HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
The Crossing of the Danube
and the Gothic Conversion
Peter Heather
A MMIANUS MARCELLINUS provides a detailed account of the re
lations between the Emperor Valens and the Goths during the
period 367-378. But essentially because Ammianus does not
mention it, there has been much controversy over the date of a
Gothic conversion to Christianity ascribed in other sources to the
reign of Valens. Equally, because the historians Socrates and So
zomen link a civil war among the Goths to the conversion, it has also
been unclear when this split might have taken place. It will be argued
here that the primary accounts found in Socrates, Sozomen, and
Eunapius can be reconciled with the secondary ones of Jordanes,
Theodoret, and Orosius to suggest a Gothic conversion in 376. Fur
ther, combined with Ammianus, they strongly indicate that Christian
ity initially affected only elements of one Gothic group, the Tervingi,
and was part of the agreement by which Valens allowed them to cross
the Danube and enter the Empire in 376. It also becomes clear that
the split too affected only the Tervingi, and occurred immediately
before the crossing and conversion.
This reconstruction in turn highlights the Huns' role in overturning
the established order in Gothic society: their attacks first divided the
Tervingi, who were unable to agree on an appropriate response, and
prompted the larger group to seek asylum in the Empire and accept
conversion to Christianity. While these conclusions emerge from a
detailed examination of the primary sources, this study is concerned
as much with history as with historiography, and it is first necessary
to place the discussion in the context of relations between the Empire
and the Goths in the fourth century.
I. Introduction
By the mid-fourth century, the Goths were the most dangerous
of the Roman Empire's neighbours north of the lower Danube fron
tier. Two large Gothic confederations, the Tervingi and Greuthun-
289
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
290 THE GOTHIC CONVERSION
gi,l are attested for the period prior to ca 375. The Tervingi, who
inhabited the territories of modern Moldavia and Wallachia, were
closest to the Danube frontier and attracted considerable imperial
attention. In the 330's the emperor Constantine pacified the Tervingi
and instituted close formal ties with them. In this settlement the
Tervingi were perhaps made legally part of the Empire, provided
troops when required for imperial campaigns (for whose upkeep the
Romans made payments), and were granted special trading rights.
The frontier was open for commerce, despite the normal official
policy of controlling cross-border traffic through a small number of
designated outlets.2 These arrangements lasted for about thirty years,
not without problems. In 347/8, for instance, there was a diplomatic
crisis during which a large number of Gothic Christians fled south of
the Danube under Ulfila. This does not seem to have occasioned any
major readjustment in the established form of the relationship.3 In
the reign of Valens, however, Constantine's accord with the Goths
fell apart. From 367 to 369 the emperor fought against the Goths
north of the Danube led by the Tervingi ruler Athanaric (who also
seems to have received aid from the Greuthungi), until hostilities
were ended by the Treaty of Noviodunum, signed on a ship in the
middle of the river. This reversed Constantine's settlement and broke
the close ties between Goths and Empire. Trade was restricted to two
centres, and payments in return for troops were discontinued.4
The mid-370's saw further changes in Valens' relations with the
Goths. Probably for some time, Hunnic invasions from the east had
posed an increasing threat to Gothic security; ca 375 a Greuthungi
king fell in battle and his followers moved west to escape their
troubles, triggering a chain reaction that completely altered the strate
gic situation north of the Danube. In 376 two Gothic groups came to
1 The common equation of these, respectively, with 'Visigoths' and 'Ostrogoths' is
anachronistic. The question cannot be dealt with here, but the reader should note that
s~~ndary authorities refer to Tervingi as Visigoths, and Greuthungi as Ostrogoths.
2 General accounts of Constantine's relations with the Goths can be found in L.
Schmidt, Geschichte der deutschen Stiimme his zum Ausgang der VOikerwanderung: Die
Ostgermanen2 (Munich 1934) 225ft"; E. A. THOMPSON, The Visigoths in the Time qf
Ulfila (Oxford 1966 [hereafter 'Thompson']) 3ft"; E. K. CHRYSOS, To Bv(avrwv Kal. oi
fOT90, (Thessalonica 1972 [hereafter 'Chrysos']) 41ft", and "Gothia Romania: zur
Rechtslage des FOderatenlandes der Westgoten im 4. Jahrhundert," Daco-Romania 1
(973) 52-64; H. Wolfram, Geschichte der Goten (Munich 1979) 60ft".
3 Philost. HE 2.17, with Chrysos 77ft", contra Thompson 13ft".
4 For the war of the 36O's the main texts are Amm. Marc. 27.5 and Them. Or, 10;
for commentary see Schmidt (supra n.2) 232f; Thompson 13ft"; Chrysos 103ft"; Wolfram
(supra n.2) 74f; all agree that the 369 treaty separated Roman and Goth, although
there is some disagreement over which side wanted this separation.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 291
the river asking for asylum within the Empire: one, the Greuthungi,
the other, the larger part of the Tervingi, who had split from Atha
naric and were now led by Alavivus and Fritigern. Valens responded
by admitting the Tervingi but excluding the Greuthungi, who man
aged an illegal crossing to unite with the Tervingi. In Valens' second
Gothic war the emperor and his army were eventually destroyed at
Hadrianople in August 378 (Amm. Marc. 31.3ft). Valens' reign thus
saw three major phases in relations with the Tervingi: the close ties
inherited from Constantine were broken by the first Gothic war after
367; a period of separation followed; and in 376 the Tervingi were
brought into unprecedentedly close contact with the Empire by the
Danube crossing.
The primary sources that report a conversion in the reign of Va
lens give differing accounts of its date and place. Orosius (Adv.pag.
7.33.19) dates a conversion to Valens' reign but does not say pre
cisely when. Socrates also records a Gothic conversion in Valens'
reign but places the action more precisely north of the Danube and
within the context of a civil war between Athanaric and Fritigern,
before the former's persecution of Gothic Christians beginning after
the treaty of Noviodunum in 369.5 For 10rdanes and Theodoret,
however, conversion was part of the agreement giving the Tervingi
legal entry into the Empire in 376.6 Sozomen largely copied Socrates'
account (specifically the sequence of civil war, conversion, and perse
cution), but located the action south of the Danube before the sec
ond Gothic war that led to Hadrianople. This, of course, would mean
376.7 Finally, there have been attempts to introduce into the argu
ment a fragment of Eunapius describing the crossing of some 'bar
barians' into the Empire, although Eunapius' editors have been unani
mous in taking this as a reference to an event in the reign of Theo
dosius 1.8
This disagreement has led to uncertainty over when, if at all, a
conversion should be dated in the different phases of Valens' rela
tions with the Goths. Four positions have been adopted. One group
has followed the account of Socrates, in which civil war leads to
conversion. Given that this takes place before Athanaric's perse-
5 Soc. HE 4.33 (ed. R. Hussey [Oxford 1853] 559-61; for the date of the persecution
see Hieron. Chron. s.a. 369 (ed. R. Helm, GCS [Berlin 1956] 245).
6 Jord. Get. 25.13lf (ed. Th. Mommsen, MGH AA VI [Berlin 1882) 92); Thdt. HE
6.37 (ed. L. Parmentier and F. Scheidweiler, GCS [Berlin 1954] 2730.
7 Soz. HE 6.37 (J. Bidez and G. C. Hanson, edd., GCS [Berlin 1960] 294-97).
8 Eunap. fr.55 MOLLER = fr.48 R. C. BLOCKLEY; cf. their respective comments at
FHG IV (Paris 1868 [hereafter 'MUlier']) 38f and The Fragmentary Classicising Histori
ans of the Later Roman Empire I (Liverpool 1981 [hereafter 'Blockley']) 104f, 160 n.56.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
292 THE GOTHIC CONVERSION
cution, the action must belong to Valens' first Gothic war, between
367 and 369.9 The second group uses Socrates' basic account but,
because of serious problems that arise when it is accepted in full,
considers it to be chronologically confused. On this understanding,
Athanaric's persecution should be placed before, not after, the con
version; the other events (including the conversion) fall between the
persecution and the crossing of the Danube, co 372 to 375.10 The
third group has followed Theodoret and 10rdanes in dating the con
version to 376, interpreting Orosius' ambiguity and the accounts of
Socrates and Sozomen as confused descriptions of the same event. l1
Finally, some scholars have considered the evidence for a conversion
in the reign of Valens too confused to be conclusive. They place
more weight on references to Goths as pagans after Hadrianople,
and suggest that conversion took place sometime in the reign of
Theodosius 1.12
The date of the conversion has great historical importance, for it
establishes the circumstances of the mass penetration of the Danube
frontier by the Goths in 376, leading up to the battle of Hadrianople.
It decides whether the Tervingi who fled south of the Danube in 376
had already been involved in a deep split, leading to a Christian con
version and a client relationship with the Emperor Valens, or wheth
er this was all part of the disorder associated with invasion by the
Huns. These events, indeed, have often been seen as heralding the
collapse of Roman power, so that no apology is necessary, especially
given previous scholarly disagreement, for returning again to the
complex problems surrounding the Gothic conversion.
On one level a single date for the conversion of a people does not
make sense. Conversion is a process, not an event, and takes time to
come to fruition. Indeed, before the reign of Valens, the Goths and
especially the Tervingi had long been exposed to Christianity from a
number of sources, and it is clear that the new religion had a consider-
9 Chrysos 109-24; L. Varady, Dos letzte lahrhundert Pannoniens 376-476 (Amsterdam
1976) 27ff; K. K. Klein, "Frithigem, Athanaric und die Spaltung des Westgotenvolks,"
Sudosrforschungen 19 (960) 37ff.
10 K. Schaferdiek, "Der germanische Arianismus. Erwagungen zum geschichtlichen
Verstlindnis," in D. Baker, ed., Miscellanea Historiae Ecclesiasticae III (= Bibl. RHE 50
[Paris 1970)) 76f, and "Zeit und Urnstlinde des westgotischen Ubergangs zum Christ
entum," Historia 28 (1979) 9Off, followed by Wolfram (supra n.2) 73ff, and (with
amplification) Z. Rubin, "The Conversion of the Visigoths to Christianity," MusHelv
38 (1981) 34ff.
11 1. Zeiller, Les origines chretiennes dans les provinces danubiennes (= BEF AR 112
[1918]) 452ff.
12 Thompson 86ff; A. Fridh, "Die Bekehrung der Westgoten zum Christentum," in
U. E. Hagberg, ed., Studia Gotica (Stockholm 1972) 130-43.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 293
able impact.13 A single date can have meaning, however, where it
marks an intention, often on the part of the leadership of a group, to
advance the process of conversion consistently. Such a date refers not
to a group adherence body and soul to a new set of beliefs, but marks
rather a determination to change public practice. A later example of a
date with similar Significance in Gothic history would be the rejection
of Arianism by the Visigothic king Reccared in 587: this was a single
decision that committed the Visigothic leadership to orthodoxy, but
followed a long process in which catholic inroads into Gothic Arianism
had made it a practical and even necessary course to choose.14 De
tailed examination of the sources will indicate that in 376, similarly,
the leadership of the Tervingi committed itself to changing the prac
tice of the group and adopted the so-called 'semi-Arian' religion of the
Emperor Valens in place of their ancestral paganism.
II. The Greek Sources
Socrates
Socrates provides a clear account of a partial Gothic conversion that
has been accepted in full by the first group of scholars:
(a) The Goths are divided by civil war, one group led by Friti
gern, the other by Athanaric (4.31.1).
(b) Athanaric gains an advantage and Fritigern flees to the Ro
mans to request their help. Valens orders troops in Thrace to
assist Fritigern; they are victorious over Athanaric north of the
Danube (4.33.20.
(c) Fritigern adopts Valens' religion out of gratitude and urges his
fellow countrymen to do likewise. As Valens was an Arian, the
Goths became so and have remained so (4.33.40.
13 Most famous are the missions of Ulfila, on which see Philost. HE 2.5; c;{. also the
letter of Auxentius printed in F. Kauffmann, Aus der Schule des Wu(ft/a. Texte und
Untersuchungen zur altgermanischen Religionsgeschichte I (Strasbourg 1899) 73ff, with
Thompson xiiiff for commentary. Other missions were carried out by the heretical
bishop Audius, who even ordained bishops (Epiph. Haer. 70). Orthodoxy was repre
sented by Cappadocian missionaries such as Eutyches, mentioned in Basil Ep. 164, and
the Goth St Saba (for the latter c;{. H. Delehaye, "Saints de Thrace et de Mesie,"
Ana/Boll 31 [1912] 216-21). While a different conversion date will be proposed, I
would accept much of the picture of Gothic society ca 375 painted by Thompson (64ft)
and Rubin (supra n.10: 36-41): that is, of a people coming increasingly under Roman
cultural influence, in which Christianity was a major element. The strength of the
Christian impact is best seen in the fact that it frightened the leaders of the Tervingi
into two waves of persecution (see 316ff illfra).
14 E. A. Thompson, "The Conversion of the Visigoths to Catholicism," Nottingham
Mediaeval Studies 4 (I960) 4-35.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
294 THE GOTHIC CONVERSION
(d) The then bishop, Ulfila, invents Gothic script and undertakes
missions (4.33.6).
(e) Ulfila's missions go beyond Fritigern's people into Athanaric's,
which rouses the latter to anger and causes the persecution.
The chapter closes with a brief account of Arius' doctrine and
error (4.33.7-9).
Although Socrates makes no distinction between the Tervingi and the
Greuthungi, his reference to Fritigem and Athanaric shows that it is
the former who are being discussed. No date is provided, but inas
much as the persecution began in 369, the rest of the action can only
reflect Valens' first Gothic war. Those who wish to follow Socrates
have therefore had to integrate his account with the independent
description of this war in the reliable contemporary historian Ammi
anus Marcellinus. At no point in Ammianus' narrative does religion
appear, and there is no mention of a civil war or of subsequent Ro
man intervention.
The absence of religion does not in itself make it difficult to reconcile
Ammianus with Socrates. Although he wrote in Latin, Ammianus be
longs among those late Roman classicising historians who otherwise
wrote in Greek, attempted to follow classical examples of historical
writing, and had a fixed idea of the subjects and language appropriate
to their work. As a relatively recent innovation, Christianity was not
considered an acceptable subject: even demonstrably Christian classi
cising historians avoided technical Christian vocabulary. Recent studies
have concluded, indeed, that although Ammianus was well-acquainted
with Christianity, he deliberately omitted Christian religious elements
from the events he recounts.I6 Even in Socrates the conversion is
linked to political events, so that its absence in Ammianus need not
present a major problem. If there is no other difficulty, he can simply
be supposed to have concentrated on secular politics and ignored the
religious element. Ammianus' omission of a civil war and subsequent
Roman intervention, however, is a more serious problem, which those
who wish to follow Socrates have had to overcome.
Representative of the way these issues have been faced is Varady's
approach. He suggests that in 367 Valens attacked all the Tervingi; but
15 Ammianus' account of the war appears at 27.5. On the question of genre see
Blockley 86ft"; classicising vocabulary is discussed by A. D. E. and A. M. Cameron,
"Christianity and Tradition in the Historiography of the Later Empire," CQ N.S. 14
(1964) 316-28; on Ammianus and Christianity see R. C. Blockley, Ammianus Marcelli
nus: A Study q{ His Historiography and Political Thought (= ColI.Latomus 41 [Brussels
1975]) 123ft"; E. D. Hunt, "Christians and Christianity in Ammianus Marcellinus," CQ
N.S. 35 (985) 186ft".
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 295
during the break enforced by adverse weather in 368, an alliance was
made with Fritigern, and in 369 Valens assaulted only Athanaric's
Goths. Integration of the two accounts along similar lines has been
proposed by other scholars who wish to follow Socrates.16 Although
attractive at first sight, such attempts are clearly unacceptable when
the two accounts are closely compared. Thompson has identified the
most obvious inconsistency: if Fritigern led the victorious faction in
369, it becomes hard to explain why Athanaric was still leader of the
Tervingi when the Huns attacked them ca 376, as Ammianus testifies.
This inconsistency is deepened by the fact, often ignored, that until
his disappearance Alavivus, not Fritigern, led those Tervingi who fled
south of the Danu beP Fritigern does not seem to have been even
second-in-line to Athanaric in 376, which is incompatible with the
view that he had been an important client of the Empire since 369.
More generally, Ammianus and Socrates understand the overall
action Quite differently. In Socrates' account Roman intervention
occurs when the Gothic civil war forces Fritigern to flee south of the
Danube for Roman help. In Ammianus it is Valens who begins the
war in response to a series of aggressive acts by the Goths, who first
threaten the frontier and then send help to the usurper Procopius.18
In Socrates, Valens exploits the division to establish a client relation
ship with one of the factions, while nothing of this appears in Am
mianus. If Socrates is describing the war of 367-369, then Ammi
anus' version is not only incomplete (which is indisputable),19 but
also fails to report the most significant outcome of the war. It would
have been a major diplomatic success for Valens to have detached
part of Athanaric's Goths, with important consequences for frontier
security; yet not a word of this appears in Ammianus. Indeed, Am
mianus describes a stalemate rather than a success. This is apparent,
above all, in Valens' inability to enforce symbolic superiority when
the peace treaty was signed. That he met Athanaric on a ship in the
middle of the Danube opposite Noviodunum indicates that he had
been unable to make the Goth come to him, either on Roman soil or
north of the Danube in Athanaric's own land, where it was custom
ary to receive 'barbarian' surrenders.20
16 Varady (supra n.9) 27ff; Chrysos 122f; KJein (supra n.9) 39ff.
17 Thompson 87; on Alavivus see Amm. Marc. 31.4.1.
18 For the threat to the frontier, 26.6.11 ~ on Procopius, 26.10.3.
19 For the campaigns north of the Danube, for example, Ammianus never mentions
time and direction.
20 With 27.5.9 c.f the historian's own gloss at 31.4.13; on submission to emperors
beyond the frontier see 17.l2.9ff, 18.2.1Sf, 30.6.lf.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
296 THE GOTHIC CONVERSION
Given these discrepancies Socrates cannot simply be combined with
Ammianus; the two versions do not complement each other, and in
proposed combinations Ammianus' account has been forced to fit
Socrates'. Some choice must be made between them, and there can
only be one response. Ammianus wrote a contemporary narrative of
wide scope and tremendous detail; in contrast, while Socrates is inter
esting and important, he does not inspire great confidence in particu
lars. Those who wish to follow him stress his accuracy, but such
assertions are difficult to justify. Even with a subject central to his
interests-the Arian dispute, for example-Socrates was capable of
major chronological errors, such as misdating the Council of Serdica
by about five years.21 Northern 'barbarians' were peripheral to his in
terests, and inaccuracy correspondingly more likely.
In this case, however, no choice is necessary. When we consider
the whole of Ammianus' account of relations between Valens and the
Goths, elements common to both Ammianus and Socrates indicate
that the latter confused an account that was originally similar to Am
mianus'. Zeiller's demonstration of this point has not received the
credit it deserves.22 He maintained that Socrates confused the perse
cution of 369 with the events of 376, when the Goths fled south of
the Danube. Indeed, the argument can be taken a stage further, since
Valens' first Gothic war also appears briefly in Socrates. Using the
sections into which Socrates' account was divided above, the chapter
can be rearranged in this way:
Events oj 367-369 Events oj 376
(a) the Gothic split (Amm. Marc.
31.3.8, where the Tervingi divide in
the face of the Huns; the only dif
ference being the extent of conflict
involved23)
21 To stress Socrates' value Chrysos 122 declares him a~7rW'7'O'TEpa 1TaVTClJI', and
Varady (supra n.9) 27 n.36 "die einzige authentische QueUe zur Ermittlung der frag
lichen Ereignisfolge." For the misdating of Serdica see G. F. Chesnut, The First Chris
tian Histories (= Theologie historique 46 [t 977]) 188f.
22 Zeiller (supra n.11) 453.
23 See 314 infra. Rubin (supra n.10) 46f argued that Ammianus' pluperfect deseruerat
implies that the break with Athanaric occurred before the troubles associated with the
Hunnic invasions. This usage in a subordinate clause is dictated by the fact that the main
verb is in the imperfect (quaeritabat) and does not carry this implication. Ammianus here
securely links Athanaric's political troubles to the problems resulting from the invasions
(31.3.4-8), and no other explanation for them is required. Rubin also overestimates the
degree of panic among the fleeing Goths: they were sufficiently organised to spend some
time (diuque deliberans) agreeing upon a sensible joint approach to the problems posed
by the Huns and sending an embassy to Valens to request asylum (31.3.8 and 4.0.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 297
(b.2) Roman campaigns north of
the Danube; c;f. the wording 1T'OI.
OVVTai. viKTJv KaTcl 'A8avapixov 1T'E
pav TOV "IuTpov = Valens' first Goth
ic war (Amm. Marc. 27.5.2-6)
(b.I) Fritigern's flight, where the
wording clearly implies movement
south of the Danube (CPPC.TI:yEPVTJ~
1T'POUc/JEVyEI. ·Pw/UXiol.~) =the flight of
the Tervingi before the Huns (Amm.
Marc. 31.4.5)
(c) the resulting conversion of the
Goths (in which Ammianus had no
interest)
. (d) A digression: the work of Vlfila
(e.n Athanaric's persecution, not in
Ammianus, but securely dated by
Jerome, following the campaign of
367-369
(e.2) Another digression: Arius' errors
Ammianus' detailed account could not be derived from one struc
tured like Socrates', while Socrates' account could well have been put
together from one structured generally like Ammianus' but also
concerned with the religious dimension. Socrates' sequence of events
is the logical outcome of a combination of Valens' first Gothic war
(and the subsequent persecution) with an account of the Gothic split
and the crossing of the Danube in 376. Athanaric's name would have
been common to both sets of information and was perhaps the cause
of the confusion. Whether the two sets of information came from
different sources is unclear. They were describing chronologically
separate events, of course, and might well have been taken by Socra
tes from different portions of the same source. Whatever the case
and there is no clear indication of where Socrates found his informa
tion - Ammianus can be used to unravel Socrates' confusions.
Analysis of Socrates, therefore, suggests the following conclusions:
The reconstruction of events proposed by those who would follow his
account must be rejected, because their main authority has confused
events of the 360's with those of the 370's. Equally, those scholars
who find chronological confusion only in Socrates' account of the per
secution (the second group of opinion) would also seem to be mis
taken. Keeping Socrates' order of events and merely redating them
after 372 avoids the most obvious problem, i.e., Fritigern's appear-
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
298 THE GOTHIC CONVERSION
ance and position, but it otherwise transmits the confusions that un
derlie Socrates' entire account. Indeed, this second position is in itself
untenable: for, as Thompson (88) has pointed out, it is inconceivable
that a major military expedition north of the Danube could have
been launched at this date without being mentioned by Ammianus.
Faced with these confusions Thompson rejected Socrates outright.
Confusion certainly exists, but complete rejection seems unwarranted.
Athanaric's persecution, Valens' campaign north of the Danube, the
Gothic split, and Fritigern's flight south (in a different order) all find
confirmation in independent, trustworthy sources. It may be sug
gested, therefore, that Socrates' association of the conversion with
Fritigern's flight to obtain help from Valens might also have some
basis in fact. This is uncertain, but the hypothesis that a Gothic
conversion was bound up with the lawful crossing of the Danube in
376 by the Tervingi under Alavivus and Fritigern (otherwise de
scribed by Ammianus) can be tested against the other sources.
Sozomen
Detailed examination suggests that Sozomen read, in addition to
Socrates, another source that dated a Gothic conversion to 376. It
was probably this source that prompted Sozomen to depart from
Socrates in placing the conversion south of the Danube. Since the
existence of this alternate source has never been established, it must
be demonstrated with care.
Chrysos has confirmed that Sozomen drew heavily and directly on
Socrates for information about the Goths and Valens. To illustrate
the debt, Sozomen's chapter can be broken down by source:24
Sozomen
(a) 6.37.1.
(b) 6.37.2
(c) 6.37.3-4
Materialfrom Socrates
An oration of Themistius
brings Val ens to religious
tolerance (4.32)
1. Introduction: public
events interrupt Valens'
plans (4.32)
2. The Goths are driven
south of the Danube by
the Huns (4.34)
Other Material
Digression on how the
Huns attacked the Goths
24 Chrysos 113ff; cf. G. Schoo, Die Quellen des Kirchen Historikers Sozomenos (Berlin
1910 150.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
(d) 6.37.5-6A
(e) 6.37.6B-7
([) 6.37.8-11 A
(g) 6.37.1IB-
12A
(h) 6.37.128-
14
(j) 6.37.15A
G) 6.37.158
(k) 6.37.16A
(I) 6.37.16B
(m) 6.37.17
PETER HEATHER 299
Transitional sentence, then
Socrates' account of the
Gothic civil war followed
by Valens' intervention and
the conversion (4.33.1-5)
Ulfila teaches, invents
Gothic letters, and carries
out the missions that lead
to the persecution (4.33.6-
8)
Goths quickly make peace
among themselves (4.34.0
Goths devastate Thrace
(4.35.0
Valens learns his mistake in
accepting the Goths (4.35)
The miscalculations behind
Valens' policy (4.34)
The Gothic revolt brings
Val ens to Constantinople;
he changes his religious
policy (4.35)
The Huns defeat the
Goths, a Gothic embassy
to Valens led by Ulfila en
ables them to settle in
Thrace
Introductory sentence, then
extra information on Ulfila
Extra information on the
persecution
Sozomen reproduces the setting for conversion familiar from Socra
tes: Gothic civil war followed by Fritigern's request for help, Valens'
intervention, a conversion out of gratitude, and the persecution. But
because the chain of events is placed south of the Danube, the end
result is a redating to the period after 376.
Jeep attempted to explain this discrepancy by supposing that So
zomen had returned to Socrates' source and drawn on it at greater
length. This is clearly impossible, for the two place the conversion in
explicitly different geographical settings. Jeep's argument would re
quire that one or other misunderstood the common source; even
given their known failings, this seems unlikely. More recently it has
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
300 THE GOTHIC CONVERSION
been suggested that Sozomen's account is a confiation of Socrates
and a chapter in Theodoret's Ecclesiastical History. There is, however,
no substantial resemblance between Theodoret and Sozomen beyond
the location of the conversion south of the Danube, and this cannot
explain Sozomen's alterations.25
The most substantial attempt to explain the relationship is that of
Chrysos (1190, who concluded that in addition to Socrates, whom
Sozomen not only used as a primary source but largely copied, Sozo
men used a source that described the arrival of the Huns (c). Indeed,
Sozomen reproduces the tale of the hunter and the doe, which origi
nated with Eunapius but may have come to Sozomen through an
intermediary. He also added to Socrates further information about
the Gothic persecution (h) that seems to have come from a hagi
ographic source. If the material on the Huns was derived from Euna
pius directly, this hagiography would represent a third source, since
Eunapius would have no interest in individual Gothic martyrs. These
observations are sound but do not explain why Sozomen changed the
order of events. Chrysos asserted that Sozomen was simply following
his own inclinations without further authority: Sozomen fabricated
the material in sections (d) and (f) to remove the taint of Arianism
from Ulfila and the Gothic martyrs; his account, in other words,
represents orthodox apologetic. In support of this interpretation Chry
sos pointed out that Sozomen omitted Socrates' explicit statement
that the martyrs were Arians, and he suggested that Ulfila's embassy
(f) was a deliberate misplacing of the Goth's known embassy of ca
340. In 376 Valens was at Antioch, and there would have been no
time for an embassy to reach him from the Danube.26
This explanation makes little sense of the evidence. Sozomen does
not "correct" (Chrysos' &Op(JWEVyovuc..
When reproducing it, however, Sozomen placed this information
among other Socratean material that appears in the original at 4.32
and 4.33. With Socrates 4.34 thus relocated between 4.32 -and 4.33,
the Gothic conversion (4.33) now follows the Hunnic attacks (4.34),
while in the original it preceded them. This could only have been
done deliberately.
Another example emphasises the care with which Sozomen edited
Socrates' account. In his version of Valens' intervention in the Goth
ic civil war, Socrates states explicitly that victory was won north of
the Danube. When Sozomen reached the same point, he virtually
copied Socrates but was careful to omit the location of the victory, for
in his reordered account the Goths had already fled south into the
Empire:
30 Athanaric did not arrive at Constantinople until 11 January 381 and died there
soon after, on 25 January: Const.Constant. s.a. 381 (Chron.Min. 1243).
31 E.g. Thompson 87, Fridh (supra n.12) 136.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 303
Sozomen 6.37.7:
TOV l>i (3aUiAEWf; EmTpE"'avT~ U1Jp,
IUlXEI.V aVr~ TO~ EV 9pc/tK'{l UTpa-
, "'a Q~ \' " ,
Twrra~, aVI71.~ U1J#£#J'A'I\.WV EVI.KTlUE Kal.
TO~ a#J4Ji 'A8avapl.Xov el~ PVP,E-VOlfi;
, , G ' , Q", KaTa TTJV uP'!KTlV uTpaTwrra~, fIU-
TJ8E'iV TOI.~ {3ap{3&pOl.~ Kanx (3ap{3&-, \...... ,
PWV UTpaTEVOVO"I.· Kal. 1TOWVVTal. VI.-
KTlV Kanx 'A8avaptxov ,"Epav TOV
"IUTpov TO~ 7TOAE,.uolJ'i) El~ f/lvrTI"
TpE"'a"TE~.
It is clear, then, that although Sozomen had only limited knowledge
and wrote well after the events, he deliberately re-structured his
major source.
Unless one attributes Sozomen's modification of Socrates to per
sonal whim - which is unlikely and cannot be documented - the most
economical explanation is that he was influenced by another source.
This must have made it clear that Socrates, by putting the conversion
north of the Danube, had misplaced it. Armed with this extra infor
mation, Sozomen returned to his main source and re-structured it
accordingly. He nevertheless used Socrates' general account of the
surrounding circumstances (civil war, appeal for help, Valens' inter
vention), which suggests that this other source was not as detailed as
Socrates. Sozomen was apparently unaware that in adhering to Socra
tes' general outline, he was being led into further confusion with the
appearance of Athanaric south of the Danube in 376 and the mis
dated persecution. Again the limitations of Sozomen's knowledge are
apparent; his other source seems to have corrected the place of con
version but provided no check for the other details.
This other source clearly combined two pieces of information, the
Danube crossing and a subsequent conversion: for without this com
bination Sozomen would not have re-structured Socrates. Of the ma
terial in Sozomen's chapter not drawn from Socrates (and which
should reflect this other source), section (d) contains half the combi
nation, bringing the Goths south of the Danube. Sozomen's account
of the conversion, of course, is taken directly from Socrates, and no
indication survives of how the other source originally reported it.
Nevertheless, reference to Ulfila in section (d) as the leader of the
embassy to Valens both makes it likely that the source contained a
note on the conversion and also explains why this has left no trace in
Sozomen. Ulfila's appearance shows that the source had an interest in
the religious history of the Goths, so that such information would
not have been out of place. Equally, if the source had stated Ulfila's
role in the conversion (as his part in the embassy of 376 would sug
gest) , no trace of this would be visible because Sozomen simply
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
304 THE GOTHIC CONVERSION
followed Socrates' account of the conversion, which had already em
phasised Ulfila's responsibility in the matter (4.33.4). Mention of
Ulfila in Sozomen's other source would thus have been masked by
his appearance in Socrates.
Only suggestions can be made on the origin of the other non
Socratean material in this chapter of Sozomen. Section (c), describing
the arrival of the Huns, may well have been attached to section (d)
and thus derive from the crucial other source. Section (c) originated
in Eunapius; and although it has been argued that Sozomen knew
him at first hand, the correspondence is not striking enough to prove
direct borrowing.32 The same origin might also be proposed for sec
tion (0 since it contains extra information on Ulfila, who is named in
(d). Section (h), however, seems to derive from a quite different
hagiographic source, perhaps even the original Passion, for it strongly
resembles fragments in the Synaxaries.33
We may conclude that the careful and deliberate changes made in
adapting Socrates are best explained by supposing that Sozomen fol
lowed another source: for there is no reason to think that he fabri
cated the material, and the limitations of his knowledge show that he
had no independent view of Gothic history. Underlying Sozomen, it
appears, is a source that linked the Gothic conversion to the Danube
crossing of 376. As far as one can tell, it seems to have been fairly
detailed and accurate, since it described an embassy that is known
to have taken place. In essentials it agreed with what has emerged
from Socrates: Fritigem's Goths were converted when they found
asylum in the Empire. Two authorities, therefore, provide the same
information independently. Neither inspires much confidence by it
self, simply because of the argumentation required to disentangle the
confusions; but the agreement of independent sources makes the
association of a conversion with the 376 Danube crossing much more
likely.
Sozomen's Source and Eunapius
Since the source used by Sozomen contained material on Ulfila and
the Goths, showed interest in political affairs, and may also have
used Eunapius, one possibility is obvious. The neo-Arian historian
Philostorgius seems to have had a greater interest in political affairs
than other church historians; and among the surviving fragments
there is some precise information from an earlier period on Ulfila and
32 See Schoo (supra n.24) 80ff on these. passages; cj. the judgment of Blockley 99f.
33 (r. Delehaye (supra n.13) 28Of.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 305
the Goths that mentions Ulfila's dealings with emperors and imperial
clergy. A continuing interest in Ulfila and the Goths would have
produced information of the kind that Sozomen read in his other
source and reproduced in part in sections (d) and (f). The Eunapian
material in section (c) may also have come to Sozomen from Philo
storgius, who had some knowledge of Eunapius. It is possible, how
ever, that Sozomen himself knew Eunapius, so that his knowledge
may have been both direct and indirect.34 It is of course easy to
assign material to sources that have not survived; Philostorgius is a
reasonable suggestion, but one that cannot be pressed.
This directs attention to Eunapius, a discussion of whose work
contributes more to an understanding of the Gothic conversion than
the question of Sozomen's source. Some scholars have attempted to
introduce into the argument a fragment of Eunapius that describes
the crossing of unnamed "barbarians" into the Empire. Although Eu
napius' editors assign the fragment to the reign of Theodosius I,
closer examination shows this judgment to be misleading, since the
fragment almost certainly refers to the crossing of the Danube in 376.
Indeed, it provides a third independent witness that a conversion to
Christianity facilitated the Goths' entry into the Empire.
The fragment appears as number 53 in the series excerpted for the
De sententiis of Constantine Porphyrogenitus. This collection provides
most of the Eunapian material that has survived directly from antiq
uity, and our argument can be confined to these excerpts with little
loss of directly Eunapian material. Moreover, the passages in the De
sententiis (like the others made for Porphyrogenitus) seem to have
been excerpted strictly in the order that they appeared in Eunapius'
original work.35 One uncertainty is removed, therefore, since no
Eunapian material in the De sententiis is out of place because of mis
takes in modern interpretation.
The argument will also assume that Zosimus can be taken to reflect
Eunapius' original work closely and can therefore act as a control for
placing the De sententiis excerpts. Discussion of the relationship be
tween Zosimus and Eunapius has proceeded from Photius' statement
that Zosimus did little more than summarise Eunapius. The accuracy
of this assertion has been disputed, but it seems agreed that Zosimus
34 Philost. 2.5; cI. Blockley 99f on Sozomen and Philostorgius' knowledge of Eu
napius.
35 Excerpta de sententiis 53 (ed. U. P. Boissevain [Berlin 1906] 89; subsequent page
references are to this edition). On the reconstruction of Eunapius' text see Blockley 97;
the working methods of the excerptors are discussed by J. M. Moore, The Manuscript
Tradition qf Polybius (Cambridge 1965) 129.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
306 THE GOTHIC CONVERSION
made such an extensive use of Eunapius's work that his history
retains much of the outline of Eunapius' origina1.36 It seems reason
able, therefore, to distribute the Eunapian excerpts from the De sen
tentiis according to the way in which subjects appear in Zosimus. (As
will emerge, the argument confirms the assumption that Zosimus
reflects Eunapius' original closely, since the order of the De sent.
excerpts neatly fits Zosimus' changes of subject.)
Examination of Zosimus indicates the absence of a strict annalistic
structure. This accords entirely with what is known of Eunapius, who
denied that an historian should concern himself with precise dates.37
The non-annalistic structure is important: although the events of 376
are referred to in De sent. 39 (corresponding to Zos. 4.20.3), Zosi
mus' narrative returns to the action of 376 at 4.26.1, after describing
the accession of Theodosius. When the intervening De sent. excerpts
are compared with Zosimus' history, there is no reason why §53,
describing the crossing, should not correspond with Zosimus' later
mention of the events of 376:
§39 clearly refers to the arrival of the Huns, and its contents are reflected at
Zosimus 4.20.3.38
§§40-44 are small fragments that would be out of order if the structure were
annalistic; they probably belong to an account of Isaurian troubles, reviewing
past outbreaks. The occasion would be an lsaurian revolt in 376, reported in
Zosimus at 4.20.1.39 It should be noted that Zosimus there promises to return
to these troubles, but never does so; he had, therefore, additional material
but chose to omit it. That Zosimus mentioned the Isaurian outbreak of 376
only once, before the passage that corresponds to the preceding fragment,
need cause no concern: Zosimus has merely omitted the information that
was the source of the fragments.
§§45-47 deal with the Gothic war, through the general Sebastianus' arrival
in the East and his military reforms. They can be positioned precisely, taking
the narrative to Zosimus 4.23.3, where the same subjects are covered.40
§§48f have led editors astray and caused the misplacing of 53; the editors
have associated them with Zosimus' account of Theodosius' moral inad-
36 Phot. Bibl. 98 (ed. R. Henry, II [Paris 1960] 66). Blockley (970 has reservations
about Zosimus' use of Eunapius, but uses Zosimus as a framework for Eunapius when
there is insufficient directly Eunapian material from the Excerpta.
37 Frr.1, 14.1 Blockley (1, 7 MUller) suggest a non-annalistic structure~ cf Blockley
6f.
38 De sent. pp.84f, Eunap. fr.4U Blockley (41 MUlier); both MUlier and Blockley
(104) agree with the association.
39 De sent. pp.85f, Blockley fr.43.1-4 (44f MUlier); for the explanation cj Blockley
104 and 160 n.53 .
• 0 De sent. p.86, Blockley fr.44.1, 2,4 (46f MUlier).
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 307
equacy.41 Closer examination suggests, however, that the fragments refer
instead to Theodosius' accession, since they comment on the evil effects of
sudden promotion: ov "lap 14>fJaCTE 11'apEA8wII Em ",II apxr1l1, Kai Ka8a11'Ep
, , ;\. \ , '\ 1"'\, \,
lUt.paKwlI IIE01T",OlJ'TOV 1TaTpo~ Em 1TOUq> XPOVq> 1TOAAa XP11/UXTa CTECTWPEV-
, ~.~, • .,I..' \ ...I..-.~.'. '8 ' '" '" ,~
KOTO~, uc.u U~pOU1JV1JV Ka£ ~c.ow a 'P0Wf; KVpuvuall TWV 1TpaYJ.UlTWV v."..,-
~ , , ""11. 8 ' ''' 8 ' , Th uPOII T£lIa Ka£ 11'aVTOWV O",E 'P0V KaTa TWII EVPE EIITWV /UXWETa£ . . . . ey
should therefore be placed earlier than is usual, at Zos. 4.24.4f (Theodosius'
accession). That this is the better position is further confirmed by the next
two excerpts.
§§5Of deal with the barbarians in Thrace: 50 discusses Nicopolis and 51
describes general suffering. Blockley associated them with Zosimus 4.33.2;
but although Thrace is mentioned at this point, the barbarians are mainly
infesting Macedonia and Thessaly. A much better context is 4.25.2, which
deals directly with "barbarians" in Thrace. Indeed, the reference to Thrace in
the later passage pointed to by Blockley refers back to this earlier one.42
Blockley could not, of course, use this passage because he had already placed
§§48 and 49 after it.
§ 52 is the most difficult to position because of the difficulty of disinterring its
point. Blockley considered that its thrust was moral decay, and that Theodo
sius was again the target. He therefore placed it at Zos. 4.33.3f.43 The first
sentence, however, contradicts this: TOWVrOIl BE T£ unoP11Ta£ -yEvECT8a£ KaTa
",V NEPWIIO~ /3aUtAEiav aAAa 1TEpi JJi,av m>A£v-suggesting rather that what
happened to the city in the time of Nero also affected the later Empire, but
more widely. The main subject of the excerpt might therefore be disease or
re-colonisation (the two subjects mentioned). It remains hard to place but is
perhaps best understood as further comment on the suffering of cities, and
placed at Zosimus 4.25.2.
§ 53 is the excerpt central to the enquiry, and has been dated to the reign of
Theodosius.44 Blockley associated it with Zos. 4.33.3f, but again the contents
do not fit the proposed context. The passage in Zosimus certainly deals with
the tricks and deceptions of "barbarians" (the main point of the excerpt),
but the setting is completely wrong. Zosimus' barbarians were established in
Macedonia and Thessaly, while in this excerpt the barbarians cross into the
Empire from outside: EW 7rapaUKEvaua~, V'1r aVTWJI VUTEPOJI E7rOn.EJ.U£TO, ~ Ka£
,."Exp£ TE£XWJI rii~ m>AE~ ac/xKEu6a£. 51 The latter part of the passage
refers to the second Gothic war, south of the Danube. Mter Hadri
anople, Fritigern led the Goths as far as Constantinople. The first
part, dealing with the conversion, might therefore be dated immedi
ately before, to 376, since there is no intervening account of the
Danube crossing-a subject of major interest; but this is by itself
inconclusive. The other secondary Greek account, that in Theodo
ret's Ecclesiastical History, places the conversion south of the Danube
and before the Gothic revolt against Valens (i.e., 376). Chrysos cited
Theodoret in his attack on Sozomen's credibility because Theodoret
did not portray the Goths as thoroughgoing Arians: he could serve,
51 Fr.184.2 MUJler~ Excerpta de insidiis 77 (ed. C. de Boor [Berlin 1905] 11). Ammia
nus records the Goths' approach to Constantinople at 31.16.4. John of Antioch's
sources are uncertain for this period~ he may have used Socrates (A. Koecher, De
loannis Antiocheni aetate, fontibus, auctoritate [Bonn 1871] 30) but also seems to have
known Eunapius (Blockley 98f)~ thus he could have found an account of the conver
sion in either, or indeed both.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 311
therefore, as another example of orthodox apologetic to confirm
Chrysos' interpretation of Sozomen. According to Theodoret, Eu
doxius of Constantinople suggested that Valens bring Ulfila's Goths
to his brand of religion; but Eudoxius died ca 370. Thus, for Chry
sos, Theodoret contains a confused account of a 369 conversion.52
Theodoret has associated two incompatible pieces of information,
for Eudoxius died before the Goths crossed the Danube in 376.
Indeed, the whole of the second part of the chapter is puzzling, since
the Goths described by Theodoret are converted not from paganism
to Christianity, but from orthodoxy to Arianism. They had been
Christians for some time: 1TaAac. 'Yap 'Ta~ rij~ 8EO'YlIWUia~ aK'Tilla~
8E€afJ,ElIOc. 'TO~ a1ToO''TOAC.KO£~ ElIE'Tpeqx)lI'To 8O'Y~0'c. (4.37.1). It can be
suggested, therefore, that the bulk of the chapter refers not to 376,
but to a time when already-Christian Goths came to associate with
the Arian party of Eudoxius and Valens through the agency of Ulfila.
The bulk of the chapter documents, in fact, the continuing struggle
for a comprehensive church settlement in the East. A suitable con
text would be after Valens' accession in 364, when Eudoxius (bishop
of Constantinople from 360) attempted, under Valens' auspices, to
bring the East into line with his views. Theodoret may have confused
information about the first group of Goths (already Christianized),
who crossed the Danube in 348 led by Ulfila, with information about
the second and more famous crossing of 376.53 Theodoret provides,
therefore, another link between the crossing and the conversion, and
in addition extends our understanding of the life of Ulfila and those
who left Gothia with him in 348.
Study of the Greek sources yields several conclusions. Socrates
confused the chronology of an account of relations between Valens
and the Goths. In this account conversion came about as part of
Valens' assistance to Fritigern against Athanaric, which, when com
pared to Ammianus' account, suggests that it took place in 376.
Sozomen used another source that also associated a conversion with
the 376 crossing, and whose authenticity there is no obvious reason
to doubt; Sozomen, at least, considered it more authoritative than
52 Thdt. 4.37; for Theodoret's sources see Parmentier and Scheidweiler (supra n.6)
xxiii; c:f Chrysos 121f, commenting on 4.37.5.
53 Valens' attempted church settlement in the 360's is discussed by Kopecek (supra
n.27) 422ff; on 348 see Philost. 2.5. The explanation offered above is not the only one
possible. If the emperor in the second part was originally unnamed, so that the account
dealt with Ulfila, Eudoxius, and an anonymous ruler, this might represent another
version of the Council of Constantinople in 3591360. However, Eudoxius never wielded
as much influence with Constantius as he did later with Valens, so that the later dating
seems preferable.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
312 THE GOTHIC CONVERSION
Socrates and re-structured the latter accordingly. Eunapius De sent. S3
almost certainly refers to 376 and provides a third account that makes
conversion part of the crossing. These primary sources are reflected
in John of Antioch and Theodoret; the latter explicitly dates the
conversion to 376. The Greek sources, taken as a group and com
bined with other information, represent a strong tradition that a con
version was one of the conditions under which the Tervingi made a
legal crossing of the Danube in 376.
III. The Latin Sources
The Latin sources that deal with the conversion are comparatively
clear and much easier to use.64 The primary source is Orosius, whose
account is by itself inconclusive: Valens is again responsible for the
conversion, but the date is unspecified. According to Orosius the
Goths asked Valens for teachers per legatos supplices. For such a re
quest to have been made, Valens and the Goths must have been on
good terms, which in itself suggests 376. This follows, of course,
from our interpretation of the Greek sources, but also serves to
clarify it. Before 369 Valens was fighting the Goths; and from 369 to
at least 372 Athanaric was persecuting Christians north of the Dan
ube. No conversion is conceivable before Athanaric's power was
broken ca 376 by the Huns, nor after Fritigern's revolt against Va
lens, which would necessarily date the conversion to 376. Indeed,
Orosius' phrase per legatos supplices is reminiscent of the words in
which Ammianus described the envoys sent by the Goths in 376.55
In contrast, Jordanes is quite explicit: the Goths promise to take
Valens' faith as part of the agreement to allow them passage across
the Danube. The only issue here is the reliability of this statement.
Jordanes' account of Gothic history has been shown to contain many
errors and confusions: men are made to lead the wrong groups, and
the relationships between them are misunderstood. In isolation Jor
danes' account cannot be considered decisive; given all other indica
tions, however, his explicit testimony provides at least supporting
evidence in favour of a 376 conversion.56
04 Isid. Hist. Goth. 7f is highly derivative and need not be discussed: cf. Thompson 88
n.1.
550roS. 7.34.19; on the 36O's war and Noviodunum, supra 29Of. Amm. Marc. de
scribes the embassy of 376 at 31.4.1: missisque oratoribus ad Va/entem, suscipi se humili
prece poscebant.
66 Jord. Get. 25.131; on Jordanes' mistakes see B. Croke, "Jordanes' Understanding
of the Usurpation of Eugenius," Antichthon 9 (975) 8Uf. The following errors are
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 313
IV. A Post-376 Conversion?
Analysis of the sources, then, would suggest that a Gothic conver
sion was part of the crossing in 376, or at least took place in the reign
of Valens. A fourth group of scholars has nevertheless argued that
these sources are indeed mistaken; their argument is based on refer
ences to the Goths as pagans after 376. Thompson is representative:
"The whole story of the alleged war and the appeal to Valens for
Christian teachers should be dismissed as a fabrication designed to
account for the Arianism of the Visigoths." He bases this assertion
on the ",ilence of Ammianus.57 But among the sources he dismisses
are our three independent Greek accounts, supported by two Latin
authorities and by secondary references, all pointing to conversion ca
376. The Eunapian fragment is particularly important because it does
not explain Gothic Arianism. Nothihg suggests that Eunapius had any
interest in the controversy between Arianism and orthodoxy: he saw
the conversion of the Goths as simply another example of their
treachery. The unanimity of the sources, together with Eunapius'
separate concern, makes their witness too substantial to be dismissed
by assertion.
Ammianus' silence can be used convincingly against those who
postulate Roman intervention north of the Danube between 369 and
376. It is not surprising, however, that he should have been silent
about the Christian element in a political event of which he gives in
other respects a full account: as we have seen, his tendency is to
ignore Christianity as far as possible.58 Thompson argued that, even
apart from the issue of Christianity, Ammianus overturns the ac
counts of at least Socrates and Sozomen because he records no
Gothic civil war. But Ammianus does record a split among the Ter
vingi, most of whom broke with their established leader, Athanaric.
evident in 10rdanes' account of events ca 376: 25.131, all the 'Visigoths' flee south of
the Danube and there is no mention of Athanaric's Tervingi, who remained behind
(c/. Amm. Marc. 31.4.13); 26.134, Alatheus and Saphrax appear as Visigothic leaders,
although in fact they were Greuthungi (Amm. Marc. 31.3.3); 28.142, Athanaric suc
ceeds Fritigern, although Fritigern deposed him (Athanaric later went to Constanti
nople after being deposed again: cf. PLRE I 1200. Schaferdiek, "Zeit und Umstande"
(supra n.10) 90, asserts that 10rdanes altogether lacks independent value and simply
interpreted Orosius to mean 376. This is an oversimplification of the source problems
in the Getica, which make it quite possible, despite his many mistakes, that 10rdanes
had independent knowledge. For a recent introduction to the Getica and its problems
see 1.1. O'Donnell, "The Aims of 10rdanes," Historia 31 (1982) 22)-40.
57 Thompson 88f, followed by Fridh (supra n.12) 136ff.
58 Supra 294.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
314 THE GOTHIC CONVERSION
This could not have taken place without a political coup; and the
question is one of the degree of conflict surrounding the coup, rather
than one of basic disagreement in our sources. The split certainly
generated enough ill-feeling for one of Athanaric's kinsmen to fight
later for the Romans against those Goths who had broken away.59
Thus there is no basic contradiction between Ammianus and the
other sources, for in all of them appear division among the Goths, an
embassy to the emperor, and a subsequent crossing by consent.
The positive evidence for a conversion after 376 consists of two
passages of Ambrose of Milan. The first is a passing reference in the
De ./ide, which can be dated to 378, where Ambrose describes the
problems affecting the Danubian lands and compares the Goths to
Gog in Ezekiel: totum ilium limit em sacrilegis pariter vocibus et Barbaricis
motibus audivimus inhorrentem? Thompson interpreted sacrilegis vocibus
to mean that Ambrose considered the Goths pagans. But in Ambrose
the term sacrilegus is not reserved for pagans alone: it is his favourite
insult for the Arians who opposed him at the Council of Aquileia in
381.60 Ambrose's language is inconclusive, therefore, and might in
fact indicate that the Goths were already Arians in 378.
The second reference appears in a letter of May 382, giving an
account of Aquileia. Ambrose here attacks a certain Arian named
Julianus Valens, who had adopted a Gothic habit unsuitable for
Christians: nisi forte sic solent idolotrae sacerdotes prodire Gothorum.
Two questions follow: are the idolotrae sacerdotes pagans and, if so,
does it necessarily follow the Goths were pagans also? Although the
natural answer to the first is 'yes', further thought makes this less
secure, for Ambrose is complaining about a specifically Christian
priest. Despite his clothing, there is no charge of apostasy; Julianus
Val ens still remained a Christian. Ambrose's polemic also concen
trates on what seems to be the priest's collaboration with the Gothic
enemy: qui . . . declinavit sacerdotale concilium; ne eversae patriae, prodi
torumque civium praestare causas sacerdotibus cogeretur. Since Julianus
Valens was an Arian who collaborated with the Goths, Ambrose's
description could also fit the Arian priests of a recently converted
people in whom old pagan habits would not have died out overnight:
Ambrose may provide a glimpse into the Gothic world where pagan
forms affected the new religion. What seems at first sight an innocu
ous comment might be the key to Ambrose's complaint, etenim
59 Amm. Marc. 31.3.8: populi pars maior, quae Athanaricum attenuata necessariorum pe
nuria deseruerat; Athanaric's kinsman is Modares, whom Zosimus discusses at 4.2S.2tr.
60 De fide 2.16.140; cf. Thompson 89 and Fridh (supra n.12) 137; on Aquileia: Ambr.
Ep. 10; l;{. the judgement of Schaferdiek, "Zeit und Umstiinde" (supra n.l0) 94f.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 315
abhorret a more Romano. Religious custom, rather than religious
adherence, may have caused the problem.6t
Even if these idolotrae sacerdotes were pagans, this would not rule
out conversion. As we have seen, it is unlikely that an entire people
would have changed their religion overnight, particularly since the
conversion of 376 seems to have affected primarily the political lead
ership of the Tervingi. More important, the conversion affected only
those Goths who entered the Empire by agreement: the Tervingi of
Alavivus and Fritigern. At least one other sizable group, the Greu
thungi of Alatheus and Saphrax, crossed the Danube illegally at more
or less the same time (Amm. Marc. 31.5.3). No promise to convert
bound them, and the existence of pagan Gothic priests in the Balkans
ca 380 need not disprove a 376 conversion.
The arguments for a conversion after 376 are not convincing. They
neither undermine the evidence pointing to 376 nor show that the
passages in Ambrose are substantial proof of conversion at a later
date. Indeed, conversion in the 380's creates a serious problem,
because the Goths became Arian rather than orthodox Christians. It
is hard to understand how the Goths became Arian when the ultra
Nicene Theodosius was ruling, unless the mass of the people had
been converted before 380. Explanations have of course been of
fered: it has been suggested that Theodosius used Arianism to sep
arate the Goths from the rest of the population of the Empire, that a
more hierarchical Trinity appealed to the Gothic leadership, and that
Arianism helped preserve Gothic independence.62 Any of these might
be correct, but all are hypothetical attempts to deal with an awkward
ness in the record; moreover, since no source actually states that the
Goths were converted in the 380's, they are also unnecessary. To
reverse Thompson's argument, a conversion in 376 provides an en
tirely logical explanation of Gothic Arianism: they simply adopted the
form of Christianity favoured by their imperial patron, Valens.
v. Conclusion
The sources thus offer no serious rival to 376 as the date of the
Gothic conversion. It remains to demonstrate that this makes sense
in the context of fourth-century relations between the Goths and the
61 Ambr. Ep. 10.9; Thompson 90 and Fridh (supra n.12) 138 see a reference to
paganism; on the alternative see Schaferdiek, "Zeit und Umstiinde" (supra n.lO) 95.
62 Separation is suggested by A. Ehrhardt, "The First Two Years of the Emperor
Theodosius," JEH 15 (1964) lOfT, hierarchy and independence by Thompson 109f.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
316 THE GOTHIC CONVERSION
Empire. The link between conversion and the legal crossing of the
Danube made by the Tervingi carries with it the implication that re
ligion was a political issue to Roman and Goth. The surviving infor
mation on relations between the Empire and the Tervingi is limited,
providing a series of episodes rather than a comprehensive account.
These suggest, nevertheless, that religion was important politically
and was thought to influence loyalties.
Before the invasion by the Huns, the Tervingi leadership resisted
the spread of Christianity with two persecutions, in 347/8 and from
369 to at least 372. The first receives only passing mention and its
causes can only be guessed; but the later persecution left more of a
mark. Two related motives are reported: according to the church
historians, Athanaric ordered persecution because the ancestral tribal
religion was becoming debased, while Epiphanius records that it was
designed to spite the Romans, whose emperors were Christians.6a It
would seem that the Tervingi were afraid that Christianity would
undermine that aspect of Gothic identity which was derived from a
common inherited religion, and that Christianity was associated with
an empire whose influence they attempted to resist.
That religion was a political issue to Gothic leaders is confirmed by
the way in which persecution followed important events in Gothic
Roman relations. The milder persecution of 347/8 was linked to a
diplomatic dispute with which Constantius was forced to deal before
attacking the Persians with Gothic help. More strikingly, the better
known persecution after 369 was launched in the same year that the
Treaty of Noviodunum separated Roman and Goth,64 and may well
have been designed to rid Gothia of the religious as well as the politi
cal influence of the Empire.
These fears on the part of the Gothic leadership had some basis in
reality, for the Empire had a manifest interest in the spread of Chris
tianity. Ulfila was ordained by the emperor Constantius' leading bish
op, Eusebius of Nicomedia; the same emperor is reported to have
received Ulfila personally when the latter was forced to flee to the
63 On the state of ancestral religion: Soc. 4.33.7 (C;(. SOl. 6.37.12); the anti-Roman
motive is reported by Epiph. Haer. 70; c.(. Thompson 98ff. The persecution of 347/8
may have been caused by evangelism, since Philost. 2.5 implies that Ulfila's mission
was to those already Christian, whereas the translation of the Bible into Gothic sug
gests that Ulfila was committed to wider missionary activity.
64 On 347/8 see E. A Thompson, "Constantine, Constantius II, and the Lower Fron
tier," Hermes 84 (1956) 379fT; there is no evidence for an invasion, nor was there a
break in the Constantinian settlement, as Thompson thought; c.(. Chrysos 77f. The
second persecution is dated to 369 by Eusebius/Jerome (supra n.5); for the separation
of Roman and Goth caused by Noviodunum, see supra 29Of.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
PETER HEATHER 317
Empire in 347/8. It may be that the idea was already current that
Christianity could help to pacify dangerous peoples.66 Whatever the
case, Christians north of the Danube had strong links with those
south of it. The priest Sansala, for instance, moved south of the river
to avoid the persecution after 369, while strong links are clear be
tween Gothic Christians and the churches in Asia Minor. This bond
perhaps originated when Goths took Christian prisoners there in the
third century, but they continued to develop: the body of St Saba was
transported to Cappadocia, for instance, even though he was of Goth
ic rather than Roman descent. That this was done through the agency
of Junius Soranus, then dux Scythiae, provides some further jus
tification for doubts among the Gothic leadership about the loyalty of
its Christian subjects. Soranus had his own links with Cappadocian
Christianity: he is the addressee of Basil Epp. 155 and 165. But his
agents in Gothic lands, from whom he presumably received Saba's
body, were no doubt also of use to him in his function as the military
commander of forces in a frontier province on the Danube.66 Indeed,
such contacts may have formed the kind of network from which, in
365, imperial officers had received advance warning of Gothic aggres
sion on the Danube (c! Amm. Marc. 26.6.11).
Religion seems therefore to have been a political issue to both Em
pire and Goths in the fourth century; a religious clause would thus
have been a natural part of the agreements that brought them into an
unprecedentedly close relationship in 376. Final confirmation is pro
vided by Fritigern's choice, when he attempted to stave off battle
before Hadrianople in August 378, to send Christians-including at
least one priest-as his representatives to Valens.67
To conclude: the sources strongly suggest that a Gothic conversion
was part of the agreement by which the Tervingi of Alavivus and
Fritigern gained legal admittance to the Empire in 376. This would
have applied to other Goths, especially the Greuthungi of Alatheus
and Saphrax, who entered the Empire illegally at the same time.
65 On Ulfila: Philost. 2.5; for the link between Christianity and pacification see
Thompson, "Christianity and the Northern Barbarians," in A. Momigliano, ed., The
Corifiict between Paganism and Christianity in the Fourth Century (Oxford 1963) 65ff; for
parallel imperial interest in Christian mission in the fourth century, this time to the
Himyarites, see A. Dihle, "Die Sendung des Inders Theophilos," Palingenesia IV
(Wiesbaden 1969) 330-36.
66 Sansala: Passio S. Sabae Gothi 4 (Delehaye [supra n.13] 218, lines 23ft); Cappado
cia: Ulfila's family had been taken by the Goths from Cappadocia (Philost. 2.5), and
Eutyches was also from there; Soranus: Passio S. Sab. 8 (Delehaye 221.11ft); cf. PLRE
1848.
67 Amm. Marc. 31.1.8 and 15.6.
HEATHER, PETER, The Crossing of the Danube and the Gothic Conversion , Greek, Roman and
Byzantine Studies, 27:3 (1986:Autumn) p.289
318 THE GOTHIC CONVERSION
From this it follows that the split among Athanaric's following imme
diatedly preceded the crossing. Socrates and Ammianus combine to
make the chain of events clear: the Tervingi split over what response
to make to the Hun invasions; then Alavivus and Fritigern led one
group into the Empire to seek asylum. No client relationship between
Valens and any part of the Tervingi should be envisaged before the
split. Gothic independence established at Noviodunum in 369 was
maintained into the 370's.
Above all, the correct order of events deepens our understanding
of the catastrophic effects of the Hunnic invasion upon the Goths.
Because of the Huns, the independent Tervingi were first deeply di
vided, then forced into a client relationship with the Empire, and at
the same time surrendered their ancestral religion. Similarly, the
Greuthungi lost a king in battle and were forced to migrate westward,
leaving their homes. Without detailed study of the sources, these
conclusions remain obscured and the history of the Gothic world
distorted. Independent Gothia north of the Danube fell to the Huns,
not to internal division.68
DUMBARTON OAKS
August, 1986
68 Completion of this article was made possible by my tenure of a Junior Fellowship
at Dumbarton Oaks, for which I would like to express my thanks to the trustees. I
would like also to thank Dr E. Umberger and the anonymous reader for their helpful
comments.