Agr. 30, 4: ‘terrarum ac libertatis extremos recessus’; id. 33, 4: ‘e latebris suis extrusi’; id. 10, 7: ‘nusquam latius dominari mare … nec litore tenus adcrescere aut resorberi, sed influere penitus atque ambire et iugis etiam ac montibus inseri velut in suo.’ The last is a most vivid description of the characteristics of the west coast, obviously by an eye-witness, to whom such phenomena would be utterly unfamiliar elsewhere in the Roman world.

⁵⁵ The identity of the ‘civitates trans Bodotriam sitas’ is not known, and it would seem as if Tacitus is using civitas, as in c. 20, 3 (see n. 38 above), of smaller units than the main tribe. Ptolemy's account of the Scottish tribes (Geogr. ii, 3, 1–16) divides the Lowlands between four tribes only—the Votadini of Northumberland and the Merse, the Selgovae, whose land stretched from Roxburgh to the Solway, the Novantae, occupying the Mull of Galloway and the Ayrshire coast, and the Dumnonii, whose territory not only embraced Clydesdale and the Isthmus, but extended some distance north (pl. i). The Epidii inhabited Mull and Kintyre, while the Caledonii possessed the Great Glen, with the Caledonian forest beyond it. Between the Caledonii and the Isthmus the Vacomagi are placed in Moray and the Taexali in Mar and Buchan, while the Venicones occupy the Mearns and Strathmore. It will be noted that this allows for the presence of Caledonians in the Tay Basin, as is suggested by place-names (Watson, Celtic Place-names of Scotland 21–22). But there are other names which Ptolemy does not include, for example, the Boresti of Agr. 38, 3, and the Maeatae of Dio lxxvii (lxxvi), 12, 1 ff., the latter being connected with place-names in Clackmannanshire and Stirlingshire (Watson, op. cit. 59 f.), though Dio is careful to note that they are a confederacy: δύο δἑ γένη τών Βρεττανών μέγιστά είσι Καληδόνιοι καἱ Μαιάται͵ καἱ ές αὑτά καἱ τά τών άλλων προσρήματα ώς εἰπεῑν συγκεχώρηκεν. For tribes of the extreme north and west, see n. 67, below.

⁵⁶ Agr. 25, 4: ‘ne superante numero et peritia locorum circumiretur’; cf. Ann. ii, 5, 3: ‘longum impedimentorum agmen opportunum ad insidias, defensantibus iniquum.’

⁵⁷ Agr. 25, 4.

⁵⁸ As the sole allusion to a fort in the territory beyond the Forth, Agr. 25, 3, is worth note.

⁵⁹ Agr. 25, 1: ‘cum simul terra, simul mari bellum impelleretur ac saepe isdem castris pedes equesque et nauticus miles mixti copiis.’ The operative word is saepe, and the whole passage gives the impression of a steady drive forward, not confined to a short distance.

⁶⁰ Harbours, apart from the Tay estuary, are Arbroath, Montrose, Stonehaven, Aberdeen and possibly Peterhead.

⁶¹ The road is well known from Ardoch northwards to Strathearn, but the course between Stirling and Ardoch has given difficulty, and does not appear to lie on the modern line.

⁶² PSAS xxxvi, 182–242, and JRS ix, 113–14, give particulars of the site; and to these should be added PSAS lii, 233–4, for a coin of A.D. 86, and PSAS lxxiv, 46, fig. 3 (JRS xxxiii, 159, fig. 10) for the relation of the site to Strathmore and the Cleaven Dyke.

⁶³ Dealgin Ross: Roy, Military Antiquities (1793), pl. xi, and JRS xxxiii, pl. iii, for plan. For comments on the site, Macdonald, , JRS ix, 135–6Google Scholar, also PSAS lviii, 326, for coins. Fendoch: Richmond, and McIntyre, , PSAS lxxiii, 110–154Google Scholar.

⁶⁴ There is a persistent tradition of a Roman site at or near Bochastle, west of Callender, see Statistical Account, xi, 607. A recent discovery there of a fragment of a quern of Niedermendig lava reinforces the notion, since such querns were Roman army imports and are unlikely to have travelled far from a permanent site: the object was communicated to me by the late Mr. A. J. H. Edwards, Director of the National Museum of Scotland, where it now is.

⁶⁵ I am now inclined to think that, while the buildings at Fendoch were timber-framed, many of them may have been of wattle and daub between the framing. This makes a durable building, warmer than wooden hutments and less extravagant in timber. The same solution is adopted by the latest reconstructors of Haltern. This conclusion does not invalidate the remarks made about standard sizes, or any of the conclusions drawn about removal of timbers, and it explains the need for sill construction.

⁶⁶ In Tac., Ann. xiv, 29Google Scholar, 3, the island is attacked as being ‘incolis validam et receptaculum perfugarum’. In Agr. 14, 4, Paulinus is described as ‘Monam insulam ut vires rebellibus ministrantem adgressus’. In Welsh tradition it is known as ‘Mon mam Cymru’, that is, ‘Mona, mother of Wales.’

⁶⁷ See Douglas Simpson, The Province of Mar 1–2: in later times the Pictish divisions involved were those of Moravia and Cat, the latter becoming the Norse earldom, later still divided into the medieval earldoms of Sutherland and Caithness. The Pictish capital seems frequently to have been in the seat of the Moravian high king, at or near Inverness (cf. Adamnan, , Vita S. Columbae ii, 33–5Google Scholar, of A.D. 565, and Bede, , Hist. Eccl. v, 21Google Scholar, 1, of the mission to Nechtan, A.D. 710). In Roman times the Vacomagi seem to have inhabited Moray and the Caledonii the Great Glen, now Inverness-shire. To the north lay the Decantae and the Lugi, the latter in Sutherland north of the Smertae, while Caithness was inhabited by the Cornavii (pl. i). The West Coast tribes from Loch Linnhe northwards were the Creones, the Carnonacae and the Caereni. Ptolemy has no information to give of the population centres of these tribes, and the peculiar type of habitation, which was certainly in use among the more northerly, namely, the broch, receives no place in Roman literature, though it was certainly open to Roman commerce (Curle, J., PSAS lxvi, 392–6Google Scholar): its architectural peculiarities are the sort of features which would have interested Roman authors.

⁶⁸ Agr. 29, 2–3: ‘praemissa classe, quae pluribus locis praedata magnum et incertum terrorem faceret… ad Montem Graupium pervenit, quem iam hostis insederat … nam Britanni … legationibus et foederibus omnium civitatium vires exciverant.’ This implies a wholesale muster of tribes waiting and prepared to give battle.

⁶⁹ On the site of Mons Graupius see Macdonald, , JRS ix, 115Google Scholar, and Collingwood, Roman Britain and the English Settlements edn. 2, 115. No indication is given by Tacitus that the Mons Graupius was a hill-fort. The action suggests a key-position by which Agricola must pass, and there are not many of these, the best known to history being the natural gateways between Mar and Moray.

⁷⁰ See the last note. Macdonald ruled out all sites south of Strathmore, JRS ix, 115.

⁷¹ Collingwood, loc. cit.

⁷² It should be remarked that few were fortunate in handling a fleet. Caesar's expeditions to Britain were almost ruined for lack of seamanship; those of Germanicus in Germany were disastrous; and the career of Paulinus was shipwrecked upon his handling of the British fleet.

⁷³ Agr. 35, 2: ‘ingens victoriae decus citra Romanum sanguinem bellandi.’

⁷⁴ Ibid. 35, 2; 29, 2: ‘ex Britannis fortissimos et longa pace exploratos; 36, 1.

⁷⁵ Ibid. 35, 2; ‘equitum tria milia’; and 37, 1: ‘quattuor equitum alas ad subita belli retentas.’

⁷⁶ Agr. 37, 6.

⁷⁷ Agr. 37, 2.

⁷⁸ See Ritterling in P-W xii, 1442 f.

⁷⁹ Suet., , Domitian 10, 3Google Scholar: ‘Sallustium Lucullum Britanniae legatum quod lanceas novae formae appellari Luculleas passus esset (occidit).’

⁸⁰ Arrian, De acie contra Alanos 1–9 : the force comprised cavalry consisting of one numerus, four alae and the mounted detachments from three milliary and seven quingenary cohorts, two legionary units and the governor's life-guards; infantry consisting of two legions, one milliary cohort, two quingenary cohorts and a detachment from a third, and two bodies of barbarian levies.

⁸¹ Cichorius, op. cit. Taf. 17–19, sc. xxiv; 47–8, sc. lxvi; 51–2, sc. Lxxii.

⁸² See n. 75 above.

⁸³ Parker, The Roman Legions 76–8, for the budgetary side of the matter. Vegetius, , de re mil. ii, 3Google Scholar, remarks of the legions, ‘magnus in illis labor est militandi, graviora arma, plura munera, severior disciplina.’

⁸⁴ Cf. Tac., Hist. iv, 74Google Scholar: ‘cetera in communi sita sunt … nihil separatum clausumve.’

⁸⁵ In Tac., Ann. xiv, 38Google Scholar, 4, Classicianus is decried as partisan (‘bonum publicum privatis simultatibus impediebat’), and as against Paulinus (‘novum legatum opperiendum esse, sine hostili ira et superbia victoris clementer deditis consulturum’). But the faults of Paulinus are admitted in Agr. 16, 2, ‘ex legato timor agitabat, ne, quamquam egregius cetera, adroganter in deditos et ut suae cuiusque iniuriae ultor durius consuleret.’ The next governor was gentler: ‘exorabilior et delictis hostium novus eoque paenitentiae mitior.’ For the tombstone of Classicianus see Antiq. Journ. xvi, Cottrill, 1–7, and Birley, 207–8, the latter noting that the secdnd cognomen of Iulia Pacata was probably Indiana; also RCHM, Roman London 83. Whether the Iulius Classicianus, of CIL vi, 33805Google Scholar, is the same person is doubtful.

⁸⁶ Tac., Ann. xiv, 39Google Scholar, notes that the result of the mission of Polyclitus went against Paulinus. If Agricola's thoughts are really reflected in the comments of Tacitus, it should be noted that it was the arrogance of Paulinus that gave offence.

⁸⁷ Tac., , Hist. iv, 74Google Scholar.

⁸⁸ Agr. 15, 19, 31, for British grievances.

⁸⁹ Agr. 12, 13, 27, 21, for striking features of British character; cf. Hist. iv, 76, on the Germans: ‘qui ab ipsis sperentur, non iuberi, non regi, sed cuncta ex libidine agere.’

⁹⁰ For the name Procilla, see SHA, Antoninus Pius I, 4Google Scholar, CIL vi, 9355 = ILS 7383, and Conway, and Johnson, , Prae-Italic Dialects of Italy i, 224Google Scholar (Histria), 283 (Venetia), 436 (Emilia) where Procillus is also known, 430. Her ancestral estates were apparently in Liguria (Agr. 7, 2).

⁹¹ Ant. Journ. xvi, 207–8.

⁹² Op. cit. vi, 208.

⁹³ Syme, The Roman Revolution 359–360.

⁹⁴ For the catholicity see Arrian, , Ars Tactica, 33, 2Google Scholar: εἱ γά ρ τοι έπʹ ἂλλω τῳ καἱ έπἱ τᾧδε ἃξιοι έπαινεῑςθαι Ῥωμαίοι, ὂτι ού τά οἱκεῑα και πάτρια, ούτως τι ήγάπησαν ώς τά πανταψόθεν καλἁ έπιλεξάμενοι οίκεῑα σφίσιν άποιήσαντο.

⁹⁵ For the double notion of civilisation and service, see Sherwin-White, The Roman Citizenship 181–193, on the policy of Claudius and its relation to military service, and, in particular, p. 192 : cf. Aristides, Είς Ῥώμην, 60: ξένος δʹ ούδεἱς άρχῆς ῆ πίστεως ἂξιος.

⁹⁶ Agr. 21, 1: ‘hortari privatim, adiuvare publice, ut templa, fora domos extruerent … iam vero principum filios liberalibus artibus erudire.’ It was no doubt for the latter purpose that Demetrius of Tarsus had been engaged : see Plutarch, de defectu oraculorum 18. If the York inscription to the gods of the legate's residency at York is by this Demetrius, as seems likely, we can picture some of the young boys in residence, half guests and half hostages. For the inscription, on bronze tablets, see CIL vii, p. 62 = EE iii, 312 = EE ix, 560 = CIG xiv, 2548.

⁹⁷ For other biographies, see Wight Duff, A Literary History of Rome in the Silver Age 269–270.

⁹⁸ Cf. Agr. 2, 1: ‘cum Aruleno Rustico Paetus Thrasea, Herennio Senecioni Priscus Helvidius laudati essent, capitale fuisse, neque in ipsos modo auctores sed in libros quoque eorum saevitum.’

⁹⁹ Agr. 46, 1.

¹⁰⁰ See Wight Duff, op. cit. 596, quoting Racine's estimate of Tacitus as ‘le plus grand peintre de l'antiquité’

¹⁰¹ Plotinus, , Enneades vi, 7, 22Google Scholar.