Britannia 54 (2023), 275–303 doi:10.1017/S0068113X23000193 The ‘Saxon Shore’ Reconsidered* By JOHN F. DRINKWATER ABSTRACT I propose that the usual role of the Notitia Dignitatum’s ‘Saxon Shore’ forts was, on both sides of the Channel, to control chronic, ‘everyday’ piracy and to support imperial operations. An exception occurred under Carausius and Allectus when the British forts were augmented to face likely Roman invasion. There was never any integrated cross-Channel system against concerted barbarian seaborne attack, Saxon or otherwise. The ‘Saxon Shore’ was a late fourth-century political expedient, confined to Britain and with minor military significance. Keywords: later Roman Britain; later Roman Gaul; Saxon Shore; Saxons; Carausius; Allectus; Constantius I; Stilicho; naval warfare THE ‘SAXON SHORE’ INTRODUCTION T he phrase ‘Saxon Shore’ comes from the Notitia Dignitatum (ND), a catalogue of senior Roman imperial positions compiled around 400. The ND lists military commands with the names and bases of the units they comprise. It associates ‘Saxon Shore’ with two groups of bases, usually referred to as ‘forts’, located along the south-eastern and southern coasts of Britain and the Channel and Atlantic coasts of Gaul respectively.1 THE BRITISH FORTS In Britain these are shown under the command of a comes litoris Saxonici per Britannias or ‘count of the Saxon Shore in Britain’, as distinct from the comes Britanniae or ‘count of Britain’. Both counts were directly responsible to the most senior general in the West, the magister peditum praesentalis occidentalis. Additional forts, not listed in the ND, are at Caister-on-Sea (known from archaeology) and Walton Castle (known from antiquarian descriptions). Additional sites have been suggested for Felixstowe and Rye, but have not been proved, and at Bitterne and *All dates are A.D. I am extremely grateful to Hugh Elton, Phil Freeman, Nick Henck, Werner Lütkenhaus and the two anonymous readers of Britannia for their invaluable comments and corrections. 1 Not. Dign. (occ.) 28/26 (Seeck 1876/Faleiro 2005), 37, 38. Jones 1964, 3.347–80; Brulet 2015, 594. © The Author(s), 2023. Published by Cambridge University Press on behalf of The Society for the Promotion of Roman Studies. This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 276 JOHN F. DRINKWATER Carisbrooke, but now seem unlikely.2 In TABLE 1 I list the remainder and, where known, their ND garrisons in order of location, clockwise east to west. TABLE 1. LITUS SAXONICUM PER BRITANNIAS ND Location [in locative case, as unit bases] Mapped as References: [Seeck 1876/Faleiro 2005] References: Fu. = Fuentes 1991 Pears. = Pearson 2002 Dh.= Dhaeze 2019 [ ] = alternative proposals Unit Branoduno [28.16/26.13] Brancaster Pears. 11–15; Dh. 142–3; [Fu. 59: ?Rye] equites Dalmatae Branodunenses Garianno [28.17/26.14] Burgh Castle Pears. 17–19; [Fu. 59: ?Portchester; Pears. 130/Dh. 142–3: or Caister?] equites stablesiani Gariannonenses Othonae [28.13/26.15] Bradwell Pears. 22–4; Dh. 142–3; [Fu. 59: Bitterne] numerus Fortensium Regulbio [28.18/26.16] Reculver Pears. 24–5; Dh. 142–3 cohors I Baetasiorum Rutupis [28.19/26.17] Richborough Pears. 26–9; Dh. 142–3 legio II Augusta Dubris [28.14/26.18] Dover Pears. 29–31; Dh. 142–3 milites Tungrecani Lemannis [28.15/26.19] Lympne Pears. 31–4; Dh. 142–3 numerus Turnacensium Anderidos [28.20/26.20] Pevensey Pears. 34–6; Dh. 142–3 numerus Abulcorum Portus Adurni [28.21/26.21] Portchester Pears. 36–8; [Fu. 60: Bradwell; Pears. 130/Dh. 142–3: or Walton Castle?] numerus exploratorum THE GALLIC FORTS In Gaul, the ND allots thirteen forts to two duces. The dux Belgicae II has three; the dux tractus Armoricani et Nervicani, whose command extends over five provinces (Aquitania I and II, Lugdunensis II, III and IV), has ten. Both are responsible to the magister equitum per Gallias who in turn answers to the magister peditum praesentalis occidentalis. There is no comes litoris Saxonici per Gallias, but two forts, one in each of the two jurisdictions (Marcis and Grannona), are labelled in litore Saxonico, ‘on the Saxon Shore’. Although the dux Belgicae II appears to have had only two forts and a fleet base, scholars usually assume that the archaeologically attested forts at Oudenburg and Boulogne were also under his command.3 In TABLES 2 and 3 I list known forts and garrisons in order of location, anti-clockwise north to south. 2 3 Fuentes 1991, 59; Dhaeze 2019, 280, 295, 299–300. cf. Dhaeze 2019, 142–3. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 277 THE ‘SAXON SHORE’ RECONSIDERED TABLE 2. BELGICA SECUNDA ND Location [in locative case, as unit bases] Mapped as References: [Seeck 1876/Faleiro 2005] References: Johns. = Johnson 1976 Fal. = Faleiro 2005 Brul. = Brulet 2015 Dh. = Dhaeze 2019 [ ] = alternative proposals Unit Portu Epatiaci [38.9/9] Boekhoute Fal. 670; Brul. 596; [Johns. 92, 140: Oudenburg?; Dh. 142–3: uncertain] milites Nervii Marcis, in litore Saxonico [38.7/7] Marck, near Calais Johns. 90; Brul. 596; [Fal. 666: Marques nr. Dieppe/Marque nr. Courtrai/Famars?; Dh. 142–3: uncertain – Marquise?] equites Dalmatae Classis Sambrica, in loco Quartensi siue Hornensi [38.8/8] Étaples Johns. 90–2; Dh. 121, 142 (perhaps with Saint Valery/Le Crotoy/Cap Hornu); [Fal. 671: Quartes, nr Tournai?; Brul. 596: uncertain] TABLE 3. TRACTUS ARMORICANUS ET NERVICANUS ND Location [in locative case, as unit bases] Mapped as References: [Seeck 1876/Faleiro 2005] References: Johns. = Johnson 1976; Fal. = Faleiro 2005; Brul. = Brulet 2015; Dh. = Dhaeze 2019 [ ] = alternative proposals Unit Rotomago [37.21/22] Rouen Johns.74; Fal. 673; Brul. 596; Dh. 142–3 milites Ursarienses Grannono [37.23/23] Le Havre Johns. 92–3; [Fal. 653: Granville; Brul. 596: disputed; Dh. 142–3: Cherbourg/Le Havre/ Guérande?] milites Grannonenses Grannona, in litore Saxonico [37.14/21] Guernsey Fal. 653; [Brul. 596: disputed; Dh. 142–3: Cherbourg/Le Havre/Guérande?] cohors I nova Armoricana Constantia [37.20/20] Coutances Johns. 74; Fal. 646; Brul. 596; Dh. 142–3 milites I Flaviae Abrincatis [37.22/19] Avranches Johns. 74; Fal. 629; Brul. 596; Dh. 142–3 milites Dalmatae Aleto [37.19/18] Aleth Johns. 74; Brul. 596; Dh. 142–3; [Fal. 653: St.-Servan, nr St. Malo] milites Martenses Osismis [37.17/17] Brest Johns. 75; Brul. 596; Dh. 142–3; [Fal. 666: Liannilis, nr Brest?] milites Osismiaci Benetis [37.16/16] Vannes Johns. 74; Fal. 684; Brul. 596; Dh. 142–3 milites Mauri Beneti Mannatias [37.18/15] Nantes Johns. 74; Fal. 660; Brul. 596; Dh. 142–3 milites superventi Blabia [37.15/14] Blaye Johns. 76; Brul. 596; Dh. 142–3 [Fal. 640: Port Louis?] milites Carronenses https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 278 JOHN F. DRINKWATER The consequent distribution of forts is shown in FIG. 1. Fig. 1. British and Gallic shore forts and walled towns. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 279 THE NOTITIA DIGNITATUM The ND is difficult to use. Its initial purpose and ultimate function are contested and it is clear that it is not homogeneous. Originally compiled around 390, its eastern section was kept up to date to around 395 while its western underwent revision to around 420. This revision was, however, unsystematic, and none of its entries can be related to any specific year, which means that the document gives only a rough idea of western imperial resources around 400.4 Furthermore, as TABLES 1–3 indicate, the linking of its place-names to modern sites is open to debate. More positively, it is also clear that the western ND was based on information collected over many decades. This is demonstrated by archaeological study of its British forts which, while confirming its main list, has shown that some were founded as early as the third century, some had gone out of service by the time the ND was compiled, and others existed of which it makes no mention. This, together with other signs of earlier arrangements,5 permits the conclusion that well before 390 something came into existence on the British and Gallic coasts, part of which was eventually called the litus Saxonicum. ‘Eventually’ because it appears that a specifically Saxon menace was a late phenomenon. From the mid-third to the mid-fourth century, it was the Franks who posed the main threat to the western provinces on land and sea.6 The Saxons, although by then settled in southern modern Schleswig-Holstein,7 do not figure prominently in the historical narrative of the period. Thus while, in the mid-fourth century, Julian classed Saxons with Franks as ‘the most warlike’ of the trans-Rhenish peoples, he campaigned only against the latter.8 The earliest reference we have to Saxons as distinctly dangerous maritime foes is in Eutropius’ condensed Roman history, written under Valens (364– 378).9 This does not mean that no Saxons raided Gaul before the later fourth century. ‘Frankish’ raiders were probably somewhat ethnically mixed and may well have included Saxons.10 However, it does suggest that it was not until then that Saxons alone caused sufficient trouble for the Romans to distinguish their attacks from those of the Franks. This appears to have begun when the Saxons mounted a great seaborne attack on northern Gaul in 370.11 For this reason, to avoid anachronism, I follow Pearson in terming the forts ‘Shore forts’, not ‘Saxon Shore forts’. THE SHORE FORTS C. 200–286 BRITAIN I The earliest extant Shore forts appear to be those of Brancaster, Caister-on-Sea and Reculver, long dated to the Severan period but now postulated as being late Antonine.12 As large (at c. 3 ha/7 acres each) new coastal and estuary installations, the obvious explanation for their construction 4 White 1961, 46–50; Pearson 2002, 129; Kulikowski 2000, 358–9, 360–1, 368–72, 375–6. cf. Jones 1964, 1.56; Kulikowski 2015, 695–6; Brennan 2015, 1034. Wood 1990, 94; Drinkwater 2007, 106. 7 Haywood 1991, 29 (map 2). 8 Julian., Or. 1.34D: ta maximotata. Bartholomew 1984, 169. Zosimus (3.6–7) says that the Kouadoi, suppressed by Julian after they had made a water-borne attack up the Rhine mouths, were Saxons, but his account is confused, and these were probably Frankish Chamavi. Cf. Julian., Ep. ad Ath. 280B; Amm. Marc. 17.8.5; Goetz et al. 2006–7, 1.290 and n. 300; contra Haywood 1991, 14, 42–3. 9 Eutr. 9.21. White 1961, 50; Bartholomew 1979, 369; Southern 2004, 396; Goetz et al. 2006–7, 1.115. Cf. below ‘Carausius and Allectus’. 10 cf. Haywood 1991, 37–8; Dhaeze 2019, 51, 56. 11 Amm. Marc. 26.4.5 (looking forward to 370), 28.5.1–7, 30.7. Bartholomew 1984, 174. 12 Pearson 2002, 14–15, 24; Philp 2005, 216; Dhaeze 2019, 21–2, 104–5, 283–4. 5 6 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 280 JOHN F. DRINKWATER is that they were intended ‘to curtail the activities of raiders and pirates’.13 Piracy, like brigandage, was a constant, albeit fluctuating, social phenomenon,14 and we should expect appropriate counter-measures in places vulnerable to depredation. In Britain, the earliest known anti-piratical installations, built probably against barbarian raiding across the Bristol Channel, are datable to the mid-first century.15 From the second to the fourth century, similar defences were built on the western and north-eastern coasts, probably against Picts and Scots.16 On the Continent, defensive structures were built along the Rhine from the first century and, after heavy Germanic sea-raiding, along the coasts of Lower Germany and Belgica from the late second century.17 Shore-defences were supported by the Roman navy. The best known of its provincial fleets is the Classis Britannica, established in the first century probably to assist the conquest of Britain and kept in service during the second and third centuries to support imperial personnel on the island and to police coastal shipping. Its main base was at Boulogne, with a subsidiary first at Richborough and then at Dover.18 The third-century ‘Crisis’ brought more attacks by neighbouring Germani, some of whom demonstrated a remarkable willingness to take to the sea. Under Gallienus (260–268), a group of Franks raided overland to Spain and then sailed to Africa.19 Under Claudius II (268–270), Goths caused havoc in the eastern Mediterranean.20 Under Probus (276–282), a group of Frankish mercenaries mutinied and went on a piratical jaunt from the Black Sea to Africa before sailing home via the Atlantic.21 It therefore used to be thought that the first British Shore forts were built against earlier Germanic raids across the North Sea, but it was then plausibly argued that their vernacular ship-building technology did not allow Germani to attack across the North Sea or even the eastern Channel until the later fourth century – the Mediterranean marauders had used local vessels.22 A recent attempt to restore the earlier date is undermined by the concession that the most sophisticated Germanic warship-type known for this period, the mid-fourth-century Nydam oak vessel, lacked a full keel and sails, making it ‘only suited for coastal navigation’. As a large open rowing-boat, it could have been deployed against Gaul and against Britain across the Straits of Dover, but probably not (it is impossible to prove a negative) much further afield.23 Chronic Germanic raiding of Gaul and Britain before the later third century is not, in fact, evident in the sources.24 An alternative explanation is that the new British forts were intended to meet an increase in home-grown Channel piracy caused by increasing social upheaval in the area even before the onset of the ‘Crisis’ proper.25 Against this, however, is that the location of Brancaster and Caister-on-Sea, far from the Channel, ‘makes no strategic sense’, and that there is evidence for a run-down in this period of the Channel fleet’s headquarters at Boulogne and Dover, and no sign of the existence of the 13 White 1961, 34–6; Johnson 1983, 209–10; Pearson 2002, 13, 16, 25; Philp 2005, 288; Symonds 2018, 54 (quotation); Dhaeze 2019, 105–6. 14 Symonds 2018, 22. 15 Symonds 2018, 36, 50–2; Dhaeze 2019, 131–2. 16 Johnson 1983, 208; Pearson 2002, 52, 65; Southern 2004, 404; Williams 2004, 17; Symonds 2018, 203–8; Dhaeze 2019, 123, 133–6. 17 Symonds 2018, 37–41; Dhaeze 2019, 89, 100–4, 212 (Aardenburg), 232 (Oudenburg). 18 Pearson 2002, 49, 50–2; Dhaeze 2019, 35–6. 19 Aur. Vict., Caes. 33.3; Eutr. 9.8; Oros. 7.22.8. 20 Zos. 1.32.1, 2–3; 1.33.1, 3; 1.34.1–2; 1.42.1; 1.43.1. 21 Pan. Lat. 8(4).18.3; cf. Zos. 1.71.2. 22 e.g. Drinkwater 1987, 211, and refs; see now Cotterill 1993, 227–8; Pearson 2002, 134–6; Williams 2004, 13. Cf. Dhaeze 2019, 66 on the weakness of the archaeological evidence for the second-century case; and Casey 1994, 158–9, on the failings of Saxon vessels. 23 Crumlin-Pedersen 1990, 105–9, 113; Haywood 1991, 11, 19–21, 45–7, 62–72, 136–7; Grainge 2005, 156–7; Dhaeze 2019, 74–6 (quotation), 83, 107, 167, 179, 186. 24 See Dhaeze 2019, 85, fig. 23; cf. 113, 183. 25 e.g. the revolt of Maternus, from c. 185, and subsequent political turmoil: Drinkwater 1983, 80, 213. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 281 fleet itself beyond around 250.26 It has therefore been proposed that the British forts were intended to frustrate coastal raiding by Picts from northern Britain, and to service late second- or early third-century Roman campaigns there.27 I suggest that they may also have played a part in warfare further south, i.e. receiving and supporting troops who moved inland to suppress unrest which led to the early walling of British cities in the late second and early third century.28 GAUL I Rising indigenous and Germanic piracy is, however, probably the best explanation for the next wave of Shore defences, along the Belgian coast, built probably under the breakaway Gallic Empire (260–274). At this time, ecological change, chronic Roman civil war and consequent increased Germanic pressure on the frontier may have resulted in further social upheaval and a rise in banditry and coastal piracy westwards from the lower Rhine.29 The Gallic emperors combated this by rebuilding existing earth-and-timber forts at Aardenburg and Oudenburg in stone.30 Again, the Germanic attackers were coastal, not trans-oceanic, raiders, at times, perhaps, aided by terrorised or opportunistic locals.31 The mercantile traffic between Britain and the Rhineland would have offered especially rich pickings, as would that between Britain and south-western Gaul, served by the port of Bordeaux.32 The destruction of the Gallic Empire by Aurelian (270–275) resulted in a breakdown of order in Gaul that provoked a savage peasants’ revolt, that of the Bagaudae,33 and encouraged an even greater wave of Germanic piracy. BRITAIN II The next and greatest wave of Shore-fort building took place on the south-eastern and southern coasts of Britain at, north to south: Burgh Castle, Walton Castle, Bradwell, Richborough, Dover, Lympne, Pevensey and Portchester. Although not identical, these forts are similar in layout and method of construction. They differ, however, from the earlier British forts by being more variable in plan and by possessing thicker and higher walls equipped with projecting bastions/artillery platforms. They are also characterised by their extensive use of brick bonding-courses. These (which allowed a significant reduction in the use of high-quality dressed stone) were already employed in civilian structures in the north-western provinces but were a new feature in defensive architecture. The change has been ascribed to the influence of eastern military practice on the western Empire, perhaps first seen in Aurelian’s new walls of Rome. In any event, his walls probably acted as the prototype for new western defences from the later third century, both military and civilian.34 For many years archaeology provided a no more reliable date for the construction of these forts than the later third century. The situation changed with the increasingly firm numismatic dating of the Shore fort at Richborough to the reign of the ‘British emperor’ Carausius (286–293) and, in particular, with the 26 27 28 29 30 31 32 33 34 Pearson 2002, 55, 57; Southern 2004, 397; Pearson 2005, 80 (quotation); Dhaeze 2019, 40, 106–7. Cotterill 1993, 236–8; Pearson 2002, 39–40, 53–5; 2005, 84; Dhaeze 2019, 42–3, 105 (earlier date). Johnson 1983, 129, cf. 196; Pearson 2002, 56, 94. cf. Haywood 1991, 28–30; see below ‘The Shore Forts during the fourth century/The Channel commands’. Dhaeze 2019, 106, 108-11, 211, 233. cf. Zos. 1.34.2–3. Cf. see below ‘The Shore Forts during the fourth century/Gaul II with Britain III’. Drinkwater 1983, 222; Casey 1994, 99. Drinkwater 1984, passim. Johnson 1983, 35, 41; Pearson 2002, 17–24, 26–38, 57–8, 71–4, 76–7; Dhaeze 2019, 276–82, 285–99. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 282 JOHN F. DRINKWATER dendrochronological dating of Pevensey to that of his successor, Allectus (293–296).35 Dendrochronology had already indicated that Allectus was probably responsible for major building work in London, the construction techniques of which resemble those of Pevensey, and the obvious inference is that all second-wave forts were built under these two rulers.36 Cardiff, which resembles Lympne and Portchester but is not usually considered a Shore fort, also probably belongs to the same period and may reflect official satisfaction with the Channel chain.37 This closer dating suits the historical context and helps resolve another chronological issue. Before the Pevensey dating, scholars sought evidence for the chronology of the second wave of British forts in Gaul where, in the late Empire, a number of towns – Nantes, Rennes, Angers, Le Mans, Orleans, Sens, Senlis, Beauvais, Saintes and Bordeaux – were given monumental walls (FIG. 1).38 Strategically sited and built according to the latest tenets of military architecture, these defences resemble the later British forts. Official milestones of Tetricus, the last ruler of the Gallic Empire, built into the walls of Nantes and Rennes give a reliable terminus post quem of 274 for the arrival of the type,39 but nothing to indicate when it was first adopted. Researchers therefore turned to a statement by Julian the Apostate that, after the barbarian incursions that followed the assassination of Aurelian, the emperor Probus (276–282) ‘revived seventy cities’.40 On this basis they argued that the creation of all major later third-century northern Gallic structures, both civil and military, and, by extension, the later British Shore forts, was initiated by Probus and continued down to the accession of Diocletian. In short, the Gallic town walls and the second wave of British Shore forts preceded Carausius’ revolt, with the forts perhaps being inspired by the walls.41 This is surely too early. Neither the Gallic walls nor the British Shore fortifications appear to have been thrown up in haste to counter an immediate threat. Furthermore, the elaborate and intrusive circuits of a number of Gallic town walls will have required imperial authorisation, close planning and the deployment of significant resources, both financial and human.42 I propose that this was most likely undertaken at a time of relative stability. Aurelian had shown the way by initiating the re-walling of Rome. Henceforth urban fortifications might be seen as patriotic, even fashionable.43 However, the defence of Rome, recently threatened by barbarian raiders, was a unique and immediate priority. Aurelian’s sudden death and renewed civil and foreign warfare will have inhibited the practice from spreading to Gaul, itself under severe Germanic attack. Probus took power in the East after Aurelian’s two short-lived successors, did not reach Gaul until 277, campaigned there for one year, and then returned eastwards. In 281 he was back in the West, but was in Rome by the end of the year and then moved to the Danube where, in 282, he was assassinated and replaced by Carus (282–283).44 Given the pressure Probus and Gaul were under, it is unlikely that any walling of Gallic cities began at this time.45 Julian does not specify that the ‘cities’ restored were Gallic; but if, as seems very likely, they were, he may 35 See below ‘Carausius and Allectus’. Richborough: Pearson 2002, 58; Dhaeze 2019, 42, 117; Wilmott and Smither 2020, 162–4. Pevensey: Fulford and Tyers 1995, 1011–13; Pearson 2002, 34, 59–60, 94–5; Williams 2004, 11; Elliott 2022, 115, 141–2. Dhaeze (2019, 42) extends the similarity to Portchester (cf. Pearson 2002, 36, 60, 163), Burgh Castle and Richborough. 37 Pearson 2002, 63; Dhaeze 2019, 136; below ‘The Shore Forts under the British Empire/Failure?’. 38 Johnson 1983, 95–6; Bachrach 2010, 38; Dey 2010, 14. 39 Johnson 1976, 101; 1983, 91–3, 113, 251; Drinkwater 1987, 42; Dey 2010, 17 and n. 40. 40 Julian., Caes. 314A: hebdomēkonta poleis anastēsas. 41 Johnson 1973, 222; 1976, 101–3, 112–18; 1983, 94, 201, 210–11, 251, 259; Drinkwater 1987, 222–3; Pearson 2002, 64; Southern 2004, 395; Dey 2010, 17–18 and n. 40. 42 cf. Johnson 1976, 114, 116, 223, 251; Cotterill 1993, 236; Bachrach 2010, 44–60; Dey 2010, 11, 17–18. 43 Johnson 1983, 118. 44 Drinkwater 2005, 53–7. 45 Contra Johnson 1976, 5; 1983, 63, cf. 251. 36 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 283 well have been using an established trope to designate a whole region by reference to its constituent communities, i.e., as Eutropius says, that Probus restored to Roman rule a Gaul that had been under barbarian occupation.46 I therefore support Butler’s proposition that the main period of construction of both military and civilian defences in northern Gaul occurred at some date after 285 and Diocletian’s victory over Carus’ son, Carinus, during the subsequent extended period of reconstruction and reform.47 This fits textual and archaeological evidence for a general acceleration in the construction of defences in the late third and early fourth century.48 (In Britain, the building of the technologically advanced walls of Caistor St Edmund may also have fallen in this period.49) The second wave of British Shore forts, the work of Carausius and Allectus undertaken during a very short period before the end of the third century, therefore preceded the Gallic urban fortifications. Their distinct character raises the question as to whether they had a different function from their British and Gallic predecessors and demands consideration of the circumstances of their construction. CARAUSIUS AND ALLECTUS In 286, Carausius, a senior military commander, revolted against Maximian and established himself in northern Gaul and Britain. I will, on analogy with the earlier Gallic Empire, use ‘British Empire’ to designate the state he created, and ‘Central Empire’ for that of his opponents. What we know of Carausius’ usurpation and its outcomes derives chiefly from: a near-contemporary Latin panegyric, delivered probably in spring 297; briefer references in other panegyrics of 289 and 310; and the later summary histories of Aurelius Victor and Eutropius.50 These receive invaluable confirmation, supplementation and correction from archaeological and numismatic studies. THE CREATION OF A BRITISH EMPIRE The reigns of Carausius and Allectus are sufficiently well known as to require only a sketch of their main events. In the autumn of 285 Diocletian became sole ruler of the Roman Empire and began to re-establish order after the long ‘Crisis’. In the same year he appointed Maximian as his junior colleague, creating a ‘dyarchy’ – the non-hereditary rule of two – and despatched him to Gaul against the Bagaudae. Maximian was quickly victorious, and in April 286 was promoted Diocletian’s full partner.51 Maximian then turned to suppress escalating Germanic piracy in the English Channel. Still probably in 286, he delegated the task to Carausius, who had distinguished himself during the Bagaudic campaign and – significant for the naval 46 Eutr. 9.17.2: Gallias a barbaris occupatas . . . restituit. Lütkenhaus 2012, passim. Butler 1959, 46–7. CIL 12.2229 (= ILS 620). Johnson 1973, 222; 1983, 62–3, 88, 94, 97–9, 101, 104, 106, 112, 113, 143, 163, 253– 5; Pearson 2002, 95; Bachrach 2010, 43, 60; Dey 2010, 9, 17 n. 40. 49 cf. Johnson 1983, 131, 133, 210. 50 Pan. Lat. 8(4).11–19 (the bracketed number is the chronological ordering of Galletier 1949 and 1952); 10(2).1.7– 12.8; 6(7).5.2–3; with Nixon and Rodgers 1994, 43, 106, 216. Aur. Vict., Caes. 39.20–21, 24, 39–42; Eutr. 9.21–2. Translations of the Panegyrici are those of Nixon and Rodgers. The full range of texts mentioning Carausius and Allectus can be found in Shiel 1977, 1–20 and Casey 1994, 191–8. I ignore all works after Eutropius because these are heavily dependent on him, or are highly fanciful, or both: see Shiel 1977, 20–30; Casey 1994, 168–90; Birley 2005, 374. 51 Chronology: Barnes 1982, 4 (with Maximian made Caesar in high summer 285); Kienast 1996, 266, 272; Birley 2005, 374. 47 48 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 284 JOHN F. DRINKWATER confrontation to come – had in his youth been a ship’s master.52 Victor states that Carausius was ordered to prepare a fleet and sweep marauding Germani from the seas. Eutropius adds that Carausius’ headquarters were in Boulogne; that his area of command extended from Belgica to Armorica, i.e. along the whole of the northern Gallic coast and Brittany; and that his foes were Franks and Saxons.53 Here both draw on a common source, usually taken to be Enmann’s lost ‘Imperial History’ or Kaisergeschichte (KG), completed around the middle of the fourth century. Since the KG is unusually well informed about affairs in Gaul at this time, we may accept what Victor and Eutropius say here as broadly reliable. Eutropius is, however, usually the less wordy which makes his three glosses unusual.54 Since it was most likely Franks who comprised the bulk of the sea-raiders of 286, Eutropius’ reference to Saxons is probably a later fourth-century inference: that piratical Germani of the later third century comprised (correctly) ‘Franks’ and (incorrectly) ‘Saxons’ in equal numbers.55 Therefore how far may we trust his identification of Carausius’ headquarters and his area of command? The Notitia Dignitatum’s two northern Gallic duces oversaw the Gallic coast from modern Flanders to Aquitaine, and the headquarters of the dux Belgicae II were probably in Boulogne. The similarity between this arrangement and Eutropius’ description of Carausius’ area of responsibility suggests another anachronistic gloss, reflecting the commencement of measures in his own day against troublesome Franks and Saxons.56 Yet there remains the KG’s basic reliability which, on balance, gives Eutropius the benefit of the doubt. Since he makes no mention of Britain and there is no sign of major piracy affecting the island at this time, Carausius probably oversaw only Gallic waters and estuaries.57 Carausius swiftly defeated the pirates but was then accused of gross financial misconduct. Probably in late 286, Maximian ordered his execution, but he escaped and declared himself emperor.58 He established his first mint, and so his capital, in Gaul, probably at Rouen.59 STRATEGY Carausius’ main weapon was his navy, comprising his anti-pirate fleet, presumably built afresh after the disappearance of the Classis Britannica, supplemented by new warships.60 His ‘conscription’ of Gallic merchants may have involved their provision of transports and experienced sailors, together with general war matériel and bullion required for the minting of coins.61 On land Carausius had the support of one full legion, probably XXX Ulpia Victrix, based at Xanten; some bodies of Germanic 52 Aur. Vict., Caes. 39.20; Eutr. 9.21. Cf. Shiel 1977, 15; Casey 1994, 49; 2015, 147: gubernator as river-pilot; second in command of a ship. 53 Aur. Vict., Caes. 39.20: parandae classi ac propulsandis Germanis maria infestantibus praefecere. Eutr. 9.21: cum apud Bononiam per tractum Belgicae et Armorici pacandum mare accepisset, quod Franci et Saxones infestabant. 54 Drinkwater 1987, 46–50; Burgess 1993, passim; Birley 2005, 373; contra Bartholomew 1979, 369. Cf. Casey 1994, 197–8. 55 So Bartholomew 1979, 369; Wood 1990, 94; Dhaeze 2019, 53. 56 Nixon and Rodgers 1994, 127 n. 39; Birley 2005, 372. 57 Britain: White 1961, 24; Pearson 2005, 77–8 (with qualifications); Dhaeze 2019, 56–7; contra Nixon and Rodgers 1994, 127, n. 39. Estuaries: Williams 2004, 13; Omissi 2018, 80. 58 Chronology: White 1961, 24; Johnson 1976, 25; Shiel 1977, 202; Barnes 1982, 11; Casey 1994, 39–43, 50–1; Nixon and Rodgers 1994, 50, 107; Pearson 2002, 45; Kienast 1996, 278; Williams 2004, 74; Birley 2005, 375. The earlier date better accommodates Maximian’s campaigning in 287: see below ‘Carausius and Allectus/Stand-off’. 59 Carson 1990, 137; Casey 1994, 71–4, 89–92; Williams 2004, 6; Birley 2005, 375. 60 Pan. Lat. 8(4).12.1: abducta . . . classe quae olim Gallias tuebatur aedificatisque praeterea plurimis in nostrum modum navibus (‘Carausius took all the fleet which once guarded the Gauls . . . and then in addition a great number of ships were built on the model of ours’). 61 Pan. Lat. 8(4).12.1: contractis ad dilectum mercatoribus Gallicanis (‘Gallic merchants were assembled for a levy’). White 1961, 28; Casey 1994, 98–9; Nixon and Rodgers 1994, 129, n. 43. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 285 cavalry (cunei); and, apparently, detachments from nine legions throughout the western Empire, including Britain.62 His recognition throughout the British provinces gave him another three full legions and their auxiliaries.63 He increased his strength by hiring ‘barbarians’, apparently mainly Franks, but that all these ‘were trained for naval warfare’ returns us to the fact that, from start to finish, Carausius’ rebellion and its suppression depended on naval power.64 Carausius’ options were limited. The overwhelming majority of the Roman army remained loyal to Maximian and Diocletian, which meant that Carausius could not expand significantly into their territory. He appears, therefore, very early to have decided to rely on his naval strength and content himself with holding northern Gaul and Britain. His aim was, apparently, by restricting his ambitions, to be eventually accepted by Diocletian and Maximian and so transform the dyarchy into a triarchy. His coinage frequently references ‘peace’ – pax -- and, probably in 291–2, he produced coins depicting himself in the company of his two imperial ‘brothers’ – fratres.65 STAND-OFF Maximian’s first reaction to Carausius’ usurpation was to campaign down the lower Rhine, taking control of the length of the river (including, presumably, the legionary base at Xanten).66 Then, however, instead of attacking Carausius in northern Gaul (in anticipation of which Carausius probably moved his minting to Britain67), he campaigned against the Alamanni.68 This suggests that Maximian soon recognised that, while Carausius could not destroy him, without a fleet he could not destroy Carausius; and also that Carausius would not attack him if he moved away from the Rhine. The stand-off allowed Carausius to maintain control over a swathe of territory in northern Gaul from 286 to 289 (and again from c. 290 to 293).69 It was interrupted by Maximian’s returning to the fray and, in 288–89, building a war-fleet on the middle Rhine and its tributaries.70 Since this fleet needed access to the sea, he sent troops to maintain free passage down the lower Rhine.71 These did not chase Carausius from Gaul. The forces involved were not Maximian’s main army; and the location of his shipyards suggests that at this time northern Gaul remained subject to Carausius.72 However, it was probably a major offensive by Maximian in the summer of 289 that weakened Carausius’ control of the region and caused him to move his headquarters to Britain. On the other hand, it 62 Pan. Lat. 8(4).12.1: occupata legione Romana, interclusis aliquot peregrinorum militum cuneis. Casey 1994, 94– 5, 98; Nixon and Rodgers 1994, 128–9, nn. 41–2. Cf. Nicasie 1998, 63–4, 110–12; Drinkwater 2007, 170; Rance 2015, 182; Dhaeze 2019, 175. Vexilations: Shiel 1977, 18; Casey 1994, 92–3; Nixon and Rodgers 1994, 128 n. 41; Williams 2004, 68–71; Birley 2005, 372; cf. Drinkwater 1972, 326. 63 Shiel 1977, 32–4, 203; Williams 2004, 6; Birley 2005, 377. 64 Pan. Lat. 8(4).12.1: sollicitatis per spolia ipsarum provinciarum non mediocribus copiis barbarorum (‘considerable forces of barbarians were attracted by means of the booty from the provinces themselves’) atque his omnibus ad munia nautica . . . eruditis. Shiel 1977, 9; Nixon and Rodgers 1994, 130, n. 45. Cf. Wood 1990, 94. 65 RIC 5.2.550: obv. CARAVSIVS ET FRATRES SVI. Jugate busts of Carausius, Diocletian and Maximian; rev. PAX AVGGG. Casey 1994, 95 and plate 5.6, 110–11; Nixon and Rodgers 1994, 107–8; Williams 2004, 6, 35 fig. 11, 71–3, 80, 90–1 pl. 2.8. 66 Pan. Lat. 10(2).7.1, 10.3–6; cf. 11(3).5.4. Barnes 1982, 57; Nixon and Rodgers 1994, 43, 68 nn. 35–6; cf. Drinkwater 2007, 182. 67 cf. Williams 2004, 8, 34–7 and fig. 11, 74. 68 Pan. Lat. 10(2).7.2–3; Barnes 1982, 57; Drinkwater 2007, 181. 69 Contra Shiel 1977, 4, 6, 205; Casey 1994, 91–2; Williams 2004, 6. 70 Pan. Lat. 10(2).12.3–4. Nixon and Rodgers 1994, 43. 71 Pan. Lat. 10(2).11.4, 11.7: iam milites vestri ad Oceanum pervenere; 11(3).7.2. Nixon and Rodgers 1994, 71 n. 39. 72 Contra Pan. Lat. 10(2).12.6. White 1961, 27; cf. Nixon and Rodgers 1994, 73 n. 43. Shiel’s (1977, 3–4) proposal that otherwise unknown ‘enemies’ kept Maximian out of the northern Gallic ports and rivers is unconvincing. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 286 JOHN F. DRINKWATER is likely that his naval power enabled him to continue to hold Boulogne;73 and there was no great invasion of Britain.74 Given the notorious unpredictability of Channel weather and the rarity of ancient pitched battles at sea, the most likely cause of the non-appearance of Maximian’s fleet off British shores was a storm; but his failure to launch another attack in better weather and Constantius I’s massive investment in his naval campaign of 296 leave open the possibility that Maximian suffered a defeat at sea.75 Maximian and Diocletian now had to accept a second stand-off. This is the likeliest time for Victor and Eutropius’ period of enforced peace, when the dyarchs tolerated Carausius but refused to recognise him as a colleague.76 By early 290, Maximian had moved away from northern Gaul.77 This was probably when Carausius re-established strong-points at Rouen and Amiens,78 but that he did not recommence minting on the Continent indicates that he now regarded Britain as his base. It may, therefore, have been from 289 that he began to improve the island’s defences by adding to the line of Shore forts.79 From 290 to 293 the West was relatively quiet.80 DESTRUCTION Diocletian and Maximian began to prepare for a new assault on Carausius from late 290, but no action was taken until 293.81 This was a crucial year in Roman history with the creation of the tetrarchy. Diocletian’s lieutenant in the East was Galerius. In the West, Maximian was joined by Constantius, proclaimed on 1 March. This removed all possibility of the acceptance of Carausius into the imperial college. Constantius was given responsibility for Gaul and the Germanies, where he arrived later in the same year and moved quickly but methodically against Carausius.82 His target was Boulogne, which he won by cutting off the city from external support both by land and, crucially (by sealing its harbour with a great jetty), sea.83 Constantius then turned to recovering Britain, for which he needed a fleet. This took time to build. He used his new ships to launch a great attack three years later, but it is possible that he attempted an earlier assault, with his lack of success being attributed to bad weather.84 In the meantime, Carausius was killed and replaced by Allectus, his senior finance minister or praetorian prefect.85 It is reasonable to assume that the coup was the immediate price of failure for the loss of Boulogne, but it may have occurred after Constantius’ first, abortive, attempt at conquest when Carausius’ political opponents, feeling that the danger of invasion had passed, felt secure enough to remove him.86 Constantius was finally victorious in 296 when he led one 73 Elliott 2022, 132. cf. Pan. Lat. 10(2).12.8; 11(3)19.4–5. Casey 1994, 106; Nixon and Rodgers 1994, 79, 107; Birley 2005, 379. 75 Haywood 1991, 39–40; Grainge 2005, 16; Elliott 2022, 121. 76 Aur. Vict., Caes. 39.39; Eutr. 9.22.2. Nixon and Rodgers 1994, 107–8, 131 n. 47. 77 Pan. Lat. 11(3).4.2–3; Barnes 1982, 58; cf. Nixon and Rodgers 1994, 79. 78 Contra Nixon and Rodgers 1994, 118 n. 24. 79 See below ‘The Shore Forts under the British Empire/Identifying the enemy’. 80 cf. Drinkwater 2007, 186. 81 Barnes 1982, 58; Nixon and Rodgers 1994, 93 and n. 53; Omissi 2018, 90. (Casey’s 1994, 108–9 proposal that an abortive attack was made down the Seine valley in 291 is unconvincing.) 82 Dates: Barnes 1982, 58, 60; Nixon and Rodgers 1994, 112 n. 8, 118 n. 21; Kienast 1996, 280. Planning: Pan. Lat. 8(4).13.1: inveteratum . . . instructum; cf. 6.1. Casey 1994, 111; Omissi 2018, 90. 83 Pan. Lat. 8(4).6.2–4; 6(7).5.2. Casey 1994, 112–14. ‘Jetty’ rather than ‘mole’ because it allowed tidal ebb and flow while preventing the movement of shipping: cf. Shiel 1977, 6; Dhaeze 2019, 148. Though Johnson (1976, 83) puts the harbour to the north of the old town, current opinion is that it lay to the south, in the cove of Bréquerecque: Grainge 2005, 148; Reddé et al. 2006, 239; Dhaeze 2019, 244. 84 Pan. Lat. 8(4).12.2: inclementia maris. Nixon and Rodgers 1994, 130 n. 46. 85 Aur. Vict., Caes 39–41: summae rei praeesset. Casey 1994, 113–14, 127; Birley 2005, 377. 86 Casey 1994, 42, 114; Nixon and Rodgers 1994, 108, 130–1, nn. 46, 47. 74 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 287 invasion fleet from Boulogne while Asclepiodotus, his praetorian prefect, took another from the Seine.87 To meet Constantius, Allectus had positioned himself in Kent. When Constantius failed to appear he moved to face Asclepiodotus, who had landed somewhere around the Solent, but was defeated and killed. Constantius did not engage Allectus because his fleet failed to land until probably just after Asclepiodotus’ victory. However, some of his troops managed to get ashore and take London, and Constantius then arrived and entered the city in triumph, as the man who had restored Britain from tyrannical darkness to imperial light.88 Constantius benefited massively from superintending the fall of Carausius and Allectus. He ruled the West first as Caesar and then, from 305, Augustus. In these capacities, he brought on his son and successor, Constantine I, with all that this meant for imperial and world history. But what of Carausius and Allectus? In particular, may it be said that their defensive fleet/fort system was just ‘a colossal waste of time and money’?89 THE SHORE FORTS UNDER THE BRITISH EMPIRE There have been various attempts to reconstruct the ‘tactical operation’ of the British and Gallic Shore forts, focusing on the fourth century and using the information given in the ND to decide if the forts’ job was to repel piratical raiding by Saxons or others, and if the British and Gallic forts formed a co-ordinated defence-system under a ‘Count of the Saxon Shore’. The most influential studies remain those of Johnson.90 He envisages an integrated Gallo-British defence system (limes) that evolved into the ND’s comital command. It comprised, on each side of the Channel, forts accommodating fleets and mobile troops, which were able to communicate with these and between themselves by means of signal towers. Germanic sea-raiders – Saxons or Franks, from around the estuaries of the Elbe and the Weser – took coastal routes that funnelled them into the Straits of Dover. Here Roman warships, despatched from the Shore forts, could intercept them before they began their attacks. Those who did manage to get through could be dealt with by the fort-garrisons when they landed. Those who avoided the Straits of Dover by attacking East Anglia faced a longer sea-crossing and other Shore forts protecting this area. Channel pirates who conducted successful raids would, on their departure, again have to face the Dover screen, now fully alerted to their presence. Johnson’s thinking has been challenged by Cotterill, who argues that, on the British side alone, high-intensity policing of around 300 miles (500 km) of coastline against random attacks would have been technically problematic and prohibitively expensive. Cotterill further contends that in the ND the British forts appear to have been largely unsupported by fleets or the ‘mixed units of infantry and cavalry’ necessary to combat Germanic pirates, and there is little archaeological evidence for ‘substantial barrack-blocks’ and stables. The forts are therefore unlikely to have accommodated an integrated land- and sea-defence system; and anyway the Saxon threat was a later fourth-century phenomenon. Cotterill therefore proposes an alternative use: that the British Shore forts were secure bases for the assembling and onward transport of state property and personnel. Cotterill’s case against Johnson has had a mixed reception but has been largely accepted by Pearson.91 It has its weaknesses, for example, the fact that Johnson does not actually envisage 87 For a fuller examination of the invasion, see below ‘The Shore Forts under the British Empire/Failure?’. Pan. Lat. 8(4).19.2; cf. the Arras medallion’s redditor lucis aeternae: RIC 5.2, 34. Casey 1994, 142; Nixon and Rodgers 1994, 140 n. 71. 89 Elliott 2022, 142. 90 Johnson 1976, 6–10, 114–31, with figs 1 and 70. Cf. Cotterill 1993, 232 (quotation); Pearson 2002, 121–4, 132– 3; contra Bartholomew 1979, 370; cf. below ‘The Shore Forts during the fourth century/Gaul II with Britain III’. 91 Cotterill 1993, 223–5 (quotation, 229), 232–3, 237–8; Pearson 2002, 136–8, 161; cf. Dhaeze 2019, 40, 155, 175, 131, 188; see above ‘The ‘Saxon Shore/The Notitia Dignitatum’. 88 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 288 JOHN F. DRINKWATER permanent high-intensity patrolling – his pirates simply swim into the Roman net at Dover and are dealt with between there and the Seine/Solent line. In addition, the absence of archaeological evidence for structures inside and adjacent to the Shore forts is no proof of their non-existence because the excavation of such buildings is notoriously difficult and so the evidence poor.92 More broadly, the ND does not offer a clear picture of conditions prevailing before its compilation so that for most of the fourth century we cannot say which particular units, if any, were in which forts at what time. On the other hand, with regard to patrolling, Johnson himself allows the deployment of reconnaissance patrols and the decades-long maintenance of vessels and garrisons on constant stand-by would have been a huge logistical strain.93 Furthermore, as Grainge observes, even if the forts were suitably garrisoned, the topography of Brancaster, Walton Castle, Caister and Burgh Castle would not have allowed the easy deployment of ground-troops in support of naval operations.94 Cotterill’s case is further strengthened by the fact that very few of the essential signal towers have been found;95 that, even if these existed, it is difficult to know how they might have operated (giving what sort of signals to whom, to do what?); that, however this was, Johnson concedes that poor visibility would have impaired their use;96 and finally that current experience shows how difficult it is to intercept small craft in the Channel, even with modern technology.97 I find Cotterill’s case basically persuasive, but in a Carausian context it is, like Johnson’s, too focused on the fourth century. In the last decade or so of the third century, a period of short but intense inter-Roman conflict, conditions were very different and Cotterill’s arguments of impracticality and expense may be discounted. What, then, was the tactical operation of the British forts in the 280s and 290s? IDENTIFYING THE ENEMY Like those of the first, the locations of all second-wave British forts suggest that they had a naval role.98 When the forts were given a Proban date it was widely believed that they were directed against post-Gallic Empire Channel piracy, i.e. that they were built, and by implication failed, to counter the growing menace that Carausius faced in 286.99 Again, however, there is no evidence for high levels of piracy along British shores in this period;100 and Burgh Castle, Walton Castle and perhaps even Bradwell seem, like Brancaster and Caister-on-Sea, to be too far from Channel waters to police them securely. More obviously, the British forts appear designed to confront trained troops rather than opportunistic raiders; and if defences of this quality were considered essential against pirates in Britain, why are they not found in Gaul? Such considerations led White to propose that all the British forts were built by Carausius against invasion by the dyarchs and tetrarchs.101 White’s chronology was dismissed as unrealistic and excessively ‘nationalistic’ by Johnson and fell out of favour.102 Although the Pevensey dating caused Fulford and Tyers to revive it, and it has consequently received some 92 Dhaeze 2019, 123, 180; Wilmott and Smither 2020, 172–3. Johnson 1976, 130; Grainge 2005, 145, 157. 94 Grainge 2005, 154–5. 95 Dhaeze 2019, 159–60; cf. 188 (though see now Elliott 2022, 142–3). 96 Johnson 1976, 128–9, 44; cf. Dhaeze 2019, 146, 155. 97 Dhaeze 2019, 155. 98 White 1961, 34–8; Pearson 2002, 99, 120–4. 99 Johnson 1976, 112, 126–30; 1983, 199; Pearson 2002, 62; Dhaeze 2019, 187. Cf. Haywood 1991, 37. 100 White 1961, 24: cf. 45, 134–5; see above ‘Carausius and Allectus/The creation of a British Empire’. 101 White 1961, 29–30, 33, 40–1, 56, 60. 102 Johnson 1976, 103–7; 1983, 251 (quotation). E.g. Haywood 1991, 37; Nixon and Rodgers 1994, 127 n. 39; Southern 2004, 399; Birley 2005, 384. 93 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 289 acceptance, this has been less than wholehearted.103 Yet apart from the fact that it is now clear that there was an earlier wave of British forts, and that Allectus contributed to the construction of the later, White’s proposal suits the political situation from 286. Moreover, the second-wave structures are large (averaging 3 ha/7.25 acres) and carefully built, requiring stable government like the Gallic town-walls of the late third and fourth century. However, their construction will have been less costly than the ‘significantly larger undertaking’ of the Gallic walls. They were ‘a modest enterprise’, well within the capabilities of an independent British regime.104 I therefore propose that towards the end of the third century the east coast British strongholds, previously used as military support-bases and naval police-stations, were augmented in the south-east to face likely attack by the Central emperors from the Rhine mouths and northern Gaul. All first- and second-wave forts were built close to safe harbours. Specialised port facilities were, apparently, not essential; these have been found at some forts, most famously at Dover, but not all.105 In the earlier period, these harbours or havens will have allowed naval support of imperial military expeditions and wider logistical operations;106 under the British Empire they will have served as anchorages for the home navy. The presence of a fort may therefore be taken as an indicator that its adjacent safe haven was a major naval facility but also, indeed, judged a potential landing-ground for an enemy invasion fleet. Pevensey’s construction on an ‘isolated and probably rather inhospitable peninsula’ suggests that it was not intended to protect a rich hinterland from coastal raiders but to overlook a strategically important harbour in a time of likely serious external – Roman – attack.107 When there was no immediate threat of invasion, the forts acted as the headquarters and maintenance depots of local flotillas of naval vessels, heavy and light, that would, at most, carry out low-intensity patrols.108 The reference in the ND to a Classis Anderetianorum, ‘the Pevensey squadron’, by then stationed in Paris, may indicate the earlier existence of such flotillas, named after their bases, in Britain.109 Carausius probably began the second wave of British coastal fort building in 289 when, having decided to base himself in Britain and facing attack from the mouths of the Old Rhine and Meuse and Scheldt (established commercial shipping routes), he strengthened the coasts of Norfolk, Suffolk, Essex and Kent.110 Although his probable continued control of Boulogne and recovery of large areas of northern Gaul from 290 meant that he remained free from attack from the Pas-de-Calais, the possibility that Central forces might establish a base on the Normandy coast may have provoked his construction of the fort at Portchester. Allectus was, with the building of Pevensey, responsible for the strengthening of this western extension of the chain after the loss of all northern Gaul left him facing attack from both the Pas-de-Calais and Normandy.111 103 Fulford and Tyers 1995, 1012–14. Cf. Nixon and Rodgers 1994, 127 n. 39; Pearson 2002, 136–7; Southern 2004, 399; Birley 2005, 384; Grainge 2005, 153, 159–60; Pearson 2005, 82; Dhaeze 2019, 42: surely wrong in saying that there is now ‘a general consensus’ in favour of White – cf. 178–80. 104 Size: Pearson 2002, 17, 23, 28, 32, 34, 36; cost: Pearson 2002, 76, 84–5, 86–9, 91–3, 96 (quotations). 105 E.g. Reculver, Richborough: Dhaeze 2019, 282, 284, 285, 287; Dover: Dhaeze 2019, 287. For the distinction between harbour and port: Grainge 2005, 90. 106 See above ‘The Shore Forts c. 200–286/Britain I’. 107 Pearson 2002, 118 (quotation); cf. Dhaeze 2019, 151: Caister and Burgh Castle. 108 cf. Grainge 2005, 64–5; Dhaeze 2019, 188. 109 Not. Dign. (occ.), 42.23. Johnson 1976, 123; Haywood 1991, 36; Southern 2004, 404–5; Dhaeze 2019, 156–7. 110 Milne 1990, passim; cf. Haywood 1991, 39; above ‘Carausius and Allectus/Stand-off’. 111 See above ‘Carausius and Allectus/Stand-off’; cf. Fulford and Tyers 1995, 1013. Portchester: Pearson 2002, 36; 2005, 76; Dhaeze 2019, 117, 299. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 290 JOHN F. DRINKWATER NAVAL POWER There was open contemporary acknowledgement of the importance of naval power in the confrontation between the British and Central Empires.112 We should, therefore, begin by considering ships, not forts. Carausius and Allectus were in an unforgiving fight to the finish with the Central emperors.113 Crucially, this meant stopping the enemy from landing an army in Britain by preventing them from gaining the initiative at sea. Even supplemented by Frankish recruits, the British army could never have matched its opponents in numbers and quality, and it suffered significant losses at Boulogne. The British Empire’s survival depended on its control of the Channel. If, as Constantius managed to accomplish at Boulogne, a Central ruler could break this hold and land an army, the advantage would pass to him but until then Carausius and Allectus had the upper hand. In the age of sail, uncertainties in weather and tides and in accessing landing places rendered taking any invasion fleet across the Channel ‘a venture fraught with risks and difficulties . . . a study of the attempts to invade Britain by Caesar, Claudius, Maximian and Constantius Chlorus, William the Conqueror, and the Spanish in the late sixteenth century and William of Orange shows that in total they involved sixteen cross-Channel fleet operations; of these, eight (50 per cent) failed to reach their destination’.114 The Central rulers had anyway first to build ships, and in this, too, they were at a disadvantage because construction took time and for them ‘ships’ meant both warships and transports. Carausius and Allectus had taken over an existing Roman war-fleet, had years to build more warships and did not need transports. In addition, ‘the most significant element in the complement of the ancient warship was the oar crew’, who needed to be sufficiently trained and experienced as to know automatically what to do in the chaos of battle.115 While Carausius and Allectus from the start had many experienced sailors at their disposal, the tetrarchs, in particular Constantius, had to recruit more or less from scratch. Finally, in confronting the enemy the British ships did not have to protect clumsier and slower-moving transports but could attack at will. Both sides, therefore, had to plan primarily for a confrontation at sea, not on land before the forts, a confrontation that the British rulers must win and that the Central rulers knew they might lose.116 Both, therefore, would be ready to pour money into naval resources and their deployment. Regarding deployment, the advantage again lay with the British emperors. They did not have to keep a significant number of ships at sea on permanent look-out, which would indeed have been too impractical and expensive.117 It was, in fact, unnecessary. Unlike pirates, Roman invaders would not come out of nowhere. There was a limited number of base-sites within safe striking-distance of Britain (the Rhine mouths; Boulogne; and the area between Boulogne and the Seine), and invasion fleets took time to create.118 It would not have been impossible for Carausius and Allectus, with their Continental contacts, to discover where and when an enemy armada had been begun, when it was ready to sail, and perhaps even when it had set out.119 Furthermore, in order to maintain cohesion, an invasion fleet would normally not dare leave in 112 e.g. Pan. Lat. 8(4).17.2, 4–6, on Constantius’ restoration of Roman naval prestige: gloria ex navalis. The speaker, who here also recounts the story of Probus’ Franks (see above ‘The Shore Forts c. 200–286/Britain I’), shows himself well aware of the crucial role played by naval supremacy. Cf. Casey 1994, 153: ‘The events surrounding the regimes of Carausius and Allectus all hinge on the possession and deployment of overwhelming naval forces’; Omissi 2018, 101. 113 Contra Elliott 2022, 135–6: that Allectus may have thought that his removal of Carausius might buy him some mercy from the tetrarchs. 114 Grainge 2005, 16, 122. 115 Grainge 2005, 61–2 (quotation), 63–4. 116 Contra Grainge 2005, 162. 117 See above ‘The Shore Forts under the British Empire’. 118 Grainge 2005, 32–4 and fig. 7. 119 cf. Haywood 1991, 39. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 291 unfavourable weather, but once it had embarked men, horses and equipment it had to sail fairly promptly. In addition, large fleets take a long time to leave harbour, and, since a convoy sails more slowly than a single vessel,120 it might have been possible for a small, swift boat to run ahead and warn of its imminent arrival. Finally, knowledge of a fleet’s place and rough time of departure would give its opponents a fair idea of its heading. The most direct route to Britain from the Seine, avoiding Portchester, would take Asclepiodotus around the Isle of Wight, explaining Allectus’ stationing of a fleet there. In conclusion, there should be no wholly unexpected attack,121 and high-intensity naval reconnaissance would have been necessary only when it was evident that a fleet was about to put to sea. So what was the British battle-plan? TACTICS The sole evidence for this is in ‘Latin Panegyric’ 8 where we are told that: 1) when Constantius and Asclepiodotus launched their invasion there was an ‘enemy fleet stationed in ambush off the Isle of Wight’; 2) when Asclepiodotus landed he met no significant resistance; 3) Allectus hurried to face Asclepiodotus; and 4) in doing so he ‘retreated from the shore he already held’ and deserted ‘his fleet and harbour’.122 It appears, therefore, that Allectus commanded at least two battle fleets, one stationed probably at Dover and the other near Portchester, both Shore forts, but only one field army.123 From this I propose that normally British warships were dispersed in safe havens such as sheltered tidal estuaries providing secure accommodation for large numbers of ships. When invasion threatened, battle fleets were assembled from these vessels in order to meet foes themselves heading for safe havens, places that facilitated invasion by allowing the simultaneous disembarkation of large numbers of men, horses and supplies. It is unlikely that the defending fleets operated in the uncertain waters of the open Channel. They probably went only as far as was necessary to make the best use of prevailing winds and tides and then, remaining in sight of land, employed tactics like those advocated by Vegetius: ‘Scouting skiffs (skafae) are attached to the warships . . . to intercept convoys of enemy shipping or by studious surveillance to detect their approach or intentions.’124 Although the skiffs also probably avoided open water, they moved further out and, being substantial oared vessels, moving faster than any invasion convoy, could detect its advance and return ahead of it to alert the heavier warships – Mediterranean-style galleys armed with bronze battle-rams to pierce enemy hulls.125 These then moved to attack their prime target, the transports, when these were at their most vulnerable: nearing the end of their journey, full of sea-sick and fearful soldiers and horses, and their crews weary with the effort of keeping their heavily laden vessels together. Only in the case of a successful enemy landing, which represented a major strategic defeat, did responsibility for defence pass to the field army. The fort-garrisons provided the policing necessary in all naval dockyards, but when the flotillas left to fight they had a different role. Dealing with a full Roman invasion force was not the same as dealing with a handful of opportunistic overseas raiders who might be trapped in a haven or estuary.126 If a Roman enemy broke through at sea, the garrisons were too small to deter or 120 Grainge 2005, 129–30. Contra Johnson 1976, 30, 131. 122 Pan. Lat. 8(4).15.1: inimica classis apud Vectam insulam in speculis atque insidiis conlocata; 15.5: ab eo litore quod tenebat abscessit . . . classem portumque deseruit. 123 Dover: Eicholz 1953, 44. 124 Veg., Mil. 4.37: Scafae tamen maioribus liburnis exploratoriae sociantur . . . commeatus aduersariorum nauium aliquando intercipi adsolet et speculandi studio aduentus earum uel consilium deprehendi (trans. Milner 1993). 125 Haywood 1991, 39; Casey 1994, 16; Grainge 2005, 46, 157–9; Dhaeze 2019, 155–6; Elliott 2022, 101, 150–2; contra Haywood 1991, 48. 126 Contra Grainge 2005, 157–9; cf. Haywood 1991, 48. 121 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 292 JOHN F. DRINKWATER repel a landing.127 However, their forts could withstand short-term siege, which would have allowed them to summon the field army and, if they still held their own haven, to provide anchorage for friendly naval vessels. In short, they might incommode an attack, but by then the advantage would have passed to the Central Empire since it would have been very difficult for the single British field army to reach even one landing-place in time to defeat an invading force; down to the end of the age of sail, most seaborne expeditions that managed to make their way safely to a landing place were unopposed.128 It is important to accept the wholly subsidiary role of the forts. We should not allow their impressive remains to make us forget that they were built to support the navy, whose ships now exist only as images on coins. Again, the main plank of British defence policy was victory at sea. FAILURE? Prima facie both ships and forts failed to do their job. Allectus, aware of an imminent double attack, stationed two fleets to meet it, but neither defeated Constantius’ ships at sea and one Central army landed and destroyed Allectus in battle. However, this does not signify that the British fleet/fort scheme was entirely useless. It took ten years and enormous effort for the Central emperors to remove their British rivals. Maximian appears to have suffered a major naval failure in 289 and unease caused by this was probably heightened by the appearance of the new Shore forts – uncomfortable reminders of the power of the navy over which they watched. On the British side, experimentation with the forts’ design suggests that the fleet/fort model was considered successful. Indeed, the contemporary fort at Cardiff, built against raiders across the Irish Sea but on the lines of those on the Channel coast, suggests that it was adopted as the norm.129 Respect for the fleet/fort system explains Constantius’ bypassing of it in 296.130 The port for the shortest and safest crossing to Britain was Boulogne. Asclepiodotus took the significantly longer and more unreliable route from the mouth of the Seine to land, presumably, in or near the Solent.131 Since he could not have foreseen the mist that hid his fleet on his arrival, he must have expected to fight his way through and land in a haven west of Portchester, or between Portchester and Pevensey.132 Constantius himself sailed from Boulogne, but it seems clear that, despite the shorter crossing, he never intended to land in East Sussex or southern Kent. It has been proposed that, because Asclepiodotus arrived first and made a successful landing, Constantius sailed without transports to divert Allectus’ attention from Asclepiodotus’ approach. Only after Asclepiodotus had established a bridgehead would Constantius set foot in Kent.133 This is unlikely. If Allectus knew about Asclepiodotus’ fleet he must also have known about Constantius’; and if Constantius failed to load an appropriate number of transports with the appropriate number of infantry and cavalry he would be informed, suspect a trick, and ignore this fleet. Constantius had to sail with a full expeditionary force which, given the constraints noted, could not remain long at sea: it must have been more than diversionary, but aimed at what? The panegyrist’s remarks that Allectus retreated from the shore he already held and deserted his fleet and harbour suggest that Allectus was waiting for Constantius on the Kent coast.134 Although aware of Asclepiodotus’ attack, he seems to have calculated that 127 128 129 130 131 132 133 134 Contra Dhaeze 2019, 130; cf. Johnson 1976, 104–5. Grainge 2005, 161. Johnson 1976, 132–4; see above ‘The Shore Forts c. 200–286/Gaul I’. Grainge 2005, 160. Grainge 2005, 15, 122–4. Perhaps in one of the estuaries of the Arun, Adur, Ouse and Cuckmere: Grainge 2005, 126. Eichholz 1953, 44, 46. Frere 1978, 381. Contra Eichholz 1953, 43; Shiel 1977, 12: inland, in defence of London. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 293 Constantius would take the traditional invasion route, and that the Caesar was the more important foe: defeating him would hamstring the whole invasion. For Constantius, therefore, attempting a landing on the southern coast would have been an enormous challenge. He is said to have sailed before Asclepiodotus.135 Although this could be just a rhetorical conceit – Caesar must precede prefect – taken with subsequent events it suggests another target. If Constantius had headed for East Sussex/southern Kent, by sailing first he would have reached Britain earlier than Asclepiodotus, endangering himself and nullifying the impact of simultaneous attacks.136 On the other hand, if he kept out to sea and, on a heading which might have been easily anticipated only if he had sailed from the Rhine, by-pass Reculver and land well up the Thames estuary, he could, notwithstanding Allectus’ prior knowledge of his attack, win an element of tactical surprise. This would allow him to evade Allectus’ fleet, remain synchronised with Asclepiodotus, and attack London – to judge from Carausius’ and Allectus’ attempts to secure the city, a recognised major strategic target.137 The troops who, despite having become detached from Constantius’ command in poor visibility, landed and took London, may, therefore, have been following orders given prior to their departure.138 Thus the fleet/fort system could not be ‘easily outflanked’: it compelled two widely distant landings against an enemy prepared for action.139 Risky even in good weather, in the poor conditions under which both were launched it was a huge gamble which almost failed. Each fleet sailed in, presumably unplanned for, poor weather. Asclepiodotus’, indeed, heading NNW and so needing a wind from the south-east, ran into what seems to have been a brisk Channel ‘sou-wester’, blowing to the north-east.140 Asclepiodotus managed the crossing, but his convoy was no doubt disordered by the adverse wind and it then faced a British war-fleet. He was able to evade this and make a safe landing without a battle because of a change in the weather that ended the wind and brought a covering mist and, perhaps, also because the enemy, not expecting his imminent arrival in such unfavourable conditions, had not deployed scouting vessels. Elsewhere, however, the onset of mist caused Constantius’ ships to lose contact with each other and prevented any chance of effecting roughly simultaneous landings.141 Yet, having been despatched, both forces had to fight it out. In particular, there could have been no thought of Constantius returning to Gaul and awaiting events:142 his desertion of his forces at this time would have been unforgiveable. This may also help explain Asclepiodotus’ ‘histrionic gesture’ of burning his boats immediately on arrival.143 However, fortune was on Constantius’ side. First, mist and, perhaps, a certain element of surprise helped his ships to elude Allectus’ Dover fleet. Second, with no sighting of Constantius, Allectus, judging that he posed no immediate 135 Pan. Lat. 8(4).14.4: prior. cf. Shiel 1977, 12. 137 London: Eichholz 1953, 44 (although viewing a line of attack beyond the Straits of Dover as accidental, caused by the poor weather conditions); Haywood 1991, 40 (although judging Asclepiodotus’ assaul as a feint); Grainge 2005, 150, 162; Elliott 2022, 116, 137–9. 138 Pan. Lat. 8(4).17.1–2. As Casey (1994, 137) observes, these troops, described only loosely as ‘your’ (vestri), may have belonged to Asclepiodotus’ forces. However, the panegyrist says that these were ‘separated’ at sea by the mist, which does not fit his description of the western invading force, actually saved from disruption by poor visibility. Cf. Eichholz 1953, 44. 139 Contra Johnson 1976, 105. 140 Pan. Lat. 8(4).14.4: fervidum . . . Oceanum . . . caelo et mari turbidis . . . die pluvio . . . ventum quia derectus non erat captaret obliquum. Eichholz 1953, 41 n. 3; Nixon and Rodgers 1994, 134 n. 53; Grainge 2005, 150; Dhaeze 2019, 77–8. 141 Mists: Pan. Lat. 8(4).15.1; 17.1: tantae . . . nebulae . . . nebulosi . . . maris. Nixon and Rodgers 1994, 138 n. 63. 142 Contra Eichholz 1953, 46. 143 Pan. Lat. 8(4).15.2: statim atque Britanniae litus invaserat, universis navibus suis iniecit ignes. Shiel 1977, 11 (quotation); contra Elliott 2022, 146: that it occurred because Asclepiodotus anticipated a swift victory; but cf. Grainge immediately below. 136 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 294 JOHN F. DRINKWATER threat, marched to fight Asclepiodotus.144 A land battle was, of course, what the Central Empire will have desired, but the reasons for Allectus’ defeat demand some consideration. The obvious explanation is that he was outnumbered, but this raises the question of the size of the invasion forces. Vegetius’ recommended size for an expeditionary force is 20,000 infantry and 4,000 cavalry.145 Adopting Grainge’s calculations, to transport the infantry would have required a minimum of 286 vessels, with the cavalry needing 276 more, making a total of 573. In addition, there would have had to be a large escort of galleys, say 200, making a total of 773 ships. This is in line with Julius Caesar’s invasion fleet of 54 B.C. and with the 600 or more ships that the emperor Julian is said to have gathered to restore the provisioning of northern Gaul from Britain in the mid-fourth century.146 However, if we assume that each of the invasion armies was around 24,000 strong, requiring around 1,600 ships in total, plausibly large becomes less plausibly massive. Constantius might have had to settle for two half-armies or two two-thirds armies, each around 12,000 or 18,000 strong, respectively. Grainge, indeed, proposes that Asclepiodotus burned his boats because his army was too small to protect them when he moved inland.147 If so, with Constantius’ army gone missing the Central Empire’s superiority in numbers would have been significantly reduced. The panegyrist accuses Allectus of attacking in haste and disarray. This could be the standard invective of the victor, but the fact that there were very few Central casualties suggests that the British army should have done better than it did, and that something went badly wrong for Allectus.148 A final piece of good luck for the Central emperors was that Constantius’ reputation was saved in his absence by his men taking London. A concatenation of circumstances meant that the Central Empire was able to avoid two pitched sea-battles with superior naval forces and turned the third-century Battle of Britain into a land conflict which Allectus, for reasons unknown, proved incapable of winning.149 At the start, however, the fleet/fort system forced Constantius to take risks which made victory far from certain. The tetrarchs’ appreciation of the riskiness of the task in hand is indicated by the fact that in 296 Maximian maintained his watch on the Rhine, but through his presence, not by force. This suggests that Germani posed no real threat and that he was on hand rather to prevent trouble in Gaul if things went wrong for Constantius in Britain.150 The system should not be judged a total failure. Famously, however, lucky generals are preferable even to good generals. Allectus was not lucky, but by 296 Constantius had already proved himself in this respect. As Eichholz remarks, this appears to have inspired such confidence and loyalty in his men that they went forward in his name even when things seemed to be going against them and confirmed his good fortune.151 THE SHORE FORTS DURING THE FOURTH CENTURY BRITAIN III After Constantius’ re-conquest, the British forts underwent another change of use. There are the usual problems in establishing a reliable chronology, with conclusions based on coin finds 144 Shiel 1977, 12–13; Casey 1994, 138; contra Eichholz 1953, 43, from Pan. Lat. 8(4).15.5: that Allectus abandoned Dover when he saw Constantius’ sails in the offing. 145 Veg., Mil. 3.1. 146 Caes., Bgall. 5.8.6; Julian., Ep. ad Ath. 279D–80A, 80C (660 ships); Lib., Orat. 18.82–3, 87; Amm. Marc. 18.2.3; Zos. 3.5.2 (800 ships). Grainge 2005, 81–2, cf. Matthews 2015, 616; see below ‘The Shore Forts during the fourth century/Gaul II with Britain III’. 147 Grainge 2005, 150. 148 Pan. Lat. 8(4).15.16.1–3. 149 cf. Grainge 2005, 61, 63. 150 Pan. Lat. 8(4).13.3. Drinkwater 2007, 187; contra Nixon and Rodgers 1994, 132 n. 48. 151 Eichholz 1953, 46. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 295 being particularly debateable,152 but different sites appear to have experienced different phases and intensities of occupation. Burgh Castle and Caister were abandoned by c. 380, while Brancaster and Bradwell persisted into the fifth century. Reculver and possibly Dover seem to have gone into a decline from c. 300. Reculver was abandoned by c. 360, but Dover probably remained in use into the later-fourth and maybe even the fifth century. Richborough was neglected c. 300, but may have been restored c. 340 and continued for the rest of the fourth century and into the fifth. Lympne was totally abandoned by c. 350 but Pevensey seems to have persisted into the fifth century. Portchester was neglected then abandoned down to c. 325, when it was reoccupied and properly maintained down to c. 365. After 365 it continued in occupation into the fifth century, although after c. 378 under increasingly less military discipline.153 Thus, while most forts were kept operational during the fourth century, there is no sign that this was in order to confront a particular major threat. However, in establishing what the new role or roles of the British Shore forts may have been, we must consider them alongside developments in Gaul. GAUL II WITH BRITAIN III At some stage the number of Gallic forts was increased, producing the line of coastal defences, from Flanders to Aquitaine, recorded in the ND and confirmed by textual and archaeological evidence (FIG. 1) However, because of a lower intensity of archaeological investigation, we know less about these than we do about the British forts.154 The most accessible study remains Johnson’s of 1976, now supplemented by the publications of Reddé and his collaborators and of Dhaeze.155 The extended Gallic chain is very different from the British. Few of its forts are new, massive, ‘purpose-built’ structures, with most being sited in existing settlements, especially towns. In addition, they are far more widely distributed.156 It is usual, following Johnson, to date the second wave of Gallic forts, like the British, to the later third century: after the walling of the Gallic towns but before the revolt of Carausius, as part of a co-ordinated defence system against Germanic pirates.157 Having earlier argued that the second British wave should be dated to the time of the British Empire and before the building of the new Gallic town walls, which began after 296,158 I propose that the construction of the new Gallic coastal defences also fell after 296. This fits what new evidence we have for them. The fort at Brest was built with a thick stone wall and external towers, characteristic of the monumental Gallic town walls, suggesting that it was begun in the fourth century.159 In the mid/late fourth century, a new fort at Alet/Saint-Malo, with a brick-bonded tower complex, was built over the purposely demolished centre of a defended civil settlement in a manner 152 Johnson 1976, 137–8; cf. Pearson 2002, 164–5; Dhaeze 2019, 153. Burgh Castle and Caister: Cotterill 1993, 235; Pearson 2002, 167; Brancaster and Bradwell: Pearson 2002, 168– 9; Reculver and Dover: Cotterill 1993, 235; Pearson 2002, 163–4, 167, 169; Richborough: Johnson 1976, 140, 150–1; Pearson 2002, 164–5, 169; Lympne: Cotterill 1993, 235; Pearson 2002, 167; Pevensey: Pearson 2002, 168; Portchester: Johnson 1976, 137–8, 140, 151; Cotterill 1993, 235; Pearson 2002, 36–8, 163, 169. Dhaeze 2019, 123, 129 fig. 45, 154, 180, 183. 154 Johnson 1976, 101–2. 155 Johnson 1976, 72–93, 94–113; Reddé et al. 2006; Dhaeze 2019 (although his catalogue goes only to the mouth of the Somme, his text considers more western sites). Cf. Monteil 2017, 25–6 for a possible additional coastal fort in Brittany. 156 White 1961, 60–1; Johnson 1976, 124, 147 (quotation). 157 Johnson 1976, 101–2, 112–13, 116; 1983, 210–11. 158 See above ‘The Shore Forts c. 200–286/Britain II’. 159 Reddé et al. 2006, 242–3; Dhaeze 2019, 137. 153 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 296 JOHN F. DRINKWATER reminiscent of the disruption caused by the building of the Gallic town walls.160 A fourth-century dating is now being suggested for other Gallic defended sites.161 The forts of the western Channel and Brittany cannot have been constituent elements in an integrated Gallo-British system against major Germanic attack. The Saxon menace was yet to come and, anyway, in addition to the logistical and technological difficulties already noted, they are too widely spaced, and their supposed British counterparts show no sign of being maintained to face a present major enemy. This stands even if we accept Johnson’s restricted specialised stretch from Flanders to the mouth of the Seine, and his proposal that some fort sites here may have been lost to the sea.162 The most likely purpose of all the Gallic forts was therefore to combat persistent everyday piracy. Carausius had suppressed the Germanic pirates of his day, and during the period of the British Empire focus on naval control of the Channel would have inhibited their revival, even under conditions of civil war. After Carausius, the firm government established by the Diocletianic reforms and a long imperial presence in Gaul will have resulted in continuing policing of these waters.163 There is no evidence for high-level piracy in the Mediterranean or the Channel during the first half of the fourth century.164 Such as there was would have been the usual maritime banditry, rising at times of public insecurity due to civil war or barbarian incursion but falling once order had been secured.165 Home-grown pirates – provincials raised in the trade or forced into it by adverse circumstances – may, up to a point, have been tolerated by local coastal populations. From Ausonius we know of one landowner who was prepared to work with ‘opportunistic’ brigands, so why not pirates?166 In addition, in the Channel we may assume continued opportunistic raiding by Germani from the Rhine/Meuse basin.167 In the Bay of Biscay, continuing trade between Bordeaux and Britain may have attracted the attention of North African raiders.168 Threats here would explain the presence of the southernmost Gallic Shore fort at Blaye, and maybe also the walling of Saintes and Bordeaux. Bordeaux’s harbour was, unusually, enclosed within its walls, suggesting that the city had a particular importance, perhaps as a centre for the collection and onward transportation of goods collected as tax in kind, the annona militaris. The nearby coastal fort at Bayonne, although listed in the ND as being under the authority of the magister militum praesentalis, not the dux tractus Armoricani et Nervicani, may also have been involved in this.169 Piracy was a bane, but posed no existential threat to the Empire. A subsidiary use of the Gallic forts was probably, as earlier in Britain, to service imperial personnel stationed inland. Although the main force of Roman troops was on the Rhine, from the early fourth century northern Gaul accommodated settlements of people known as laeti and gentiles, under military prefects. We have no clear understanding of the nature or occupation of either category, but both probably comprised Germanic migrants given asylum within the Empire on 160 Reddé et al. 2006, 379–80; Monteil 2017, 22; Dhaeze 2019, 125, 137–8; see above ‘The Shore Forts c. 200–286/ Britain II’. 161 Cribellier 2017, 44, 47; Kasprzyk and Monteil 2017b, 6; Monteil 2017, 19, 26; Kasprzyk 2017, 112; Dhaeze 2019, 125, 138, 185. 162 Johnson 1976, 125, 127 and fig. 79. 163 cf. Johnson 1983, 199. 164 Bartholomew 1984, 84; Haywood 1991, 40; de Souza 1999, 228; contra Casey 1994, 10: ‘frequent and devastating attacks’. 165 See e.g. Amm. Marc. 28.2.10 for a strange outbreak of lawlessness on land in Gaul in 369, under Valentinian I. 166 Auson., Ep. 13.23–27 (Green). Green 1991, 628 (quotation); cf. Tomalin 2022, 246–7; see above ‘The Shore Forts c. 200–286/Gaul I’. 167 Cotterill 1993, 231; see above ‘The Shore Forts c. 200–286/Gaul I’. 168 British trade: CIL 13.634 (=ILS 7523); Auson., Parent. 7, 18. Drinkwater 1983, 222 and n. 44; Sivan 1993, 43. Piracy: Johnson 1983, 131. 169 Amm. Marc. 14.10.2; 17.8.1; Not. Dign. (occ.) 42.19. Johnson 1983, 130; Esmonde Cleary 2013, 129–30; see below ‘The Shore Forts during the fourth century/The Channel Commands’. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 297 condition that they maintained themselves and supplied recruits to the army.170 In addition, Belgica I and II housed a number of government workshops producing arms and textiles. Overall, therefore, the Gallic coastal forts formed no emergency ‘Atlantic Wall’ but comprised a series of bases constructed over many years and used for routine maritime policing and government logistics. They presumably housed naval flotillas, but the existence of these lacks proof.171 In 296, the British forts must have been immediately taken over in order to secure communications with the Continent and hinder ‘a disaffected population’ from rejecting Central control.172 Richborough was used in this way in 368, when Count Theodosius landed there to suppress barbarian and civil disorder on the island.173 More broadly, however, the job of the British forts probably became much the same as the Gallic: to discourage common piracy and safeguard government communications and supplies. With regard to piracy, the southern forts would revert to the work done by those of the west coast, e.g. like the new fort at Cardiff. They again watched over havens not as potential landing-grounds but rather as potential targets for casual sea-raiders.174 The distribution and type of the British west coast chain is, as Johnson remarks, tellingly like the installations of the Gallic Atlantic coast.175 The relative neglect of the southern chain suggests, however, that here such raiding was a lesser problem. Logistical operations return us to the annona militaris, the exaction of which may have been particularly enforced in Britain (unlike northern Gaul and Germany, relatively unscathed by the conflicts of both the third and fourth centuries) in order to provision Roman personnel in the Rhine region. Dhaeze challenges this interpretation on the grounds that archaeologists have not found the ‘large numbers’ of warehouses (horrea) necessary to store provisions in or around the forts. However, these must have existed even if, as in Gaul, late Roman horrea are hard to find in the archaeology. As we have seen, in 358–59 Julian prepared a fleet of several hundred ships to convey supplies from Britain up the Rhine.176 Libanius remarks that this was a ‘long established’ practice that Julian was simply ‘renewing’ because it had been interrupted by the recent usurpation of Magnentius. The ‘regular transfer of products from Britain’ is confirmed by Ammianus.177 Osteology has indicated large-scale meat processing around some forts, but it is likely that most operated mainly as secure collection points, with foodstuffs and other items being produced inland and transported to them by water for temporary storage in warehouses.178 The mass collection and storage of goods in, and onward transport from, British ports would have necessitated the use of all forts within easy sailing distance of the Rhine, including the Kentish.179 This has implications for how we reconstruct the development of the post of comes litoris Saxonici. THE CHANNEL COMMANDS Who was in charge of the Gallic and British Shore forts during the fourth century? Current common opinion, based on Johnson’s concept of a co-ordinated, cross-Channel, maritime 170 Drinkwater 2010, passim; Edwell 2015, 351; Le Bohec 2015b, passim; Syvänne 2015a, 32; 2015b, 407–8. cf. Dhaeze 2019, 40–1,248. 172 Wood 1991, 315. 173 Amm. Marc. 27.8.6. Cotterill 1993, 239. 174 cf. Dhaeze 2019, 150, 162. 175 Johnson 1976, 136, 147. 176 Casey 1994, 148; Pearson 2002, 127, 161; Birley 2005, 423–4; Dhaeze 2019, 43; Drinkwater 2019; see above ‘The Shore Forts under the British Empire/Failure?’. 177 Lib., Orat. 18.83: palai . . . ananeoumenos; Amm. Marc. 18.2.3: annona a Britannia sueta transferri. 178 Pearson 2002, 161; 2005, 83; Grainge 2005, 158. 179 Contra Pearson 2002, 161. 171 https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 298 JOHN F. DRINKWATER limes, is that after the fall of the British Empire Carausius’ broad anti-Germanic pirate command was revived, maintained and expanded and, by the end of the fourth century, was held by a ‘count of the Saxon Shore’.180 Most cited here is Ammianus Marcellinus’ account of the great northern barbarian attack on Britain, mounted in 367, which resulted in the death of Nectaridus, comes maritimi tractus, ‘count of the coastal zone’, interpreted as a periphrasis of the ND’s comes litoris Saxonici.181 However, there is an awkward oddity in Ammianus’ account. Immediately after the death of Nectaridus he describes the capture of the dux Fullofaudes. In the late Empire, frontier troops were commanded by regional duces, so Fullofaudes was probably the current dux Britanniarum.182 Since he oversaw the security of Britain’s four provinces, he was a very senior officer, so why do we find him mentioned alongside another of even higher rank whose responsibilities were only coastal in a period when piracy was only just becoming a major problem? In addition, there are again the difficulties involved in maintaining a permanent cross-Channel defence system, the irregular maintenance of the British Shore forts and the absence of any advanced, second-wave British-type fort in Gaul. A discrete Gallo-British marine defence system under a single official is implausible, and I agree with Cotterill that Ammianus’ ‘count of the coastal zone’ probably had a senior ‘logistical’ command concerned with the provisioning of the Rhineland from Britain.183 Since Roman Britain had a long tradition of troops operating away from the main northern frontier area, I propose that for most of the fourth century all British troops, including those in the Shore forts, were commanded by the dux Britanniarum. After Count Theodosius had restored order, the post of dux Britanniarum was continued and so listed in the ND.184 Later fourth-century British duces will have supervised the strengthening of anti-piratical installations on the eastern and western coasts against increased raiding from the north and across the Irish Sea. In Gaul, although there was much less of a tradition of troops requiring supervision away from the Rhine, under the high Empire and in the fourth century we find some guarding key points in the communication network behind the frontier.185 We may therefore envisage the coastal forts being placed, like the British, under a regional dux. For north-east Gaul – from Flanders to the Seine estuary – this is likely to have been the dux Belgicae II of the ND.186 This is because of the semi-abandonment of the lower Rhine due to flooding caused by a rise in sea-level: the ‘Dunkirk II marine transgressions’. This, as we have seen, may have already provoked regional piracy in the later third century.187 Rome held the lower Rhine as a shipping route but no longer as a frontier.188 Germania II was effectively abandoned, along with its ducate, and an inner defence line was created in Belgica II along the highway that linked Cologne, the western 180 Johnson 1976, 142–4, 147; Drinkwater 1987, 221; Wood 1991, 314–15; Casey 1994, 115, 144; de Souza 1999, 226; Pearson 2002, 132–4; Southern 2004, 401. 181 Amm. Marc. 27.8.1. Johnson 1976, 144; Esmonde Cleary 2004, 410. 182 Jones 1964, 1.52, 54–56, 60, 2.608; Esmonde Cleary 2004, 410; Southern 2004, 401. 183 Cotterill 1993, 238. 184 Amm. Marc. 27.8.9; Not. Dign. (occ.) 40; Birley 2005, 400. 185 Early Empire: Drinkwater 1983, 84–5. Late Empire: Not. Dign. (occ.) 42.13–17. Jones 1964, 1.99; Nouvel and Venault 2017, 73. Not. Dign. (occ.) 38. 186 Not. Dign. (occ.) 38. 187 Haywood 1991, 27–30; Esmonde Cleary 2013, 48; Brulet 2017, 120–1 (figs 1, 2); Dhaeze 2019, 111, 187. See above ‘The Shore Forts c. 200–286/Gaul I’. Other factors, including human damage to the environment, may also be to blame: Dhaeze 2019, 45–7, 72. 188 Johnson 1983, 145; Drinkwater 1987, 28; Casey 1994, 48; Drinkwater 2007, 52–3; Esmonde Cleary 2013, 47–8, 52, 70–1; Brulet 2015, 594–5; 2017, 120–1, 122, 128; Dhaeze 2019, 119–20; cf. Roymans et al. 2020, 277–84; see above ‘The Shore Forts c. 200–286/Gaul I’. Cf. (in Britain) Pearson 2002, 102. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 299 military capital, to Boulogne, the provincial capital and seat of the dux Belgicae II.189 From the Seine to the Garonne, however, we must look elsewhere, because the internal provinces of Lugdunensis II and III, Aquitania II and Novempopulana did not have duces and because their coastal fort system appears to have grown only slowly and not to any great intensity. Here, the most likely supervising officer – treating the ND’s reference as part of its chronological depth – is the magister militum praesentalis, shown as still being in overall command of the fort at Bayonne, together with a miscellaneous collection of western fleets (including those near the Rhône) and garrisons, and of units of laeti settled in Gaul.190 THE SAXON SHORE From the ND we may infer that major change took place before 400. While most of its frontier commands are still those of provincial duces, the Gallic list appears incomplete and irregular, with traditional duces of Sequania and Belgica II but novel ones of Moguntiacum (Mainz) and of a tractus Armoricanus et Nervicanus.191 In Britain, the dux Britanniarum is accompanied by two comites: a comes Britanniarum and a comes litoris Saxonici.192 In addition most of the units listed by the ND as forming the garrisons of the Shore forts appear to be late creations.193 I follow the view that this was the work of Stilicho, regent of the young emperor Honorius in the West after the death of Theodosius in 395, which returns us to the ‘Saxon’ threat. Down to the mid-fourth century Saxons were, like all other Germani, no worse than adventitious raiders. From 370, however, they became a particular menace and took Germanic marauding to new heights of socio-economic disruption.194 This may have resulted from improvements in their maritime technology, including the ability to build bigger vessels propelled by sail.195 We know from the court poet Claudian that in the period 395–400 Stilicho made a whistle-stop tour of the Rhine from its source to the sea, not campaigning but ensuring that his Germanic allies made no trouble when he turned his attention to urgent matters in the East.196 To explain the entries in the ND it has long been conjectured that at this time Stilicho, in order to increase his control of the system, restructured the military administration of the West. The result was either the re-shaping or the dismantling of a long-evolving, integrated, Gallo-British ‘Saxon Shore’ command, under a comes, to suit new circumstances.197 I propose instead that Stilicho actually created this command. In his account of Stilicho’s Rhenish activities, Claudian prioritises his dealings with Franks and Alamanni and has a personified Gaul praising him for having subdued Germani and Franci.198 Next, however, he describes Britannia as thanking Stilicho for relieving her from attacks by Hibernians, Scots, Picts and Saxons, and declaring that she no longer needs ‘to keep watch along all my coasts for the Saxon who would come whatever wind might blow’. The same sentiment is expressed more briefly in Claudian’s invective against Eutropius: ‘the Saxon is 189 Johnson 1983, 138: Wightman 1985, 208; Drinkwater 1987, 218–19, 221–2; Haywood 1991, 27–8; Esmonde Cleary 2013, 48; cf. Dhaeze 2019, 141; see below ‘The Shore Forts during the fourth century/The Saxon Shore’. 190 Not. Dign. (occ.) 42. 191 For Nervicanus see below. 192 Jones 1964, 2.609–10. 193 White 1961, 64–5, 68; Birley 2005, 400. 194 White 1961, 50–1, 55; Haywood 1991, 41, 45; Pearson 2002, 128; cf. Dhaeze 2019, 81; see above ‘The Saxon Shore/The Notitia Dignitatum’. 195 See above ‘The Shore Forts c. 200–286/Britain I’; cf. Haywood 1991, 64–5; Pearson 2005, 80. 196 Claud., Cons. Hon. 439–59; Cons. Stil. 1.189–245; 2.186–9, 240–4. Goetz et al. 2006–7, 2.210–17 and nn. 78, 83. 197 Bury 1923, 1.118 n. 3; Demandt 1970, 616–27; Scharf 1990, passim; Wood 1990, passim; Haywood 1991, 53–4. 198 Claud., Cons. Stil. 2.240–6. Drinkwater 2007, 46. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 300 JOHN F. DRINKWATER conquered and the seas safe’.199 The reference to Stilicho’s salvation of Britain, which he is unlikely to have visited in person, is surprising but implies that his Rhenish tour caused him to reorganise the military administration here as well as in Gaul. If so, Claudian’s associating of ‘Saxon’ and ‘coast’/‘shore’ is suggestive: the island was now under pressure from Saxons capable of striking at will and without warning.200 The question of how to defend the specifically British litus Saxonicum – a coast now, indeed, unusually named after its attackers201 – may have been in the air in military circles at this time, but was dealt with in a wider, Gallo-British context. I propose that the coastal command of the dux Belgica II was replicated westwards, with the responsibility for all the coastal forts of the western Channel and Atlantic, except Bayonne, being transferred from the magister militum praesentalis – i.e. relieving Stilicho himself – to a new dux tractus Armoricani et Nervicani. The inclusion of Nervicani, i.e. ‘of the Nervii’, in this title defies easy explanation because the civitas of the Nervii lay inland in Belgica II. It could have been that, as a result of the marine transgressions, the dux Belgicae II lost some of his previous responsibilities. If so, supervision of the Cologne–Boulogne highway, with its major node at Bavai, the former capital of the Nervii, could have been transferred to the nominally junior but actually more important dux tractus Armoricani et Nervicani.202 In Britain, the perception of a Saxon threat forced the appointment of a dux litoris Saxonici alongside the dux Britanniarum. Not long afterwards, again perhaps in recognition of the severity of the danger and because, as we shall see, his authority extended to two bases across the Channel, the dux litoris Saxonici was promoted comes.203 In addition, maybe before this promotion, Britain was also given a general commanding a field army, a comes Britanniarum. An exact chronology is impossible. Scharf locates all the changes between 395 and 408, but the crucial point is that they fell very late. The impression is one of improvisation. The comes litoris Saxonici has no fleet and no field troops, only a mixed bag of frontier units – limitanei.204 In Gaul, the ND lists two other regiments of limitanei, under the dux Belgicae II and dux tractus Armoricanus respectively, in bases described as being ‘on the Saxon Shore’ – in litore Saxonico.205 However, although the lists give each a location – at Grannona and Marcis respectively – in the illustrated pages introducing each of the ducal commands and the garrisons for which they are responsible this is missing. Since Claudian implies that contemporary Saxons were now a threat only to Britain, I propose that the ‘Saxon Shore’ was established as a purely British measure, and that the two Gallic bases were not part of a matching Gallic ‘Shore’ but, though under the nominal command of the two Gallic duces, were placed at the disposal of the British comes litoris Saxonici.206 This should have increased British strength, but the whole suggests limited effectiveness. Restoring the British forts to full preparedness would have been hugely difficult given the political, military and financial strains of the time. In addition, assuming that here the ND’s information actually applies to the period around 400, the units assigned to these and the Gallic bases would, for reasons already outlined, be unequal to the task. They lacked cavalry and supporting fleets; and they could never have communicated effectively with each other and 199 Claud., Cons. Stil. 254–5: ne litore toto prospicerem dubiis venturum Saxona ventis; In Eutrop. 1.392–3: domito quod Saxone Tethys mitior (trans. Platnauer 1922). 200 cf. Sid. Apoll., Epist. 8.6.14–16. Elliott 2022, 93. 201 cf. Johnson 1976, 145–6; Pearson 2002, 130–1; 2005, 73–4; Dhaeze 2019, 144–5. 202 cf. Wightman 1985, 208–9. 203 cf. Dhaeze 2019, 146. 204 Fleet: Haywood 1991, 54. 205 TABLES 2–3: Not. Dign. (occ.) 38.7/7: Marcis (at Marck); Not. Dign. (occ.) 37.14/21: Grannona (in ?Guernsey). 206 Johnson (1976, 88–93, 143–4) comes close to this thinking, but on the assumption that a Gallo-British command under a Count of the Saxon Shore existed throughout most of the fourth century, only to be broken up towards its end. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 301 across the Channel.207 I propose that Stilicho’s answer to the problem of increasing Saxon attacks on Britain was more on paper than in men, ships and bricks. The ‘Saxon Shore’ could have done little against new-found Saxon strength. It may have helped to protect cross-Channel communication in the turmoil caused by other barbarian incursions and civil wars of the early fifth century, but thereafter it probably soon disappeared. Department of Classics and Archaeology, University of Nottingham john.f.drinkwater@nottingham.ac.uk BIBLIOGRAPHY Bachrach, B.S. 2010: ‘The fortification of Gaul and the economy of the third and fourth centuries’, Journal of Late Antiquity 3, 38–64. Barnes, T.D. 1982: The New Empire of Diocletian and Constantine, Cambridge MA. Bartholomew, P. 1979: Review of Johnson 1976, Britannia 10, 367–70. Bartholomew, P. 1984: ‘Fourth-century Saxons’, Britannia 15, 169–86. Birley, A.R. 2005: The Roman Government of Britain, Oxford. Brennan, P. 2015: ‘Units in the Notitia Dignitatum’, in Le Bohec 2015a, 1034–9. Brulet, R. 2015: ‘litus Saxonicum’, in Le Bohec 2015a, 594–602. Brulet, R. 2017: ‘Les agglomérations de Germanie Seconde aux IVe et Ve s. apr. J.-C.’, in Kasprzyk and Monteil 2017a, 119–46. Burgess, R.W. 1993: ‘Principes cum tyrannis 2. The KG and the lost History of Eusebius Nanneticus’, Classical Quarterly 43, 495–99. Bury, J.B. 1923: History of the Later Roman Empire from the Death of Theodosius I to the Death of Justinian (2 vols), London. Butler, R.M. 1959: ‘Late Roman town walls of Gaul’, Archaeological Journal 116, 25–50. Carson, R.A.G. 1990: Coins of the Roman Empire, London. Casey, P.J. 1994: The British Usurpers Carausius and Allectus, London. Casey, P.J. 2015: ‘Carausius’, in Le Bohec 2015a, 147–8. Cotterill, J. 1993: ‘Saxon raiding and the role of the Roman coastal forts of Britain’, Britannia 24, 227–39. Cribellier, C. 2017: ‘Les agglomérations du Centre de la Gaule: types d’occupations et évolution du réseau (IIIe–VIe s. apr. J.-C.)’, in Kasprzyk and Monteil 2017a, 39–60. Crumlin-Pedersen, O. 1990: ‘The boats and ships of the Angles and Jutes’, in McGrail 1990, 98–110. de Souza, P. 1999: Piracy in the Graeco-Roman World, Cambridge. Demandt, A. 1970: ‘magister militum’, RE Suppl. 12, 553–790. Dey, H. 2010: ‘Art, ceremony and city walls’, Journal of Late Antiquity 3, 3–37. Dhaeze, W. 2019: The Roman North Sea and Channel Coastal Defence. Germanic Seaborne Raids and the Roman Response, Wetteren. Drinkwater, J.F. 1972: ‘A new inscription and the legionary issues of Gallienus and Victorinus’, Numismatic Chronicle 7.11, 5–26. Drinkwater, J.F. 1983: Roman Gaul. The Three Provinces, 58 BC–AD 260, London. Drinkwater, J.F. 1984: ‘Peasants and Bagaudae in Roman Gaul’, Classical Views n.s. 32, 349–71. Drinkwater, J.F. 1987: The Gallic Empire. Separatism and Continuity in the North-Western Provinces of the Roman Empire A.D. 260–274, Historia Einzelschriften Heft 52, Wiesbaden. Drinkwater, J.F. 2005: ‘Maximinus to Gallienus and the “Crisis”’, in A.K. Bowman, P. Garnsey and A. Cameron (eds), Cambridge Ancient History vol. XII (2nd edn), Cambridge, 28–66. Drinkwater, J.F. 2007: The Alamanni and Rome 213–496. Caracalla to Clovis, Oxford. Drinkwater, J.F. 2010: ‘Laeti’, in H. Heinen et al. (eds), Handwörterbuch der Antiken Sklaverei (im Auftrag der Akademie der Wissenschaften Mainz), CD-ROM-Lieferung I–III, Stuttgart. 207 See above ‘The Shore Forts under the British Empire/Identifying the enemy’. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press 302 JOHN F. DRINKWATER Drinkwater, J.F. 2019: Review of Martin, S. (ed.), Rural Granaries in Northern Gaul (6th century BCE–4th century CE): From Archaeology to Economic History (Radboud Studies in Humanities 8), Leiden, Bryn Mawr Classical Review 2019.12.11. Edwell, P.M. 2015: ‘Eastern Auxiliaries: Late Empire’, in Le Bohec 2015a, 351–2. Eichholz, D.E. 1953: ‘Constantius Chlorus’ invasion of Britain’, Journal of Roman Studies 43, 41–6. Elliott, S. 2022: Roman Britain’s Pirate King. Carausius, Constantius Chlorus and the Fourth Roman Invasion of Britain, Barnsley. Esmonde Cleary, S. 2004: ‘Britain in the fourth century’, in Todd 2004, 409–27. Esmonde Cleary, S. 2013: The Roman West AD 200–500, Cambridge. Faleiro, C.N. 2005: La Notitia Dignitatum. Nueva edición critica y comentarió historico, Nueva Roma 25, Madrid. Frere, S.S. 1978: Britannia (rev. edn), London. Fuentes, N. 1991: ‘Fresh thoughts on the Saxon Shore’, in V.A. Maxfield and M.J. Dobson (eds), Roman Frontier Studies. Proceedings of the XVth International Congress of Roman Frontier Studies, Exeter, 58–64. Fulford, M., and Tyers, I. 1995: ‘The date of Pevensey and the defence of an “Imperium Britanniarum”’, Antiquity 69, 1009–14. Galletier, E. 1949: Panégyriques latins. Tome I (I–V), Paris. Galletier, E. 1952: Panégyriques latins. Tome II. Les panégyriques constantiens (VI–X), Paris. Goetz, H.W., Patzold, S., and Welwei, K.-W., 2006–7: Die Germanen in der Völkerwanderung. Auszüge aus den antiken Quellen über die Germanen von der Mitte des 3. Jahrhunderts bis zum Jahre 453 n. Chr. (2 vols), Darmstadt. Grainge, G. 2005: The Roman Invasions of Britain, Stroud. Green, R.P.H. 1991: The Works of Ausonius, Oxford. Haywood, J. 1991: Dark Age Naval Power. A Re-Assessment of Frankish and Saxon Seafaring Activity, London. Johnson, S. 1973: ‘A group of late Roman city walls in Gallica Belgica’, Britannia 4, 210–23. Johnson, S. 1976: The Roman Forts of the Saxon Shore, London. Johnson, S. 1983: Later Roman Fortifications, London. Jones, A.H.M. 1964: The Later Roman Empire 284–602. A Social and Administrative Survey (3 vols and maps), Oxford. Kasprzyk, M. 2017: Les agglomérations/vici/castra du Centre-Est de la Gaule: morphologie et fonctions (IIIe–VIIe s. apr. J.-C.), in Kasprzyk and Monteil 2017a, 91–117. Kasprzyk, M., and Monteil, M. (eds) 2017a: Agglomérations, vici et castra du Nord de la Gaule entre Antiquité tardive et début du haut Moyen Âge, Gallia 74.1, Paris. Kasprzyk, M., and Monteil, M. 2017b: ‘Agglomérations, vici et castra du Nord de la Gaule (IIIe–VIe s. apr. J.-C.): esquisse d’un bilan’, in Kasprzyk and Monteil 2017a, 1–12. Kienast, D. 1996: Römische Kaisertabelle (2nd edn), Darmstadt. Kulikowski, M. 2000: ‘The Notitia Dignitatum as a historical source’, Historia 49, 358–77. Kulikowski, M. 2015: ‘Notitia Dignitatum’, in Le Bohec 2015a, 695–6. Le Bohec, Y. (ed.) 2015a: Encyclopedia of the Roman Army, London. Le Bohec, Y. 2015b: ‘Gentiles’, in Le Bohec 2015a, 427. Lütkenhaus, W. 2012: ‘Observations on Zosimus’ British “cities”’, Britannia 43, 268–70. Matthews, S. 2015: ‘Logistics: Principate’, in Le Bohec 2015a, 612–19. McGrail, S. (ed.) 1990: Maritime Celts, Frisians and Saxons, Council for British Archaeology Research Report 71, London. Milne, G. 1990: ‘Maritime traffic between the Rhine and Roman Britain: a preliminary note’, in McGrail 1990, 82–4. Milner, N.P. (trans.) 1993: Vegetius: Epitome of Military Science, Translated Texts for Historians 16, Liverpool. Monteil, M. 2017: ‘Les agglomérations de la province de Lyonnaise Troisième (Bretagne et Pays de la Loire). Entre abandon, perduration et nouvelles creations (IIIe–VIe s. apr. J.-C.)’, in Kasprzyk and Monteil 2017a, 15–37. Nicasie, M. 1998: Twilight of Empire. The Roman Army from the Reign of Diocletian until the Battle of Adrianople, Dutch Monographs on Ancient History and Archaeology 19, Amsterdam. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press THE ‘SAXON SHORE’ RECONSIDERED 303 Nixon, C.E.V., and Rodgers, B.S. 1994: In Praise of Later Roman Emperors. The Panegyrici Latini, Berkeley and Oxford. Nouvel. P., and Venault, S. 2017: ‘Les agglomerations du Centre-Est de la Gaule. Quel devenir durant l’Antiquite tardive?’, in Kasprzyk and Monteil 2017a, 61–89. Omissi, A. 2018: Emperors and Usurpers in the Later Roman Empire. War, Panegyric and Legitimacy, Oxford. Pearson, A.F. 2002: The Roman Shore Forts. Coastal Defences of Southern Britain, Stroud. Pearson, A.F. 2005: ‘Barbarian piracy and the Saxon Shore: a reappraisal’, Oxford Journal of Archaeology 24, 73–88. Philp, B. 2005: The Excavation of the Roman Fort at Reculver, Kent, Dover. Platnauer, M. (ed. and trans.) 1922: Claudian (Loeb edn, 2 vols), Cambridge MA. Rance, P. 2015: ‘Cavalry: Late Empire’, in Le Bohec 2015a, 178–86. Reddé, M., Brulet, R., Fellmann, R., Haalebos, J.K., and von Schnurbein, S. (eds) 2006: Les fortifications militaires. L’architecture de la Gaule romaine, Bordeaux. Roymans, N., Dirks, T., and Heeren, S. 2020: ‘Roman imperialism and the transformation of rural society in a frontier province: diversifying the narrative’, Britannia 51, 265–94. Scharf, R. 1990: ‘Die Kanzleireform des Stilicho und das römische Britannien’, Historia 39, 461–74. Seeck, O. 1876: Notitia Dignitatum accedunt Notitia Urbis Constantinopolitanae et Laterculi Provincium, Berlin. Shiel, N. 1977: The Episode of Carausius and Allectus: The Literary and Numismatic Evidence, British Archaeological Reports British Series 40, Oxford. Sivan, H. 1993: Ausonius of Bordeaux. Genesis of a Gallic Aristocracy, London. Southern, S. 2004: ‘The army in late Roman Britain’, in Todd 2004, 393–408. Symonds, M. 2018: Protecting the Roman Empire. Fortlets, Frontiers and the Quest for Post-Conquest Security, Cambridge. Syvänne, I.Y. 2015a: ‘Allies: Late Empire’, in Le Bohec 2015a, 32–3. Syvänne, I.Y. 2015b: ‘Foederati’, in Y. Le Bohec 2015a, 407–9. Todd, M. (ed.) 2004: A Companion to Roman Britain, Oxford. Tomalin, D. 2022: Roman Vectis. Archaeology and Identity in the Isle of Wight, West Cowes. White, D.A. 1961: Litus Saxonicum. The British Saxon Shore in Scholarship and History, Madison WI. Wightman, E.M. 1985: Gallia Belgica, London. Williams, H.P.G. 2004: Carausius. A Consideration of the Historical, Archaeological and Numismatic Aspects of his Reign, British Archaeological Reports British Series 378, Oxford. Wilmott, T., and Smither, P. 2020: ‘The plan of the Saxon Shore fort at Richborough’, Britannia 51, 147–74. Wood, I. 1990: ‘The Channel from the 4th to the 7th centuries AD’, in McGrail 1990, 93–7. Wood, I. 1991: ‘Internal crisis in fourth-century Britain’ (review of A.S. Esmonde Cleary, The Ending of Roman Britain, 1989), Britannia 22, 312–15. https://doi.org/10.1017/S0068113X23000193 Published online by Cambridge University Press