The "Notitia Dignitatum" as a Historical Source Author(s): Michael Kulikowski Source: Historia: Zeitschrift fr Alte Geschichte, Vol.

49, No. 3 (3rd Qtr., 2000), pp. 358-377 Published by: Franz Steiner Verlag Stable URL: http://www.jstor.org/stable/4436586 Accessed: 13/08/2009 13:16
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=fsv. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Franz Steiner Verlag is collaborating with JSTOR to digitize, preserve and extend access to Historia: Zeitschrift fr Alte Geschichte.

http://www.jstor.org

THE NOTITIA DIGNITATUM AS A HISTORICAL SOURCE

TheNotitiaDignitatum is a peculiar, illustrated list whichitemisesthe administrativehierarchy, both civil and military,of the late Romanempire.1 It is our greatestsurvivingsourcefor the bureaucracy of thatempire,but it is a unique source and nothing comparable to it survives from its epoch. Study of the Notitiais therefore tangledup in thornyquestionsthatperrnit no sureanswers. In the first place, the Notitia lacks a self-evidentcontext.We possess deluxe copies of a lavishly illustrated Carolingian copy of a late Romanoriginal,but whatpurposethatoriginalservedis a matter of controversy.2 We cannotknow for certainwhetherthe originalwas a uniqueexample,drawnup for a specific purpose or, as mostassume,one of manysuchadministrative lists whichplayed a regularpartin the functioning of late Romangovernment. The questionhas recentlybecome morecomplicated still, as a recentarticlehas arguedthatthe Notitia as we know it did not have an administrative purposeat all, but was ratheran ideological text meant to illustrateand emphasizethe unity and cohesionof the Romanempireat a time when this unityandcohesionhad all but disappeared.3 This proposalmeritsseriousattention. The copies of the Notitiawhich we possess wereall madefroma single Speyercodex.TheNotitiaappeared in this manuscript alongsidean odd set of texts with a primarily interest antiquarian thathad probably been transmitted in one another's companysince antiquity.4
All references are to the flawed but standardedition of 0. Seeck, Notitia Dignitatum (Berlin, 1876). A et Laterculiprouinciarum acceduntNotitia urbis Constantinopolitanae andcommennew Teubneredition is expectedfromS. Irelandandan Englishtranslation Texts for Historiansseries. tary from P. Brennanin the LiverpoolTranslated The textualhistoryof the Notitia is discussedin M.D. Reeve, "TheNotitia Dignitatum", (Oxford, 1983), 253-7. The illustrauions in L.D. Reynolds,ed., Textsand Transmission are shown schematicallyin Seeck's edition and do no justice to the rich colours of the manuscript.A numberof them are shown in T. Cornell and J. Matthews,Atlas of the Latomus RomanWorld(New York, 1982), 202-3. J.H. Ward,"TheNotitia Dignitatum", theory for the originalNoti33 (1974), 397-8, propoundsan attractive,but unprovable, tia's codicological history. techniquesdans l'antiquit6 P. Brennan,"The Notitia Dignitatum",in Les litte'ratures workto romaine.Entretiens Hardt 42 (Geneva, 1995), 147-78 is muchthe most important have appearedon the Notitia since the exhaustivestudy in Appendix2 of A.H.M. Jones, TheLaterRomanEmpire(Oxford, 1964), 3.348-80. These texts were the De rebus bellicis, the Altercatio Hadriani Augusti et Epicteti and the De gradiphilosophi, the Notitiae urbis Romaeand urbis Constantinopolitanae, (as note bus cognationum. See Reeve, "Notitia" (as note 2), 253, with Brennan,"Notitia" 3), 166-9. Historia,BandXLIX13(2000) ?) FranzSteinerVerlagWiesbadenGmbH,Sitz Stuttgart

The Notitia Dignitatumas a HistoricalSource

359

These texts have in common the goal of constructing an ordered world where none necessarily existed, of imposing a shape on disordered reality. What they construct, most especially, is a united empire that encompasses the whole oikoumene and is carefully structuredby a hierarchy of officials. This almost self-delusionary outlook has been linked quite plausibly to the court of Galla Placidia, whose interest in an ideology of unity was only underscored by the circumstances in which she and her son regained the western throne in 425.5 This same ideology of unity, the same insistence on imperial strength at a time of dissolution, marks the whole era's literary output.6This new understanding of the Notitia has serious consequences for the way we approach the document. If the Notitia was not primarily an administrative document, then using it to reconstruct the history of the late imperial bureaucracy, whether civilian or military, is simply wrong-headed. We would, in that case, be asking a document that constructs an imaginary reality to inform us about things as they actually were. This, however, is precisely the way that generations of scholars have approached the Notitia. Partly because the question of the Notitia' s context has always been controversial, scholars have for years concentrated instead on the use to which it can be put as a historical source, as a mine of information on late Roman administration. The new approach to the Notitia, which sees it as an essentially antiquarian work with a more or less purely ideological purpose, would seem to invalidate this old methodology. There are problems with the new conception, however, and the old approach is not necessarily invalid. The new reading of the Notitia is at one level a matter of scholarly convenience. That is, it allows us to escape trying to understand the information preserved in our text: if the Notitia was not intended as a working document, then the contradictions in its text are not only insoluble but also irrelevant. Moreover, the new approach assumes that the Notitia Dignitatum as we have it is a composite document, made up from different notitiae produced in the bureau of the primicerius notariorum at various times.7 This assumption, which has a very long history in the older secondary literature as well, requires a
5 Brennan,"Notitia" (as note 3), 166-9, thoughhis ascriptionof the Notitia to the praetorian prefectandantiquarian Macrobius Ambrosius Theodosiusis verybold. We shouldalso note that the recent attemptby G. Purpura,"Sulle origini della Notitia Dignitatum",
Annali del Seminario Giuridico dell'Universita di Palermo 42 (1992), 469-84, to see the

Notitia as a birthday gift fromTheodosiusII to the ten-yearold ValentinianIII, while not in itself convincing, also focuses on the ideological and not the practicalpotentialof our text. This was pointed out long ago by F. Paschoud,RomaAeterna (Rome, 1967). That the same concerns prevailedat the court of Theodosius II has recently been shown by A. Gillett, "The Date and Circumstances of Olympiodorus of Thebes",Traditio43 (1993), 1-29. Brennan,"Notitia" (as note 3), passim.

360

MICHAELKULIKOWSKI

demonstration it has never received. In fact, the Notitia was in origin a single base text, divided at the time of composition into eastern and western partes. This original base text can be dated with certainty to between 386 and 394. In all likelihood, however, it comes from shortly before the latter date, and represents the state of the Roman world at the start of Theodosius' campaign against Eugenius. The whole document is therefore eastern. Its western section in origin represented the dispositions aligned against the eastern emperor before the start of the civil war. Despite these eastern origins, however, the copy we possess is western, and its eastern portion ceased to be updated after the final division of the empire in 395. Its western portion underwent continual revision until at least 419, and possibly much later. Now, as the most recent study suggests, the actual version that has come down to us may very well have originated at Valentinian's and Galla's court, in which case its ideological uses will have been to the fore. Nevertheless, an earlier version which resembled in many ways the one we possess was created at Theodosius' court early in the 390s. It too may have had an ideological purpose. Though for us the Theodosian Notitia illustrates a defacto division of the empire, its arrangementalso asserts an ideological unity. The Theodosian text therefore need not have been an administrativetool in any precise sense of the word, but was instead, like extant Placidian version, an ideological text that insisted on the unity of the empire at a time when it was quite patently divided. In point of fact, however, this question of function must remain open. Even if we admit an originally ideological impetus, however, we do not thereby throw up our hands and abandonthe Notitia's contents. An ideological document can deliberately falsify reality, or at least misrepresentit, but to do so it uses real fragments of fact. In the case of the Notitia, the building blocks of its ideology are readily accessible in its text. Those building blocks, the pieces of information out of which the Notitia was created, can in the end tell us certain things about the bureaucracyand administrationof the later Roman empire, just as generations of scholars have known. The limitations on that useable information are, however, narrowerthan most have realized. They are imposed not just by questions of ideological purpose but also by more concrete issues of textual history. As extant, the Notitia is reliable evidence for the eastern empire around 394. On the other hand, it is nowhere reliable evidence for the western empire. Because the western list contains an unknown numberof progressively overlaid changes, it cannot be subjected to precise or unitary dating, and therefore only those pieces of information confirmed by other sources can be used reliably. This interpretationis not uncontroversial and diverges in many particulars from others which have been offered. The major point of difference, however, lies in the fact that the Notitia is not generally regardedas a single base text, but ratheras two separate texts joined together some time after their initial composition. This conviction is shared both by those who see the text as primarily ideological or antiquarian, and by traditionalists who see it as a practical

The Notitia Dignitatumas a HistoricalSource

361

administrativetool. Now, it has long been recognized that nothing in the eastern half of the Notitia is post-Theodosian, while the western list is riddled with later information, most of it dating from the reign of Honorius (395-423), but some arguably pointing to that of Valentinian III (425-55).8 No amount of ingenuity can make all the information in both eastern and western lists consistent. That is to say, the text of the Notitia as extant can correspond to no single date in the history of the later empire. The obvious way out of this quandary is to view the Notitia as a cobbling together of two separate base texts, one eastern and one western, each of a different original date.9 By positing two separate base texts, the chronological discrepancies between eastern and western lists can be explained away. Each base text can be dated with greater or lesser precision and the work of mining their contents can begin. This approach is fraught with hazards, because it always requires the discarding of one or more pieces of contradictory evidence. It is also unnecessary, as there is a better way forward. If we posit a single base text that has undergone a long series of corrections and accretions in only one of its sections, all the text's anomalies are explained while none has to be explained away.

Thoughall his evidence is Theodosian,Jones, LRE(as note 3), 3.347-9, actually maintains that the extant Notitia Orientalisrepresentsa thoroughrevision of 395. D. Hoffmann,Das spatromischeBewegungsheer unddie Notitia Dignitatum(Dusseldorf, 196970), chapterone, adds greaterprecisionto Jones' conclusionson this point. Thus J.B. Bury, "TheNotitia Dignitatum", JRS 10 (1920), 131, refers to it as "the document (or rathertwo documents)which has come down underthe title Notitia Dignitatum...",a verdict accepted by F. Lot, "La Notitia Dignitatumutriusqueimperii, ses tares,sa date de composition,sa valeur",REA38 (1936), 285-338. This approach can be takenwith greaterand lesser degreesof incaution.On the one hand,there can be posited a westernbase text as late as its latest citationappendedto a substantiallyearliereastern text. Thus Hoffmann,Bewegungsheer(as note 8), 22. Anotherapproachis to ignore the easternlist, admitthatthe westerntext containstoo manyanomaliesto allow of a single, unitary date, and thereforedate it according to the bulk of its datable entries, while discardingthe rest as later corrections.Thus, E. Demougeot, "LaNotitia Dignitatumet l'histoire de l'Empired'Occidentau d6butdu Ve siecle", Latomus34 (1975), 1133: "La date de cette redactions'etabliten fonctiondes donneesr6centesles plus nombreuses, qui ne sont pas necessairementles plus recentesde toutes, celles-ci pouvantn'etre que des retouchesapporteesau documentdeja r6digV". She is followed in this approachby W. Seibt, "Wurde die NotitiaDignitatum408 von Stilicho in Auftraggegeben?",MIIEG 90
(1982), 339-46 and J.C. Mann, "The Notitia Dignitatum
-

Dating and Survival", Britan-

nia 22 (1991), 215-9. G. Clemente, La 'Notitia Dignitatum'(Cagliari, 1969) and "La Notitia Dignitatum", Attidei ConvegniLincei45 (1980), 39-49, follows more or less the same approach,though he maintainsthat both eastern and western sections accurately representthe situationof the early fifth century.

362

MICHAELKULIKOWSKI

One objection has long prevented scholars from looking at the Notitia as a single, unitary document and that is the way it portraysthe status of the Balkan dioceses. These dioceses of Pannonia, Dacia, and Macedonia, often referred to imprecisely as the prefecture of Illyricum, are divided between the eastern and western partes imperii in the Notitia. This division, it is universally argued, can only obtain after c. 396. Even though, as all admit, the contents of the eastern half of the Notitia are entirely Theodosian, the western list must date from 396 or later. Therefore, a date of unitary composition cannot be maintained. There are two problems with this conventional argument. First, the diocesan division in the Balkans is entirely consistent throughout the whole of the Notitia. This consistency in fact argues for, not against, a unitary composition. Second, there are periods before 396 when the Balkan dioceses may have been divided between east and west. The mere possibility of a division of the Balkan dioceses before 396 removes the main objection to viewing the Notitia as single base text. As we shall see, all the rest of our information confirms that hypothesis. The first point, though, is that the consistency of the division of the Balkan dioceses actually points to an originally unitary composition. The division is not only entirely consistent throughout the Notitia, but also integral to its composition. Macedonia and Dacia pertain to the east, Pannonia to the west. Nowhere in the western list do any offices with reference to Macedonia or Dacia occur; nowhere in the eastern list is there reference to offices in Pannonia. The illustrations of insignia reflect the same division. Under the eastern prefect of Illyricum, one finds depicted personifications of two dioceses only, viz. Dacia and Macedonia (Or. 3.2,3). Correspondingly, under the praetorian prefect of Italy, one finds personifications of the three great regions subject to him, Italy, Africa, and Illyricum (Occ. 2.2-4). Illyricum, in this instance, means only the six provinces of the diocese of Pannonia (Occ. 28-34), as the subsequent list makes clear. The division is borne out by the later chapters as well. All the Balkan dependencies of the west are found within the diocese of Pannonia. There are military commands for Valeria and first and second Pannonia (Occ. 32-5). Balkan fabricae subject to the magister officiorum are all Pannonian (Occ. 9.16-22), as are the fiscal servants of the comes sacrarum largitionum (Occ. 11.10-1; 21-5; 46-8) and the comes rerum privatarum (Occ. 12.20). The eastern list is exactly parallel. Under the magister officiorum, there arefabricae only in Dacia and Macedonia (Or. 11.35-9), while Balkan military commands occur only in Dacia (Or. 41, 42). Eastern and western lists are thus perfectly consistent in the picture they draw, and this consistency is unlikely to be the result of late revision. The Notitia is notorious for the inconsistency with which it records revisions, corrections, and changes of every kind. In no case of revision have the scribes of the extant text succeeded in removing all trace of previous dispositions,

The Notitia Dignitatumas a HistoricalSource

363

hence the large number of discrepancies with which the western list is littered. It is, moreover, particularly significant that the division of the Balkan dioceses is also confirmed by the illustrations, since these are especially hard to explain as late corrections.10 If the division of the Balkan dioceses were the result of correction, some discrepancy would surely be visible. None is, and their division between east and west is integral to the composition of the extant Notitia. This strongly suggests, but does not in itself prove, that the Notitia was in origin a single base text, to which a single date of composition can be assigned. The consistency of the division between the Balkan dioceses is not the only issue, however. Our second point is the date at which such a division could have been possible. The literature on the status of the three dioceses of Pannonia, Dacia, and Macedonia is vast, and tends to be referred to as the problem of Illyricum.II Two problems encumber many discussions of the topic. The first is a legal formalism that privileges theoretical models of constitutional history over the empirical analysis of political history. That is to say, the question of what ought to have happened is given more attention than the question of what did happen. The second encumbrance is the traditional means of stating the problem, which does so in terms of prefectures ratherthan dioceses. Now, dioceses were the basic building blocks of the Roman administrative system. They were created long before regional prefectures developed, and prefectures were created out of them.12If we take a prefecture of Illyricum as the normative model for Balkan administrationwe miss the point. The jurisdiction of prefects and the shape of prefectures varied according to circumstance, but the diocesan structure of the Balkans stayed unaltered from Diocletian onwards. Illyricum was a word to describe a region without juridically definite shape, a region whose boundaries changed according to circumstance. The Balkan dioceses, however, did possess juridically definite shapes. Each was a clearly defined administrative unit. None, however, possessed a necessary
10 The illustrations have produced a largebibliography, for which see Brennan,"Notitia" (as note 3), 170-1. 11 V. Grumel,"L'Illyricum de la mortde ValentinienIer (375) Ala mortde Stilicon (408)", REB9 (1951), 5-48 with referencesto earlierliterature. 12 There is a vast, and at times controversial,literature on the subject. See Jones, LRE(as
note 3), 1.373-5; T.D. Barnes, The New Empire of Diocletian and Constantine (Cam-

bridge,Mass., 1982), 209-24; K.L.Noetlichs,"ZurEntstehung der Diozesen als Mittelinstanz des spatromischen Verwaltungssystems", Historia 31 (1982), 70-81; T.D. Barnes, "Praetorian Prefects, 337-361", ZPE 94 (1992), 249-60; and J. Migl, Die Ordnungder Amter(Frankfurt, 1994), 54-68. Both Zosimus2.33 and the Notitia itself of course show the empiredivided into fourprefectures. This is not the place to enquireinto the accuracy of Zosimus' information,who is at this stage in his text following Eunapius.It is likely that both the Notitia and Eunapius' original reflected the dispositions of their own, roughlycontemporary, eras ratherthanthat of Constantine.Regardlessof this question, the extension of prefecturescould clearly vary, while dioceses, which concern us here, were administratively fixed regions.

364

MICHAEL KULIKOWSKI

jurisdictional connection to any other. It is true that, generally speaking, they were administered together, as part of the prefectureof the Italian prefect. Very occasionally, the three dioceses were administered together as a separate prefecture of Illyricum, but this situation was an anomaly that occurred only in response to specific crises. The 'problem of Illyricum' should not be posed as such. Instead, the dioceses of Pannonia, Dacia, and Macedonia must be examined in light of their separate histories. Until 379 all three dioceses were subject to the praetorianprefect of Italy, and thus to the western emperor. This changed with the disaster at Adrianople and the appointment of Theodosius as Augustus on 19 January379 with a view to restoring order to the Balkans.'3 Under Valentinian and Valens, the dioceses of Pannonia, Dacia, and Macedonia had been administered by the prefect of Italy.'4 Valens' defeat at Adrianople inspired Gratian to create a separate prefect to administer the Balkan dioceses in 378. He chose Q. Clodius Hermogenianus Olybrius.15By January380, following Olybrius' promotion to the consulate, the historian Eutropius appears as prefect of Illyricum.16 At this point, however, the three dioceses were subject to the eastern emperor Theodosius. They had probably been transferredto his jurisdiction at the beginning of the fiscal year 379, that is on 1 September, in order to help organise the Gothic
war.17

The three dioceses remained subject to the eastern emperor, under their own prefect, until just before the conclusion of the Gothic war. This war lasted longer in Thrace than it did in any of the Balkan dioceses, and so the jurisdiction of Pannonia, Dacia, and Macedonia was returnedto the western emperor at the start of the indiction of 38 1. The three dioceses had pertained to the eastern empire for just two years, during which time one finds Theodosius directing a law to the vicarius Macedoniae.19 As soon as the government of the three dioceses was resumed by the western emperor, the separate prefecture of Illyricum was eliminated and Pannonia, Dacia, and Macedonia were once again subordinated to the prefect of Italy, as surviving legislation makes clear.20
and the Goths",Chiron26 (1996), 1-27, andthe more "Theodosius 13 See R. M. Errington, detailedstudy of P. Heather,Gothsand Romans,332-489 (Oxford, 1991), 147-81. (as note 11), 6-7. 14 Grumel,"Illyricum" 15 CIL6.1714 = ILS 1271 and Aus., Grat.Act. 12.55. See PLRE1.640-1. as prefectof Illyricumis CJ 1.54.4 of 6 January 16 The earliest law addressedto Eutropius he receiveda very long series of laws in thatcapacity,the latestof which, 380. Thereafter CJ 5.34.12, dates from 28 September381. See PLRE1.317. (as note 13), 24. By 17 This is the very plausible contentionof Errington,"Theodosius" placing the date of the transferat the startof the indictionin September,one can explain the fact that Gratianwas still legislatingfor Illyricumon 5 July 379 (CTh. 13.1.11). (as note 13), 25. "Theodosius" 18 Errington, 19 CTh. 13.1.11, 27 March380. 383 addressedto PetroniusProbus,the praetorian 20 See CTh. 11.13.1, a law of 19 January prefectof Italy.

The Notitia Dignitatumas a HistoricalSource

365

Purely formalist argumentsabout the constitutionality of the Balkan administration in this period may be left out of the picture. Much effort has been expended trying to establish the precise constitutional jurisdiction of the socalled Mittelreich of Valentinian II, but this speculation is unnecessary. When Gratian and Theodosius divided the task of ruling between them, Valentinian was of no concern to them.21 After Gratian's murder and before Maximus' invasion of Italy, the Balkan dioceses remained subject to Valentinian II and his prefect of Italy.22 When Maximus invaded Italy, however, the position of the Balkan dioceses changed once again. Maximus crossed the Alps in May 387 and Valentinian fled to Thessalonica in Macedonia. Theodosius had been ready to accept Maximus so long as he confined himself to Gaul. Whatever the emperor's own feelings, Maximus was unquestionably a more useful imperial colleague than either Gratian or Valentinian. The invasion of Italy, however, could not be tolerated and Theodosius' response was immediate. His first move was to secure the Balkan dioceses, for which we find him legislating in a law which probably belongs to June 387, that is to say, only a month after Maximus had crossed the Alps.23 In doing this, Theodosius may have hoped to forestall the sort of difficulties which Constantius had faced during the revolt of Magnentius, when the Balkans were only saved for the legitimate emperor by a loyalist coup.24 At any rate, though he assumed command of the Balkan dioceses in 387, Pannonia, Dacia, and Macedonia were almost certainly administered by the western prefect of Italy after Theodosius conquered Maximus in 388. This prefect was one Polemius, twice attested in the early months of 390.25 His successor was Virius Nicomachus Flavianus, the great pagan aristocrat, who was the recipient of a long string of laws between August 390 (CTh. 9.40.13) and April 392 (CTh. 10.10.20).26 Then the usurpationof Eugenius supervened. The status of the Balkan dioceses in the years between 392 and 394 is confused by contradictory evidence. This confusion, as we shall see, is the result of contemporarypropagandaand shows that both Theodosius and Eugen21 The supposed"decision"of Gratianand Theodosiusto put an end to a "constitutionally anomalous"position and reuniteIllyricumis a scholarlyconstruct,Grumel,"Illyricum" (as note 11), 11-12. The meetingat Sirmiumto whichthis decision is ascribednevertook place: Heather,Goths(as note 13), 154 with Errington, "Theodosius" (as note 13), 25, n. 140. 22 CTh. 1.32.5, of 26 July 386, shows him legislating for the specifically namedregions of Macedonia,Dacia, Moesia, and Dardania. 23 CTh.8.4.17. Grumel,"Illyricum" (as note 11), 17 n. 1, is correctin emendingthe date of this to 387. 24 See J. Sasel, "The Strugglebetween Magnentiusand ConstantiusII for Italy and Illyricum", Iva antika21 (1971), 205-16. 25 CTh. 15.1.26, 16 January;15.1.28, 4 April. 26 PLRE1.348 for othercitations.

366

MICHAELKULIKOWSKI

ius claimed control of the Balkan dioceses by appointing competing officials to govern them. Which emperor's officials actually administered the dioceses remains somewhat unclear, but the value of the different pieces of evidence is susceptible to some analysis. Valentinian II died on 15 May 392. As recently as 4 April, a law of Valentinian and Theodosius had been addressed to Nicomachus Flavianus as PPO per Illyricum. By 28 July, the Code shows a different man, Apodemius, in that post.27 Apodemius is again named in that office on 15 February 393, and appearsas PPO Illyrici et Italiae II on 9 June 393.28 At the same time as all this, however, Nicomachus Flavianus reappears as prefect of these regions, that is, PPO Italiae, Illyrici, et Africae iterum.29There is a contradiction here which requires explanation. Most solutions have involved the emendation of one or more of our texts.30Now, the dates in the Code do frequently require emendation, but offices are usually attested more accurately. We should therefore be cautious in accepting conclusions which requirethe emendation of Apodemius' prefecture from Italy to Africa or ignoring its iteration in CTh. 11.30.51. A different solution is available. In civil wars, the loyalty of prefects was much sought after, since the adhesion of the prefect brought with it the resources of the territory he governed. Theodosius may not have felt confident in the loyalty of Flavianus after Valentinian's death. In moving to secure the Balkans, he would have appointed a new offlcial in his place. Flavianus, rather than putting down his office, simply transferredhis allegiance to Eugenius and continued on.31 This solution makes good sense of the available evidence, and also explains the apparently simultaneous tenure of Flavianus and Apodemius in the same office. The iteration expressed in Flavianus' inscription, in this scenario, simply acknowledges the transfer of his allegiance from Theodosius to Eugenius.32We cannot
27 CTh. 12.12.12: PPO Illyrici etAfricae. See PLRE 1.82-3. 28 CTh. 13.5.21, following the necessaryemendationof PLRE 1.82;CTh. 11.30.51. Apodedesignationat CTh. 12.12.13, 10 mius also appearsas prefect,withoutany geographical September392.
29 CIL 6.1783. 11.

(as note 11), 23, Hoffmann,Bewegungsheer (as note 8), 21030 E.g., Grumel,"Illyricum" 31 PLRE 1.348 and 1052 maintainsthatFlavianussteppeddown from his prefecturein 392 it afterhe decidedto join Eugenius.It is morelikely thathe neverleft and then re-entered
office.

32 J.F. Matthews,"CodexTheodosianus9.40.13 and NicomachusFlavianus",Historia 46 prefectures "Thepraetorian (1997), 196-213 effectively shows, contra R.M. Errington, of Virius Nicomachus Flavianus",Historia 41 (1992), 437-61, that the iteration of held legitimatelyunderTheodcannotreferto an earlierprefecture Flavianus'prefecture osius, because it is clearly impossible for Flavianus' quaestorship,which is securely datedto 388, to postdatethe first prefecture.

The Notitia Dignitatumas a HistoricalSource

367

be sure of how much control Flavianus actually retained in the Balkan dioceses. That same uncertainty, however, also applies to Apodemius, whose jurisdiction may have been fictitious in part. That is to say that the Balkan dioceses may have remained under the control of Flavianus.33 This confusion in the sources is itself an importantpoint. The prefectures of Apodemius and Flavianus represent propagandist impulses, of a kind with the mutual rejection of consular nominees by Theodosius and his opponent and the raising of Honorius to the purple. That is, they represent the refusal of each government to recognize the other's legitimacy, which each signalled by appointing rival officials to the same offices. The iteration of Apodemius' prefecture can be seen in the same light. The attestation of Apodemius as prefect of Italy is not an error,but a declaration of war. Theodosius reappointed Apodemius - hence the iteration - to an office in which Eugenius recognized Flavianus. No part of the evidence, then, requires emendation in order to be explained. On the other hand, no part of the evidence can be taken to faithfully reflect the actual situation on the ground in the Balkan dioceses in the years after Eugenius' usurpation. That defacto situation cannot be described with confidence. Three pieces of evidence are relevant. First, the final battle between Theodosius and Eugenius was fought at the Frigidus, a tributaryof the Isonzo on the Italian side of the Julian Alps.34 Second, Claudian describes Eugenius hiding behind the Alpine passes.35 Finally, on his way west at some date after 20 June 394, Theodosius halted at Sirmium in Pannonia and minted a very large stock of coins there.36 This evidence is inconclusive. Although it is probably safe to assume that no fighting took place east of the Julian Alps, this does not imply that Theodosius controlled the Balkan dioceses right from the start of Eugenius' usurpation.The coins minted at Sirmium are usually thought to have been minted in order to pay the army, and are thus taken as evidence for eastern control of the Balkan dioceses. This may be so, but if the coins were minted to finance the campaign against Eugenius, they should have been struck at Constantinople. We might do better to see the Sirmium issue as yet another propagandaexercise, which made public the adherence to Theodosius of a diocese that had hitherto been loyal to Eugenius' prefect Flavianus. We simply cannot tell. It is possible that Pannonia, Dacia, and Macedonia remained loyal to their then prefect Flavianus in 392 and
33 This conclusion is reached independentlyby Errington,"Flavianus" (as note 32), 444, and accepted as possible by Matthews,"Flavianus" (as note 32), 211. Africa seems to have remainedloyal to Constantinoplethroughout:CTh. 9.7.9, 30 December 393, to Gildo. 34 The JulianAlps hadbeen the geographicalfrontierbetweeneast andwest in the civil wars earlierin the century:Sasel (as note 24), 205. 35 1IICons. 89f. 36 RIC 9.160-2. For the date, 0. Seeck, Regestender Kaiser und Papste (Stuttgart,1919), 284.

368

MICHAELKULIKOWSKI

equally possible that Theodosius secured their allegiance with the appointment of Apodemius. Alternatively, the dioceses may have been split between east and west. We can only say with certainty that by some time between late May and early September 394 the westernmost of the three dioceses had been brought into the eastern camp.37 This uncertainty about the status of the three Balkan dioceses between 392 and early 394 is, however, very important to a discussion of the date of the Notitia Dignitatum. When one realises that all or some of the Balkan dioceses may have remained subject to the west in 392, the strongest objection to a unitary date of composition for the Notitia disappears. As we have seen, the division of the Balkan dioceses between east and west is integral to the structure of our document. Such a division may have existed in the first years of Eugenius' usurpation. As it happens, these are precisely the years to which the eastern half of the extant Notitia can be shown to date. The way is thus clear for us to date the initial composition of the entire Notitia to those years, with an integral structuraldivision that reflects the political realities of the time.

Before that possibility can be pursued, however, we must turn to the date of the extant eastern list, which can be determined with some precision. The internal evidence of the Notitia shows clearly that nothing in the eastern list must postdate the death of Theodosius in January395 38 Such discrepancies as there are represent either clerical errors or changes which may be post-Theodosian, but may equally well date from his reign.39The military lists, on the other hand, are
37 It seems likely that the three dioceses were subject to Constantinopleat the time of Theodosius' death. Zosimus 4.59.4 shows that Illyricumbelonged to the inheritanceof Arcadius, while Alan Cameron, Claudian: Poetry and Propaganda at the Court of the Honorius(Oxford, 1970), 60-1, enlists Claudian(Ruf.2.153f.; 304f.) to demonstrate same thing. None of this can have made much difference with Stilicho standingat the head of the unitedcomitatusof bothpartes imperii. 38 The only real obstacle to a Theodosiandate is the entry for a tabulariumdominarum in the castrensis' chapter(Or. 17.8), which mightbe takento mandatea date Augustarum after 423 when there were in fact two Augustae. The entry is better explained as a standingoffice in place for such times as therewas an Augusta,the pluralmeretradition, dating back perhaps to Constantineunder whom there were, briefly, two Augustae, Helena and Fausta. Thus Ward, "Notitia"(as note 2), 400-1, whose argumentsurely proposedby Jones,LRE(as note 3), 3.349. explains more thanthe "blunder" 39 For clerical errors in the civil lists of the east, Ward "Notitia"(as note 2), 398-408. Substantive errors are given at Ward, 408. None of them need post-date 394. The of Paphlagonia first appearsin othersourcesin July 395 (CTh.2.8.22), but correctorship the Notitia's referencemay in fact be the earliest. Likewise, MacedoniaSalutarismay already have been dissolved by 394. If the 394 date is correct, the argumentsof T.D. and the Notitia Dignitatum", Phoenix32 (1978), 81-2, for a division Barnes,"Claudian of Galatiaby Eutropius, cannotbe accepted.Thereis no positive reasonthatthe division

The Notitia Dignitatumas a HistoricalSource

369

certainly Theodosian.40 Finally, the officia of three of the eastern field armies are clearly temporary, since they are administered by officia drawn from the army, and not, as normally, the civil service: Officium autem magisteriae ... potestatis in numeris militat et in officio deputatur.4I By contrast, the second praesental magister and magister per Orientem are served by bureaux drawn from the civil service: Officium autem magisteriae per Orientem potestatis cardinale habetur.42If the difference indicates a distinction between campaign armies and field armies on regular service, a context is suggested by the preparationsfor the campaign against Eugenius. To reiterate, nothing in the eastern lists necessitates a date later than 394, and the officia of the field armies certainly suggest a campaign force. A terminus ante quem for the list is thus established by the departureof Theodosius' army from Constantinople.43It had left by 20 May 394."4 Even those who accept this date as the terminus ante quem for the bulk of the eastern list have felt that Or. 9, the army of the magister militum per Illyricum, must postdate it.45 Part of the reason for this is the assumption that Pannonia, Dacia, and Macedonia cannot have been separatedfrom each other before 396. As we have seen, this need not be the case, and the division may have existed in 392. There is another problem, however. The Illyrian list contains a correction which must have been made after the other eastern lists were first redacted. However, this fact does not make the list significantly later than the other oriental lists and it certainly need not be post-Theodosian. It has been observed that when, in the military lists of the Notitia, identically named units appear in more than one list, the phenomenon is to be explained as a troop transfer.46
of Galatia should not be Theodosian.The hierarchyof honorati also provides no firm dates.Jones,LRE(as note 3), 3.349, rightlydismisses the questionof the rankof thepraepositus sacri cubiculi. His position in the Notitia directly beneathprefects and magistri may have been attainedas early as 385. Otherquestions of rank, as discussed by Demougeot, "Notitia"(as note 9), 1083-93, are not probative. Jones,LRE(as note 3), 3.347-8; Hoffmann,Bewegungsheer (as note 8), 22. Units named afterTheodosius, Honorius,and Arcadiusoccur in an orderwhich can have arisenonly before the death of Theodosius and the beginningof the cold war between eastern and westernempires.On the otherhand,thereare only five Arcadianunits, too low a number to date very far into his own reign, especially when contrastedwith the western lists which contain 17 Honorianunits. Or. 5.67; 8.54; 9.49. Or. 6.70; 7.59. Hoffmann,Bewegungsheer(as note 8), 52. Seeck, Regesten(as note 36), 284. If the dioceses of Pannonia,Dacia, andMacedoniaare held to have been divided between east and west only after 395 then even those who accept a Theodosiandate for the other eastern army lists are forced to see Or. 9 as a post-395 addition. Thus Hoffmann, Bewegungsheer(as note 8), 19-21, followed by Demougeot,"Notitia"(as note 9), 1094. This is one of Hoffmann'smost originaland important observations,at Bewegungsheer

40

41 42 43 44 45

46

370

KULIKOWSKI MICHAEL

Thus, when the Illyrian Equites sagittarii seniores (Or. 9.19) reappear in the Thracian list (Or. 8.30), one may confidently postulate a transfer. Either the Thracian or the Illyrian list has therefore been slightly corrected after the initial redaction of the eastern list as a whole. It is marginally more likely that the Thracian list is earlier than the Illyrian rather than the reverse, and we may assume for the sake of argument that the Illyrian list preserves a correction recording a troop transfer.47Regardless, the Equites unit is certainly Theodosian or earlier.48 This alone does not show the Illyrian list to be Theodosian, but other factors do. In the first place, Or. 9 describes a campaign army, an observation confirmed by the fact that it lists nine units of pseudocomitatenses, or units raised from the garrison army into the comitatus.49 This is an unusually high number, and eight of the nine units are certainly of Theodosian origin.50 The army shown in Or. 9 is thus entirely consonant with preparationsfor a major campaign. This evidence cumulatively suggests that the Illyrian list is nearly contemporary with the rest of the eastern Notitia. Better put, it is precisely contemporarywith the rest of the Notitia orientalis save that it contains a single correction not registered elsewhere. The terminus ante quem for the eastern Notitia is therefore 20 May 394. A terminus post quem is much harderto discover. It has recently been customary to fix it in 392 on the basis of the eastern Notitia's military lists. The argument runs as follows. Troops which must have been western before the defeat of Maximus in 388 appear in the eastern lists.5' Their incorporation into the eastern army must therefore post-date 388. Since Theodosius did not returnto Constantinople with the eastern army until July 391, it must in fact date to that year, while some minor points probably refer to 392.52 The logic is impeccable, but it is founded on error. The argument assumes that units with the byname seniores must be western in origin, those with the byname iuniores eastern. One may thus trace troop movements between east and west in the distribution of
(as note 8), 28. Earlieropinionhad inclinedtowardsseeing these duplicationsas separate units bearingthe same name,or, at best, as new units formedout of old ones. The Equitessagittariisenioresappearin Thracewith theirsister unit, the Equitessagittarii iuniores (Or. 8.31), but alone in Illyricum,so their stationingin the Illyrianarmy is likely to be the laterof the two. This must be the case because it appearsin the Thracianlist aheadof a definitely Theodosian unit, the EquitesprimiTheodosiani,Or. 8.32. Or. 9.40-8. or the Thracian units in the firstpraesental Or.9.41-8. Thereare no pseudocomitatensian armies, only one in the second praesentalarmy (Or. 6.69), though there are eleven in Oriens(Or. 7.49-58). (as note 8), 488. Hoffmann,Bewegungsheer (as note 8), 490-506. The 392 date is derivedfromsubsidiary Hoffmann,Bewegungsheer arguments,Hoffmann,516-9.

47

48 49 50

51 52

The Notitia Dignitatumas a HistoricalSource

371

units with this nomenclature.53This assumption is demonstrably incorrect. The bynames seniores and iuniores do not originate, as this argument requires, with a division of the comitatus by Valentinian and Valens in 364, but are in fact attested already under both Constantius II and Julian.54An archaeology of the troop movements enshrined in the Notitia cannot be conducted on the basis of this assumption."5Only units attested independently in outside sources can be claimed with certainty as belonging at any point to eastern or western armies. If the seniores-iuniores criterion is discarded, units often identified as having been transferred from west to east in 388 may in fact have belonged to the eastern army for many decades before. The 388 terminus post quem thus disappears, and with it the derivative date of 392.
53 Hoffmann,Bewegungsheer(as note 8), chaptersfive and eight, is devoted to the topic. His conclusions are acceptedby Demougeot,"Notitia"(as note 9), 1094. The same conclusion was reachedindependently by R. Tomlin, "Seniores-lunioresin the late Roman field army",AJPh 93 (1972), 253-78, and restatedin his review of Hoffmann,JRS 67 (1977), 186-7. 54 T. Drew-Bear, "A Fourth-century Latin soldier's Epitaph from Nakolea", HSCP 81 (1977), 257-74. The epitaph belongs to one Flavius Aemilianus, a ducenarius in the Cornutiseniores, who died DDNN ConstantiVIIIet Juliani c(on)s(ulatu),that is, in 356. This clearlyshows thatthe bynamesseniores and iunioresexisted alreadyin 356 andmay well date back still further.For HerculianisenioresunderJulian,see D. Woods, "AmmianusMarcellinusandthe deathsof BonosusandMaximilianus", Hagiographica2 (1995), 31-7. 55 Hoffmann,Bewegungsheer (as note 8), 117-30, andTomlin,"Roman Army"(as note 53), 259-60, maintainnot only thatthe seniores-iunioresdivision dates back to 364, but that in thatyear seniores were allocated-tothe westernarmywith the senior brother,iuniores to the eastern with the junior. For Hoffmann,moreover,the iuniores representedthe inferiortroops,the seniores the superior.This, in his view, helps accountfor the annihilation of the eastern army at Adrianople.The distinction, however, is most valuable to Hoffmannas a tool of textual archaeology.Any occurrenceof seniores in the east and iuniores in the west must be accountedfor by specific, and thus determinable, historical circumstances.Trooptransfersare carefullyplotted,the originalpostings of units excavated, and hypothesis is stacked on hypothesis, until it is concluded that certain units which appearas easternin the extant Notitia must until 388 have been western. In this way 388 becomes the terminus post quemfor the easternNotitia. Nakolea, however, is in the east. Aemilianusserved in the Cornutiseniores, who were stationed there before 356 (Drew-Bear, "Epitaph",[as note 54], 269). The assumed connectionbetweenseniores units and the westernarmycannotstand.The result undermines nearly the whole of Hoffmann'swork. Almost every conclusion in his chapters eight, nine, and ten, no matterwhat immediatelogic it is basedon, ultimatelyturnsupon the 364 army division. Nearly every conclusion is thus suspect, though many are not implausible.None, however,is susceptibleof proof.Thereareequally graveconsequences for Hoffmann'schapterssix and seven, on the comitatusthroughthe reign of Julian. Forheremanyunitsareexcludedfromconsideration on the basis of the seniores-iuniores nomenclature. The direstconsequenceis reservedfor the 388/391 terminuspost quem.It is foundedon a reconstruction basedon a false premise.Withoutthatfoundation,it stands wholly unproved.

372

MICHAELKULIKOWSKI

One is therefore left to establish a terminus post quem for the eastern list from other sources. In the civil portions there are some indications of date, the two firmest of which are the inclusion of the provinces of Honorias and Arcadia.56The province of Honorias in Pontus, which appears in the list, was founded between 384 and 387.57 The province of Arcadia in Egypt, which also appears in the list, was founded after February 386.58 By this same year there were also two provinces of Cappadocia which the Notitia, too, describes.59The earliest terminuspost quem one can fix, therefore, is 386. The military lists tend to confirm the picture, since Honorius was born in 384 and units named after him appear in the eastern Notitia.60Some argumentshave been ventured on the basis of the nomenclature of command, but its evidence is ambiguous.6' The terminus post quem of the eastern Notitia must therefore remain 386. We can, however, say with certainty that the eastern Notitia dates from between 386 and May 394.

The western list will have originally been composed at the same time. Because the division of Pannonia, Dacia, and Macedonia between east and west is integral to the structureof the whole Notitia, both its eastern and western halves should be presumed to have a common origin unless the contrarycan be proved. The common origin has been denied on the grounds that the division of the Balkan provinces cannot have obtained before the death of Theodosius. As we have seen, there might very well have existed just such a division in the years between 392 and 394. The Notitia in this context should be viewed as a document drawn up during the preparationsfor Theodosius' campaign against Eugenius. It was probablyjust a new draft of the sort of document always kept in the bureau of the primicerius notariorum so he could draw up codicils of office.62 It need not, however, be in any way representative of other notitiae
56 Ward,"Notitia"(as note 2), 411-3, lists the points which he thinksdate the easternNotitia to between 388 and 395. Apartfromthe cases of HonoriasandArcadia,his citations are suggestive ratherthanprobative. 57 Libanius,Or. 19.62. See Jones, LRE(as note 3), 3.348.
58 59 CTh. 1.14.1. CTh. 13.1.11 and Or. 1.105-6.

(as note9), 136, thereis no 60 As Jones,LRE(as note 3), 3.348, notes contraBury,"Notitia" justificationin thinkingthatsuch unitsmustreflectthe periodafterhis becomingAugustus. 61 Hoffmann,Bewegungsheer(as note 8), 496-8. The argumentturnson the discrepancy betweenthe easterngenerals,who are magistriutriusquemilitiae, and the westernones, who are magistripeditumand magistri equitum.Attestationfrom outside the Notitia is contradictory,and there is no sure way of determiningwhy two different command systems should have existed simultaneously. of the use to which the Notitia's chapterswere put, based on Bury, 62 This interpretation "Notitia"(as note 9), 131-3, is convincinglyrevived by Brennan,"Notitia"(as note 3),
150-3.

The Notitia Dignitatumas a HistoricalSource

373

which preceded or followed it. Because there is no comparative evidence one way or another, the version of the Notitia we possess must be assumed to be unique in all its details for all purposes of argument. We may, however, propose a reconstruction of the circumstances that surrounded the genesis of our version. In May 392 Valentinian died. Theodosius realised that accepting the fait accompli could not reflect well on him, since the death of Valentinian had done his own dynasty a service. He temporised anyway. The delay forced the hand of Arbogast, who proclaimed Eugenius Augustus. Theodosius made no open show of hostility, but set about preparing for war. He appointed Apodemius to the prefecture of Illyricum, but this was in part a fiction. The diocese of Pannonia remained loyal to the praetorianprefect of Italy, an office held by Nicomachus Flavianus who had been reappointed to that post by Eugenius. Dacia and Macedonia, however, were made subject to Constantinople, and it is this situation which is reflected in the Notitia Dignitatum. This was set down at some point between 392 and 394. Its eastern half showed the campaign army recently organised by Theodosius, and included many pseudocomitatenses, perhaps raised as a counterweight to the Pannonian units which had adhered to the west. The western list showed the dispositions of the western empire as they were imagined by the eastern court. The army described in the eastern half of the Notitia set out from Constantinople by May 394, and Theodosius advanced as far as Sirmium unopposed. At Sirmium the emperor halted to strike an issue of coins. These had the dual function of paying the troops and advertising the resumption of Pannonia by Theodosius. Possibly at this point, the Equites sagittarii seniores were transferred from the command of the Thracian magister to that of the magister per Illyricum. If any battles were fought between the stop at Sirmium and the confrontation at the Frigidus, they are not recorded. On 6 September, Theodosius was victorious and Eugenius executed. Arbogast killed himself on the 8th. In the same month, Theodosius moved on to Milan. Henceforth, the whole comitatus was united under the command of the emperor himself in north Italy. In such circumstances, the de iure position of the three Balkan dioceses hardly mattered, but they were perhaps administered by the prefect mentioned in a funerary inscription from Concordia.63 The army remained together near Milan at the time of Theodosius' death and continued thus under the command of Stilicho until the revolt of Alaric forced him to lead it to Illyricum. Having gone out to suppress Alaric, Stilicho was greeted with the open hostility of the eastern court.64Arcadius, that is to
63 PLRE 1.1006 (Anonymus 10). The prefect,noted in the epitaphof anotherbureaucrat, is anonymous,but there is no reasonthathe should not be identicalwith Apodemius. 64 On this see Cameron,Claudian(as note 37), 156-88, and Heather,Goths (as note 13), 193-213.

374

MICHAELKULIKOWSKI

the return of the easterncomitatus and the breaking say Rufinus,commanded off of the campaign.Alaric was thus granteda reprieve,while Stilicho was command of the armywhichmightotherwisehave allowed forcedto relinquish him to act the arbiter of the whole empire.He did as he was asked,for he could of both his claimto be actingin the interests not do otherwiseandstill maintain brothers. imperial Despitethat,however,he keptbackfor himselfthe best units of the duplicaof the easterncomitatus.65This no doubtaccountsfor a number tions betweenthe easternandwesternlists of the Notitia,as easternunitswere to Italy with the Stilicho then returned incorporated into the westernarmy.66 of Ruflnusby Gainas'troops westerncomitatus. The murder now-strengthened successor,the led to an ententebetweenStilichoandthe magisterofficiorum's saw the diocese of It is likely thatthis rapprochement praepositusEutropius. into the westernempire,becomingagain subjectto Pannoniare-incorporated the Italian prefecture,while the eastern dioceses of Dacia and Macedonia remained as they hadbeen since 392.67 subjectto Constantinople

WithStilichobackin Italy,the originalof the Notitiawhichwe possess would have servedits purpose.The campaignagainstEugeniushad long since been
65 Zosimus 5.4.2. 66 Thus Hoffmann,Bewegungsheer(as note 8), 25-39. He attemptsto distinguishbetween to the western those units, duplicatedin easternand westernlists, which were transferred comitatusin 395 by Stilicho, and the six units sent to Honoriusby TheodosiusII in 410 (Zosimus 6.8.2; Sozomen 9.8.6). These dates, he determines,are the only two times between 394 (his correct terminusante quem for the eastern Notitia) and 423 (his mistakenterminusante quem for the westernlist) at which a transferfrom east to west would have been possible given diplomaticrelationsbetweenthe two partes imperii.One His sortingprocessdepends may agree with thatjudgement,thoughwith nothingfurther. upon the Illyrianlist (Or. 9) being laterthan395, in which case the Ascarii seniores and iuniores, which appearin the Illyrianlist's (Or. 9.24-5) and then in Spain (Occ. 7.11920), must belong to the transferof 410. But, as shown above, the Illyrianlist need not be later than 394. The selection criterionis thereforevalueless. Hoffmannalso contradicts transfers,when he denies that his own dictum,p. 28, thatduplicatedunits mustrepresent the Equitessagittariiseniores which appearin Thrace(Or. 8.30) and Illyricum(Or. 9.19) are identical with the Equitessagittariiseniores which appearin Tingitania(Occ. 6.84 = 7.208). The consequenceis thatof the seventeen(not sixteen, Hoffmann,26) units which appearin both easternand westernlists, not a single one can be fixed with certaintyto One may agreewith some of his hypotheses,that, eitherthe 395 or the 410 trooptransfers. for instance, crack units would not have been sent willingly in 410 and thus the Schola gentiliumseniorum,Cornutiiuniores,Bracchiatiiuniores,and Bataviseniores went west in 395, but this remainsspeculative. cannotbe statedwith precision.The evidence of Claudian,Theod. 67 The date of the transfer 201-2 is probably valid only for the time of composition, and so cannot be used to postulate a transferbefore Theodorus' prefectureof 397. See T.D. Barnes, AJPh 96 (1975), 419.

The Notitia Dignitatum as a HistoricalSource

375

won, and the de facto partition of the Balkan dioceses was now officially recognized as well. The document's basic structurealready reflected the situation which now became a lasting reality. There was therefore no reason for any structuralrevision if the western list was to continue to function as a practical document. If, on the other hand, the original purpose of the text had been ideological, it retained that function without requiring redrafting. The eastern section fell into abeyance during the protractedcold war between the two partes imperii which began as soon as Eutropius proved less pliable an ally than Stilicho had hoped he would be. The extant Notitia Orientalis therefore still reflects the situation at the time of Theodosius' death. The western list, however, began its long career of progressive revision. This revision was haphazard, and perhaps halfhearted if the document had more symbolic than practical value. The most obvious alteration is the appearance of the Distributio Numerorum (Occ. 7), an intrusion into what was originally the chapter of the magister equitum per Gallias which has no parallel in the eastern lists. There are also many internal contradictions on account of changes not having been carried out systematically. The most famous example of this is the case of Valeria in Pannonia, which does not appear among the provinces of Illyricum under the prefect of Italy (Occ. 2.28-34), even though a chapter devoted to it (Occ. 33) occurs later in the western list.68This example may stand for many others, but confusion is also very commonplace in the army lists. Hence the many limitanei upgraded to the comitatus appear in both the distributio numerorum, and in their original chapters. There are countless other changes, the latest of which date from July 419 or later.69No other changes need date later than this, though some may. Finally, there is no indication that any systematic effort was ever made to delete the record of obsolete offices or of destroyed or disbanded units. The state of the text of the extant Notitia Occidentalis, in conjunction with our analysis of how it got to be the way it is, has momentous consequences for our use of the western Notitia as a source. To a very real extent, it becomes almost worthless as a mine of precise information on the history of the late
68 As Jones,LRE(as note 3), 3.351, showed,this errorwas clerical,causedby a confusionof the Italianand Pannonian Valerias.Before Jones, it was long takenas confirmationfor a statementof MarcellinusComes (Chron.min.2.76) on the Hunnicoccupationof Pannonia: 0. Seeck, "Die Zeit des Vegetius",Hermes 11 (1876), 61-83. 69 ValentinianIII was bornon 2 July of that year, and this is the earliest date at which the units namedafterhim, e.g., the Placidi Valentinianicifelices (Occ. 7.36) and the Equites constantesValentinianenses(Occ. 7.165), could have been formed. (That Valentinian's fatherConstantiusonly became Augustusin 421 has no bearingon the question, since Valentinian'sdynasticclaims passed throughhis motherGalla.) It should be noted that these units do not allow Occ. 7 to be "precisamente datata" (Clemente"Notitia"[as note 91, 48), since it merely providesthe date of the latest correction.OtherValentinian-units (Occ. 7.47;61;71), areas likely to have been called afterValentinianII as ValentinianIII.

376

MICHAEL KULIKOWSKI

Roman army and administration. As a base text with a large number of later alterations, it can provide no foundation for hypothesis, because there is no way to distinguish original material from later correction or interpolation unless a specific entry is confirmed by outside evidence. Yet even where such confirmatory evidence exists, it can be valid only for each individual item to which it pertains. Extrapolations cannot be made with confidence, neither within the Notitia nor to sources outside it. As is usual, the army lists provide the best examples. The armies of all the western provinces, especially those of Gaul and Britain, but also those of Africa and Spain, show a great many disturbances in the Notitia. Many of these changes have been explained with reference to other sources for the political history of the empire. Thus, with a great deal of effort and considerably more speculation, we have plausible accounts of the way in which the Roman armies in Britain and Africa arrived at the state in which the Notitia portrays them.70These exercises have merit insofar as they illustrate the way in which outside sources can illuminate passages of the Notitia itself. But they are also misleading, in that they create the false impression that the Notitia is a tractable source for the fourth- and fifth-century west. It is not, for the many reasons outlined in this paper, and bringing it into the discussion of historical questions almost inevitably produces a circular argument. Herein lies the great problem with using the western Notitia as a source for the history of the west in the fourth and fifth centuries. The issue goes beyond the question of practicality or ideology, and back to the textual history of our extant document whatever its purpose. If it is useable only where it can be dated, and if it can be dated only in those precise cases where external evidence duplicates the information it provides, then it is not in fact useable. The Notitia's most perceptive interrogatorthought that the eastern list provided a fairly accurate picture of the eastern empire at the end of Theodosius' reign.7' In this he was correct. He thought, however, that the western list did the same for its subject around the death of Honorius, and in this he was mistaken. The western half of the Notitia has long been drawn upon, section by section, to confirm or deny any number of arguments on which it has some potential bearing. This employment, however, has rarelytaken account of the larger issue of its composition and date as a whole. When that is examined, it becomes clear

70 For Britain,Demougeot "Notitia"(as note 9), 1115-20; for Africa, Ward,"Notitia"(as note 2), 432-3. The Notitia's Spanish army has never been adequately treated: all accounts have hithertobeen vitiated by a misconceptionabout the comes Hispaniarum (Occ. 7.118), who is not attestedoutsidethe Notitia till 420 (Hyd. 66, ed. Burgess).This to see the Spanishcomitatus(Occ. 7.118-134) as laterthanthatyear. has caused inquirers The fact is, however,thatthe Notitiamaycontainthe earliestmentionof the office, which can in thatcase date to any time between392 and 419. 71 Jones, LRE(as note 3), 3.355.

The Notitia Dignitatumas a HistoricalSource

377

thatno piece of the westernlist can fairlybe usedin isolation- andthattakenas partof a whole, each piece loses muchof its potentialutility.72 Washington andLee University,Lexington,Virginia MichaelKulikowski

72 I should like to thank ProfessorsT.D. Barnes and Walter Goffart for a great deal of valuableadvice.