¹ see Appendix.

² Western province: Birley, E., ‘The Fourth-Century Subdivision of Britain’, V Congressus internationalis limites Romani Studiorum (Zagreb, 1963), 83Google Scholar (based on memorandum written and circulated in 1958); Ann Dornier, ‘The Re-Organization of the North-West Frontier in Britain in A.D. 369’, Roman Frontier Studies (ed. E. Birley, M. Jarrett, B. Dobson, 1974), 102-5. Eastern province: M. Hassall, ‘Britain in the Notitia’, Aspects of the Notitia Dignitatum (ed. R. Goodburn & P. Bartholomew, 1976), 109.

³ Stevens, C. E., ‘Magnus Maximus in British History’, Étudies Celtiques iii (1938), 90Google Scholar.

⁴ Hind, J. G. F., Historia xxiv (1975), 101–11Google Scholar.

⁵ Ammianus Marcellinus xxviii, 3 (ed. J. C. Rolfe, III (1958), 134); Notitia Dignitatum (ed. O. Seeck (1962 reprint), 171-2); Laterculus Polemii Silvii (ibid., 260).

⁶ Bibl. Nat. ms. fr. 22, 321, p. 730. Relevant section quoted in G. H. Doble, The Saints of Cornwall, Part 5 (1970 reprint), 124. The saint is variously called Sulian/Suliau/Suliavus/Sulinus. Either dicitur is in the wrong tense or a verb such as vocatum est has been omitted.

⁷ Gualent (<Latin Valentia) could theoretically be either a borrowing of the Welsh rendering of the name, *Gualeint, or a direct borrowing from Latin into Primitive Breton, *Ualint (with the characteristic loss of the epenthetic vowel after i-affection), for both ei and i were reduced to e in Old Breton; but if it is an extract from a Welsh Life (see below) it is most likely to be derived from Welsh *Gualeint. (Clearly this statement cannot have been culled directly from a late classical source, for such a source could not have included a sixth-century British ruler.) The writer is indebted to Professeur Leon Fleuriot for giving his blessing to the contents of this footnote and also footnote 9.

⁸ Doble, op. cit. (note 6), 113-21.

⁹ Removing its (incorrectly) restored Latin ending, the orthography of Gualent indicates a form that would have been current in Breton from the mid-ninth to the late eleventh century. Doble maintained that this Welsh Life did not reach Brittany until the fifteenth century (op. cit. (note 6), 119-21), but if it were borrowed as late as that one would have expected the ei to be preserved. Moreover, the fact that Gu has not been written Gou suggests a pre-Middle Breton date, that is to say, before the twelfth century.

¹⁰ J. G. F. Hind, op. cit. (note 4), 108.

¹¹ As Ammianus Marcellinus states explicitly that it was a province which had been lost to the enemy (op. cit., note 5), the clear implication is that this fifth province had already come into existence sometime between the compilation of the Verona List, 312-4, and 367. Its original name has now been lost owing to its being re-named Valentia. In a letter which the writer received from the late C. E. Stevens just before his death he suggested that this fifth province was created by Constans in 343 and was originally called Constantia. It is to be hoped that his work on the subject will be published.

¹² A later medieval Welsh source suggests that south-west Wales was known to have been in Prima - see Davies, W. S., ‘The Book of Invectives of Giraldus Cambrensis’, Y Cymmrodor xxx (1920), 24Google Scholar. Internal evidence indicates that this information could have been taken either from the Breviarium of Festus or the Notitia Dignitatum. If the latter, it implies the area remained in Prima after the creation of Valentia. In which case, as it is unlikely to have been a detached portion, south-east Wales must also have remained in Prima. On other grounds it seems probable that south-east Wales was not in Valentia (see note 30).

¹³ Jones, A. H. M., The Later Roman Empire III (1964)Google Scholar, Map 11: ‘Dioceses and Provinces According to the Notitia Dignitatum’, shows Valentia comprising Wales and north-west England. On the map is the following note: ‘The geographical arrangement of the British provices is conjectural (based on Mann, J. C., Antiquity xxxv (1961), 317–20)’. He offers no explanation as to why he included Wales which Mann assigned to PrimaGoogle Scholar.

¹⁴ For the division of the north into two provinces see Mann, J. C., ‘The Administration of Roman Britain’, Antiquity xxxv (1961), 320Google Scholar, note 22; Ann Dornier, op. cit. (note 2). There is no evidence for a sixth province. It is curious, however, that the heading of the Verona List gives Britain six provinces and that Polemius Silvius adds the Orcades as a sixth province.

¹⁵ see A. L. F. Rivet, ‘The Notitia Galliarum: Some Questions’, 137, fig. 5, in Goodburn & Bartholomew, op. cit. (note 2).

¹⁶ see note 13.

¹⁷ Eric Birley, Roman Britain and the Roman Army (1961), 64-8.

¹⁸ The writer intends to discuss elsewhere the evidence for these two points.

¹⁹ Information from T. Strickland.

²⁰ The writer will review this evidence in a forthcoming publication.

²¹ M. Hassall, loc. cit. (note 2).

²² J. C. Mann, op. cit. (note 14), 318.

²³ see note 11.

²⁴ Notitia Dignitatum, Oc. i & ii.

²⁵ The writer has argued elsewhere that the west coast was more vulnerable - see Ann Dornier, ‘Was there a Coastal Limes in Western Britain in the Fourth Century?’, Roman Frontier Studies 1967 (Tel Aviv, 1971), 19.

²⁶ e.g. Dux Belgicae Secundae with only three forts (Notitia Dignitatum, Oc. xxxviii).

²⁷ For distribution see C. M. Daniels, Roman Frontier Studies 1979, ed. W. S. Hanson and L. J. F. Keppie (1980), 173-93.

²⁸ Ann Dornier, op. cit. (note 2), 103; J. C. Mann, ‘What was the Notitia for?’, in Goodburn & Bartholomew (op. cit., note 2) 1-5.

²⁹ Only the place-names of the first group of units (York province - other units) are listed at the top of the return, within the frame. This in itself suggests that this first part was a self-contained separate ‘sheet’. Item per lineam valli seems to have been tacked on to this, omitting the appropriate list of place-names at the beginning of its two sections.

³⁰ Before the creation of Valentia presumably any return would have been made up from lists from the York province and Prima. The writer intends to argue elsewhere that from c. 343 south-east Wales was not within the command of the Dux and that therefore there would no longer be a list for Prima.

³¹ Britannia viii (1977), 358; ix (1978), 404.

³² There is no evidence that the Seguntienses, subsequently found serving in Illyricum, evacuated the fort as early as 383; and even if some were transferred at that date this does not necessarily mean that the whole garrison went.

³³ Ann Dornier, op. cit. (note 25), 19.

³⁴ Moreover, if the north-west was the most vulnerable area, communications may have been quicker by sea from Chester than overland across the Pennines from York.

³⁵ F. H. Thompson, Roman Cheshire (1965), 40.

³⁶ Although at a later date one might have to consider the possibility of civil buildings encroaching on the fort area, this is unlikely in the late third century.

³⁷ Zosimus, Historia Nova iv, 35 (ed. J. Buchanan & H. Davis, 1967), 168). His soldiers makes it clear that he held some military post at the time of his usurpation.

³⁸ ibid. He crossed to Gaul immediately after his elevation. The Anonymous Gallic Chronicle has got events in the wrong order because for some reason his crossing is mis-dated to 381.

³⁹ The Irish Picti of Powys belong to a later period (see note 20).

⁴⁰ It has been suggested that he was consular governor of Valentia with responsibility for a local militia (loc. cit. note 3). Apart from the fact that there is no evidence of such a local militia, one might wonder whether such a force would have been large or powerful enough to have enabled him to seize power with such ease.

⁴¹ Ifor Williams, Breuddwyd Maxen (1908), i-xxi.

⁴² ibid., 9-10: Ac yna y gwnaethant wynteu amherawdyr newyd; ac yna y gwnaeth hwnnw lythyr bygwth ar Vaxen. Nyt oed hagen o lythyr, namyn, ‘O deuy di, ac o deuy di byth i Ruuein –.’ Ac hyt yg Kaer Llion y doeth y llythyr hwnnw ar Uaxen a'r chwedleu; ac odyna yd anuones ynteu lythyr ar y gwr a dywedei y uot yn amherawdyr yn Ruuein. Nyt oed yn y llythyr hwnnw heuyt dim, namyn, ‘Ot af ynheu y Ruuein, ac ot af-.’

⁴³ see note 20.

⁴⁴ W. S. Davies, op. cit. (note 12), 130-1. The writer is grateful to Dr A. K. B. Evans for pointing out that this is a more reliable text than that printed in the Rolls Series.

⁴⁵ F. Haverfield, ‘A Roman Inscription from Cirencester Illustrating Fourth-Century Britain’, Oxoniensis (1948), 225.

⁴⁶ C. E. Stevens, ‘The Notitia Dignitatum in England’, in Goodburn & Bartholomew, op. cit. (note 2), 212-3. Certainly Giraldus’ numbering and order does correspond with that of the Laterculus. Stevens does concede that Giraldus must have tampered with this information.

⁴⁷ He may have acquired such information during one of his visits to Ireland where he spent some time studying Irish sources, although one cannot rule out the possibility of direct contact between Wales and Scotland bringing such information south at an earlier period.