Gothia SubmergedThe Impacts of Severe Flooding on Valens's First Gothic War
According to the standard narrative of the outcome of Valens's First Gothic War, the Goths sued for peace after being subjected to three years of conflict. Yet not only did the Romans fail to engage the Goths in a major battle, but they were also prevented from campaigning in Gothia during the second year of the war due to flooding of the Danube River. This article argues that the natural disaster described by Ammianus played a significant, and hitherto unexplored, role in compelling the Goths to surrender. It demonstrates how the hydroclimatological conditions that cause floods on the Lower Danube also tend to cause equally devastating flood events in other rivers that flowed through Gothia, thereby potentially wreaking havoc on Gothic territory that had remained unscathed by Roman raiding. It further utilizes Western and Central European dendroclimatological proxy evidence to show that the spring months of both 368 and 369 were among the wettest of the entire century, which provides corroboration for Ammianus's account and raises the possibility of additional flooding in the final year of the war. Finally, the article explores the ways in which flooding could have had deleterious impacts on the Gothic agrarian economy and their war effort.
In the history of the Romano–Gothic clashes of the fourth and fifth centuries ce, Valens's First Gothic War is often overshadowed by the events that followed it. This is the case because the conflict did not have the profound implications of the Second Gothic War or the Gothic sack of Rome in 410, which [End Page 351] fundamentally altered the relationship between the Roman Empire and the Goths. It is also because there is an established consensus about the course of the First Gothic War and its eventual outcome. However, upon closer examination, it becomes apparent that the existing explanation for the curious manner in which this war abruptly came to an end in the summer of 369 leaves a great deal unexplained.
Most modern historians follow the late Roman historian Ammianus Marcellinus in assuming that the Goths surrendered after suffering from the crushing blows of Roman incursions into Gothia and a Roman trade embargo during the three-year war. This narrative, however, overestimates the impact of Roman military and economic aggression for three reasons. First, Ammianus reports that Valens's forces did not even cross into Gothia during the second year of the conflict because of severe flooding of the Danube River. Second, the Romans failed to draw the Goths into a decisive battle at any point during the course of the war. Third, it is unclear why the cessation of trade with Rome would have been so economically devastating to the Goths. In light of these problems, it is necessary to revisit our assumptions about the motivations behind the Goths' decision to surrender, and to explore what other factors could have precipitated this outcome.
As it happens, the account of this war offered by Ammianus hints at an additional factor that may help to explain why the Goths were so motivated to sue for peace: the severe flooding that occurred in 368. As we will endeavor to show, the hydroclimatological conditions that cause severe floods on the Lower Danube also tend to cause equally devastating flood events in other rivers that flowed through Gothic lands, such as the Siret, Prut, and Dniester. Therefore, severe flooding of these rivers could have potentially caused significant devastation in parts of Gothic territory that had avoided harassment by Roman forces. This possibility is also borne out by Western and Central European dendroclimatological proxy evidence, which has been used to reconstruct annual spring precipitation during the fourth century, and which shows that the spring months of both 368 and 369 were among the wettest of the entire century. Consequently, we propose that natural disaster(s) in 368 (and possibly in 369), in conjunction with Roman raiding, weakened the capacity of the Goths to continue the conflict, and thus drove them to sue for peace despite not having been defeated by Valens's forces.
The Goths in the Fourth Century and the Collapse of Romano–Gothic Relations
The Goths did not form a single political grouping during the reign of Valens. Instead, they were a hegemonic ethnic group comprised of different polities that have been called tribes, kingdoms, and confederations. And collectively they controlled much of the northwestern Black Sea basin beyond the Roman [End Page 352]
Empire.1 Ammianus records the names of the two most significant Gothic confederacies with which the Romans interacted during this period—the Tervingi and Greuthungi.2 He also indicates that the boundary between these two tribal confederations was the Dniester River, with the Tervingi controlling the region west of the river to western Wallachia and the Greuthungi ruling over the great expanse of land eastward from the Dniester through the Crimea to the Don River (Fig. 1).3 But these were hardly the only two subdivisions. There were, in fact, at least twelve different groups of Goths during the fourth century.4 Unfortunately, besides the Tervingi, little is known about the other Gothic polities or their relations with each other. [End Page 353]
Yet, despite our knowledge of such political subdivisions among the Goths, the Greco–Roman sources that preserve the only written record of the interactions between the Goths and the Roman Empire during this period are often ambiguous when describing Goths.5 For example, although Ammianus is aware of different Gothic polities, he typically refers to the barbarians in his treatment of Valens's First Gothic War as simply "Goths" (Gothos).6 Themistius and Zosimus are even less helpful, as they employ the classicizing name "Scythian" for the Goths in their accounts.7 Moreover, even when Ammianus offers specifics, this information can raise more questions than answers. For example, he reports that Valens negotiated with the leader of the Tervingi, but it is unclear whether the terms of the treaty were binding for all Tervingi, much less all Goths. Therefore, while it stands to reason that Valens primarily waged his war against the Tervingi confederation, especially given the geography of his campaigns, we will utilize the generic label of "Goth," except in those cases when our sources allow for more precision.
Relations between Rome and the Goths fluctuated considerably during the third and fourth centuries. While the earliest period of interaction was heavily characterized by conflict,8 Constantine's defeat of the Goths in 3329 ushered in three decades of peaceful relations and sustained trade between the Romans and their barbarian neighbors.10 However, this period of relative tranquility came to an end after the revolt of Procopius.11
The Procopian Revolt had considerable ramifications for Romano–Gothic relations because Gothic involvement in this internal Roman power struggle provided Valens with the justification for terminating the state of peace between Rome and the Goths. Prior to the usurpation, Valens received a report that the Goths were stirring and preemptively dispatched several units [End Page 354] to bolster the Danubian border.12 But while en route, these forces were won over to Procopius's uprising against Valens.13 The usurper additionally called upon the Goths to honor their treaty obligation to supply troops to the Constantinian dynasty, of which he claimed to be a member. The Gothic leaders were apparently persuaded by this petition,14 and dispatched several thousand soldiers to support Procopius.15 But the usurper's bid for power ended in failure the following year. And in the aftermath, Valens had these Gothic soldiers rounded up and, instead of repatriating them, he forcibly resettled them in Roman towns along the Danube.16 He also dispatched an envoy to the Goths to inquire into why they had supported the usurper.17 Unsatisfied with the Gothic reply, which claimed that they were merely honoring their treaty with Rome, and clearly in need of shoring up his precarious hold on the eastern imperial throne with an easy victory over barbarians, Valens made preparations in the winter of 366/367 to launch a punitive war against the Goths.18
Valens's First Gothic War (367–369 ce)
Valens's First Gothic War began in the summer of 367,19 when Roman forces deployed a pontoon bridge across the Danube at the old Constantinian bridgehead between Transmarisca and Daphne and then advanced into Gothic territory [End Page 355] in what is now the Romanian region of Wallachia.20 The Goths, who had probably observed the construction of the bridge, decided to evade the Roman forces and fled en masse into the nearby mountains.21 By September, Valens must have recognized that further campaigning would be futile and returned to Marcianople.22 But before doing so, he dispatched bands of marauders to wreak havoc across Wallachia and capture any lingering Goths.23 Therefore, despite failing to achieve a decisive victory over the barbarians, the Romans had "no doubt worked widespread destruction on farmsteads, burning houses and crops and capturing or killing stragglers."24 But, at the same time, Valens probably felt increasing pressure to deliver a noteworthy triumph, especially after hearing about senatorial concerns over the cost of this campaign.25
In order to facilitate a swifter advance against the Goths the following year, Valens moved his forces downriver to a camp near Vicus Carporum, where the Danube is closest to the Carpathian mountains.26 But the campaign season of 368 never materialized because excessive flooding of the Danube thwarted the Romans from crossing the river.27 Ammianus depicts a rather restful summer for the emperor and his forces, but Valens appears to have spent these months busily strengthening the Roman fortifications along the Danube.28 Across the river, the Goths should have enjoyed this respite from Roman harassment. And, as Wolfram suggests, they may have even viewed [End Page 356] the flood as an act of deific support from the Danube itself.29 But, as we shall see, there are compelling reasons for supposing that this flooding did not offer the Goths much opportunity for celebration or rebuilding, and instead it could very easily have made their situation even worse.
The subsequent campaign of 369 also commenced rather late in the season, as the Romans did not cross the Danube until July. Utilizing a wholly different route, the Roman army crossed at Noviodunum and advanced toward the northeast into Bessarabia, where it found some Greuthungi Goths and attacked them.30 Unfortunately, the extent of the engagement with the Greuthungi remains unknown. Athanaric and his Tervingi forces also offered some measure of resistance to this Roman advance; it is unclear, however, whether this fighting entailed only light skirmishing or something more substantial.31 In all likelihood, no major battle was fought between the Romans and Goths in 369.32 Shortly thereafter, Valens returned once again to Marcianople.33 During this campaign, the Goths also made several overtures to cease hostilities and the emperor finally received one of these embassies at the end of summer.34 Ammianus attributes the mutual desire for peace, on the one hand, to Valens's eagerness to turn his attention toward the growing threat of Persia and, on the other, to the Goths' need to alleviate the suffering caused by the cessation of commerce with Rome during the war.35 After further negotiations between Valens's generals and the Goths, the peace treaty was finalized [End Page 357] during a ceremony conducted on boats in the middle of the Danube.36 The treaty entailed the exchange of hostages, the restriction of trade to two sites on the Danube, the end of Roman subsidies to the Goths, and probably the termination of an earlier treaty requirement that the Goths had to provide soldiers for Roman campaigns elsewhere.37 Despite his failure to defeat the Goths on the battlefield, Valens was nevertheless celebrated as the victor upon his return to Constantinople with a triumphal procession and the assumption of the title "Gothicus Maximus."38
The Problem of the Gothic Surrender
What may have ultimately motivated Valens to accept the Gothic petition for peace remains a mystery, but his decision to do so certainly makes sense. In the three years spent on the Danubian frontier, his forces had been unable to draw the Goths into a decisive battle and he probably lost faith that subsequent campaigns would produce different results. Themistius's orations also suggest that the emperor was under increasing pressure from the Senate to cut his losses and devote his energies elsewhere. Moreover, given the recent Persian advances in the East, Valens was undoubtedly ready to march on an enemy that was not only more of a perceived threat to Rome, but also one that might actually meet him on the field of battle.39
It is, however, more difficult to explain the sudden eagerness on the part of the Goths to accept terms that were economically unfavorable, especially since they had never been defeated in a major engagement during these three years. The most popular explanation for this development has been that the three years of conflict, which entailed the cessation of trade with Rome and considerable damage inflicted on Gothic territory by the Romans, compelled the Goths to surrender.40 However, this explanation is problematic on two levels.
For one thing, it is not entirely clear why the cessation of trade with Rome would have been so economically crippling for the Goths. Even Thompson, who has maintained that the blockade was an integral component to the [End Page 358] Gothic surrender, notes, "It would be of the utmost interest to know how the ban on trade with the Roman provinces could have had such an effect."41 According to Themistius, pre-war Roman subsidies to the Goths consisted of only gold, silver, and clothing.42 So while these goods were certainly valuable, it is difficult to see how their absence would have crippled the Gothic economy. It is, of course, possible that the Goths had also traded for Roman grain or other essentials, but our sources remain silent on this point.
Secondly, it is often asserted that the destruction of Gothic infrastructure by the Romans was another key factor behind the war's sudden end. But while it certainly makes sense that the Roman invasion of Wallachia in 367 could have caused considerable damage in that particular area, much of the territory controlled by the Tervingi confederation remained untouched by that campaign. Moreover, given the absence of fighting in 368, even the Goths in Wallachia could have utilized the lull to restore their damaged infrastructure. And while it stands to reason that the Romans also pillaged Gothic lands during the campaign of 369, the full scope and impact of such activity remains unknown, and there is no evidence that Valens employed bands of marauders, as he had done two years earlier. Furthermore, since it is not possible to determine precisely when the Goths sent their delegations, Roman pillaging during this summer may not have played a significant role in the initial Gothic decision to pursue peace. Consequently, it is difficult to understand how the Roman invasions of 367 and 369 were sufficient to force the Goths to surrender, even if we consider the added economic impacts of the cessation of trade with Rome. For this reason, we propose that in addition to the Roman invasions, there is another important causal factor that, up to now, has not been given much consideration: the severe flooding of the Danube and its tributaries.43
Natural Causes of Severe Flooding on the Lower Danube
A useful way to make some general inferences about the natural causes of the Lower Danube flood that Ammianus tells us took place in 368 is by examining the climatic causes of severe flood events in the present. Both the hydrology of the Danube watershed and the causes of severe flood events on the Lower Danube are fairly well established, as are the impacts of severe flood events upon modern human populations. Accordingly, it can be presumed that, barring localized impacts of modern hydroelectric dams and other [End Page 359] similar anthropogenic alterations, the geography and hydrology of this watershed region in the present are generally similar to the conditions that prevailed in the area during the late fourth century. Thus, by understanding the causes of recent severe flood events in this watershed area, it becomes possible to extrapolate the meteorological and climatological circumstances that probably caused the intense Lower Danube flood described by Ammianus.
The Lower Danube is typically defined as the segment of the river that runs downstream from the so-called "Iron Gates" at the Serbian–Romanian border to the Black Sea.44 Its catchment area includes parts of northeastern Serbia, northern Bulgaria, much of Romania, and southern Moldova. Important tributaries of the Lower Danube, all of which originate in the Eastern Carpathians to the north of the main river channel, include the Olt, Prut, and Siret Rivers. However, the Lower basin adds comparatively little input into the Danube in comparison with the major tributaries of the Middle Danube, such as the Sava and Tisza Rivers, which also consequently provide a significant portion of the total water input for the Lower Danube.45
Several major floods of the Lower Danube have occurred during the past two decades, and have been the subject of multiple scientific studies. Two such events, which took place in the spring/summer of 2006 and 2010, will be considered here.
A joint Russian–Ukrainian study published in 2008 links significantly higher-than-average winter and spring precipitation in the eastern Carpathians to a major flood of the Middle and Lower Danube between March and July of 2006.46 The main cause of this flood was the combination of heavy snows during the winter months, which was followed by a warm and wet spring that not only resulted in heavy rainfall but also earlier than usual snowmelt.47 As a result, especially high precipitation levels were observed in the eastern Carpathians during this period. For example, at Miskolc, a city located in the upper Tisza River basin of northeastern Hungary, the amount of snow and rainfall received in the area was 144% of the amount that would fall in an average winter–spring period, with much of it falling in April and May. The flooding that resulted was unprecedented. Twelve hydrological gauge stations situated along the river's banks in Serbia, Bulgaria, Romania, and Ukraine recorded high water levels that exceeded historical flood maxima prior to 2006. In some cases, such as at the stations of Lom in Bulgaria and Braila in [End Page 360] Romania, this flood broke previous records by as much as 50 centimeters or more.48 Given these historic levels, the damage caused by the flood in the Lower Danube basin was considerable. In Romania alone, approximately 90,000 hectares (or 900 km2) of agricultural land were inundated by the floodwaters, while more than 1,000 houses were either damaged or destroyed, and some 12,000 people were forced to evacuate from their homes.49
The 2006 Lower Danube flood was followed four years later by another major spring–summer flood in 2010. As was the case in 2006, the main cause of the 2010 flood was higher than average snowfall during the winter months, which was followed by heavy precipitation beginning in the early spring.50 The maximum water discharge of this flood managed to exceed even that of 2006; for example, the 2010 flood maximum was some 19 centimeters higher than that of 2006 at the hydrological gauge station at Reni in Ukraine.51 Unlike in 2006, the 2010 flood had two distinct flood "waves" which peaked in March and July, respectively.52
Significantly, the conditions that enable powerful floods in the Lower Danube also tend to generate major floods in other rivers to the north of the Danube that flow through what was, in the fourth century, the heart of Gothic territory (now eastern Romania, Moldova, and southwestern Ukraine). According to a study of extreme precipitation events in Romania between 1980 and 2009, "the eastward mountain sides of the Eastern Carpathians are most prone to receiving larger amounts of precipitation" during years of significantly higher-than-average rainfall, and can receive considerably more snow and rain than the slopes on the inside of the Carpathian arch.53 Consequently, excessive precipitation during spring and summer months in this sector of the Carpathian range has the potential to cause flooding in the Upper Dniester Basin on the northern side of the Carpathian shield, and in the Prut and Siret Rivers, two major Romanian tributaries of the Lower Danube that have sources in the eastern Carpathians.54 This was the case in 2010, when the spring rains that helped to cause the Lower Danube flood also generated extensive flooding in these two rivers and caused widespread damage in the aforementioned region.55 [End Page 361]
Dendroclimatological Paleoclimate Proxy Evidence for Spring Precipitation in the Fourth Century CE
Having established the principal causes of severe floods on the Lower Danube and other nearby rivers in the present, it is necessary to consider whether there is any paleoclimate proxy evidence for similar conditions during Valens's First Gothic War. One form of proxy evidence that provides information about spring precipitation levels in Europe during the fourth century ce at an annual temporal resolution is a Western and Central European dendroclimatological proxy record. Because in numerous modern and historical cases (for example, 2002, 2005, 2006, 2010, and 2016)56 flooding in Central Europe was caused by the same meteorological and climatic phenomena as flooding on the Middle and Lower Danube and in eastern Romania, Moldova, and southwestern Ukraine, these proxy data can be used to make inferences about climatic conditions in the northeastern Carpathians and the Lower Danube watershed during the late 360s.57
We begin, however, with a brief introduction to the principles of dendroclimatology, in order to help orient readers who are unfamiliar with this method of paleoclimate reconstruction. To begin with, dendroclimatology is a subdiscipline of dendrochronology, which is the "science of dating annual growth layers (rings) in woody plants."58 Dendroclimatology involves the use of dated tree rings to understand both short- and long-term climate trends in the present and the past. The reason that tree ring sequences can be an excellent proxy for past climates is that tree ring growth is determined to a large extent by climatic conditions such as temperature and precipitation.59 Consequently, the analysis of the thickness of individual tree rings can be used to obtain climate information at an annual temporal resolution (whereas most other kinds of terrestrial climate proxy have millennial, centennial, or at best, decadal temporal resolutions).
Not surprisingly, the process by which tree ring samples are converted into proxy data for use in paleoclimate reconstructions can be very complex.60
For one thing, "not all woody plants produce ring-width sequences that are datable and usable for climatic inference," so careful selection of samples is required to ensure that useful data can be obtained.61 It is also necessary to [End Page 362] take numerous biological and ecological factors into account, such as the species of the tree, its age, and the environmental conditions in the vicinity. It is also crucial for individual tree rings to be properly dated to the year in which they were formed, and to be accurately measured in order to determine the amount of growth that occurred during that year62—and since the layers of individual trees grow at different rates as they age,63 this must also be considered. Even after tree rings have been dated and tree ring width measurements are complete, they must be correlated with other quantifiable climate statistics in order to produce dendroclimatological models for past conditions.64
For the purposes of the present study, the best available annual-resolution dendroclimatological record available is a regional Late Holocene dendroclimatological sequence for Western and Central Europe,65 which has been used to generate a comprehensive, annual-resolution paleoclimate reconstruction for the course of the past two and a half millennia. This reconstruction is pertinent because it provides a detailed record of spring (more precisely, April-May-June, or AMJ) precipitation totals for each year in the 2500-year dendroclimatological sequence. And because regional climate trends for the northeastern Carpathians are broadly similar to those of Central Europe, as they are both heavily influenced by the North Atlantic Oscillation (as noted above), the dendroclimatological proxy data not only provide a reasonable facsimile for AMJ precipitation totals in the northeastern Carpathians during 367–369 ce; they also make it possible to contextualize the spring weather in these years within the complete fourth-century sequence.
According to the reconstruction of fourth-century AMJ precipitation based on the dendroclimatological proxy data (which are shown in Fig. 2), between 300 and 366 ce, mean total spring precipitation was 178 mm; only once, in 361 (258 mm) did precipitation exceed 250 mm. The reconstructed AMJ precipitation totals during Valens's First Gothic War, on the other hand, are all relatively high in comparison: in 367 (236 mm) this value lies at the upper end of the "normal" range for the "pre-war" fourth century, while the total amount of precipitation in 368 (267 mm) and 369 (327 mm) was greater than at any point earlier in the century. In fact, the dendroclimatological reconstruction of AMJ precipitation shows that for the entire fourth century, the spring of 368 received the ninth highest total amount of precipitation and the spring of 369 was the wettest by a significant margin.66 [End Page 363]
Severe Flooding and Its Potential Impacts on the Goths
Several conclusions about the possibility of severe flooding in Gothic territory during Valens's First Gothic War can be drawn from the combination of the evidence for causes of modern floods in eastern Romania, Moldova, and southwestern Ukraine and the dendroclimatological evidence for abnormally high amounts of spring precipitation in 368 and 369. First and foremost, the dendroclimatological proxy data appear to provide empirical corroboration for Ammianus's account of a swollen Lower Danube that prevented Valens's army from crossing the river in 368. Additionally, considering that modern flooding patterns have demonstrated that the same climate patterns that cause Lower Danube flooding also tend to cause floods on other rivers that flowed through the heart of Gothic territory (the Siret, Prut, and Dniester, for instance), it is not unlikely that other severe floods could have taken place elsewhere within Gothic lands during the spring–summer of 368. Admittedly, Ammianus does not mention any such flood-related disasters during the lull in the fighting in 368. But this is perhaps not surprising, as he would have had very little, if any, knowledge of natural disasters that transpired so deep into Gothic territory.67 [End Page 364] And even if he had been aware of such an event, Ammianus's own biases would have likely led him to attribute the Gothic surrender to the actions of the Romans themselves, rather than to distant natural disasters.68 Furthermore, because the spring of 369 was evidently the wettest of the entire fourth century, and given that the Roman forces did not embark on that year's campaign until July, it is very possible that additional floods occurred during the spring–summer of 369 on any or all of these rivers.
How, then, might this potentially bear upon our understanding of Valens's First Gothic War? We propose that the rather abrupt ending of this conflict may have been precipitated by the economic impacts of severe flooding upon the Goths during the spring–summer of 368 (and possibly in 369). For the flooding would have considerably weakened their position and thus given them a strong incentive to sue for peace the following year, despite the absence of a clear Roman military victory. The key to understanding why floods could have forced the Goths to sue for peace lies in the geographical distribution of Gothic settlements and the nature of the Gothic economy.
Despite the common perception of the Goths as a migratory people, they lived mostly in permanent settlements in the fourth century.69 To judge from archaeological evidence,70 it appears that a substantial portion of these Gothic settlements (and, no less important, of their agricultural hinterland) in the northwestern Black Sea basin was clustered around rivers,71 where they generally occupied "wooded land, protected from the wind, and, naturally enough, close to water."72 The density of these settlements, moreover, appears to have been particularly high along the Lower Danube, Prut, and Dniester.73 Besides the economic reasons for establishing settlements along rivers, the location of these Gothic villages may have been further influenced by the potential presence of river deities in the Gothic pantheon.74 [End Page 365]
The settlement pattern of the Goths living north of the Danube is highly significant because, as we have already seen, the rivers around which so many of the Goths settled are also known to be flood-prone—and when floods do occur, they are often quite severe. Even today, despite the presence of sophisticated and extensive modern flood control infrastructure in the region, flood events on the Dniester and Prut have been known to inflict considerable damage and loss of life in northeastern Romania, Moldova, and southwestern Ukraine, particularly in settlements and agricultural areas.75 Consequently, given the comparatively primitive flood-management infrastructure that we must assume was available to the Goths, any severe flooding of these rivers that occurred during the spring and summer months of 368 could have very easily devastated the Gothic agricultural communities clustered along them.
Although there is some evidence for the developing sophistication of the Gothic economy during this period, including commercial workshops capable of mass production and specialized manufacturing, it was still primarily agrarian-focused.76 With such heavy dependence on staple crop agriculture, especially cereals like barley, wheat, and millet, their agrarian economy was particularly vulnerable due to the close proximity of Gothic farmland to the flood-prone rivers.77 Moreover, while the primary focus of Gothic animal husbandry was cattle-raising,78 in the lowland areas of what is now southwestern Ukraine, another important component of the pastoral economy was pig-raising.79 This is not surprising in view of the fact that the location of many Gothic settlements in wooded areas near water would have been ideal for pig-raising, as pigs cannot be easily herded across landscapes, but are well-adapted to foraging for food in forests, and the proximity of water sources would help to keep the animals cool during the warm summer months. But pigs were also disproportionally susceptible to loss during flooding because, unlike sheep, goats, or cattle, they cannot be herded quickly and easily, especially over long distances.
The human impacts of severe flooding in 368 would have been similar in character to those of modern severe floods: extensive damage to settlements, agricultural infrastructure, and roads, coupled with substantial loss of life. The widespread destruction of agricultural fields and potentially heavy losses of livestock would have, in all likelihood, constituted a serious, perhaps even crippling, economic blow to the Goths. Moreover, the situation of the Tervingi confederation would have been presumably made worse by the fact that Roman troops had inflicted substantial damage on Gothic agricultural [End Page 366] infrastructure during their previous campaign in Wallachia and possibly in Bessarabia the following year.
Although Ammianus does not mention another flood in 369, the tree-ring proxy evidence makes it possible to speculate that further flooding could have occurred in the spring–summer of that year. Indeed, given the inordinately high levels of spring precipitation for 369 that can be observed in the tree-ring proxy data, flooding in that year might well have been worse than in 368. Additional destructive flooding would, needless to say, have further worsened the Goths' situation and sapped both their capacity and resolve to continue the conflict. If additional flooding did take place in 369, this might also help to explain why Valens's army waited until July to cross the river—relatively late in the campaigning season—and perhaps also why they did so much farther downriver than they had in 367 (or than they evidently planned to in 368). However, additional research is required before anything conclusive can be said about whether or not further flooding occurred in the spring–summer of 369, and if so, what the impacts of these flood events might have been upon the course of Valens's brief campaign in that summer and the abrupt cessation of hostilities which followed it.
Even if it is difficult to say with any precision how badly the Gothic economy was devastated by the combination of flooding and Roman pillaging, there are good reasons for supposing that flooding could have played a key role in forcing the Goths to sue for peace in the summer of 369. For one thing, the devastation of Gothic agriculture caused by floods would have, in all likelihood, undermined the ability of the Goths to sustain a renewed campaign against the Romans. It is also very plausible that flood damage to vital transportation infrastructure in Gothia would have been substantial enough to hamper the Gothic war effort. Therefore, even if the Goths were capable of raising a defensive force to challenge Valens's army, the destruction caused by Roman pillaging and raiding during incursions into Gothic territory, and by flooding in 368 (and possibly in 369), still could have easily prompted the Tervingi leaders to decide that it was better to sue for peace. In doing so, they would avoid risking any additional damage in places that had remained relatively unscathed up to then, and focus on rebuilding efforts in the areas that had been afflicted by Roman raids, flooding, or both. Certainly, the desire to rebuild the sociopolitical cohesion of the confederacy appears to have been a priority of the Tervingi leadership, as is suggested by their subsequent persecution of Gothic Christians, whose adherence to the Christian faith was perceived to constitute a threat to the stability of Gothic tribal society.80 [End Page 367]
Given what we know about the causes of modern severe floods on the Lower Danube and the dendroclimatological evidence for abnormally high winter– spring precipitation in 368, it is highly likely that excessive precipitation in the eastern Carpathians caused the flood that Ammianus says prevented the Romans from launching a campaign against the Goths in that year. Furthermore, these same climatic conditions would have, in all likelihood, caused even greater damage throughout the Tervingi confederacy because other rivers, such as the Siret, Prut, and Dniester, were similarly fed by precipitation and runoff in the eastern Carpathians. So although the Romans were unable to cross the Danube in 368, it is nevertheless plausible that the Gothic agricultural hinterland suffered considerable damage because of flooding. Accordingly, these floods would have compounded the damage done by Valens's army in the previous year, when it ransacked the southern parts of Gothic territory. As a result, these separate events may have delivered a serious double blow to the agrarian economy of the Goths by the time that the Romans crossed the river again in 369. And if this were the case, it would certainly explain why the Goths were eager to sue for peace even though they had not been defeated decisively in battle.
Finally, it is worth noting that the Western and Central European dendroclimatological proxy record also raises the possibility that severe flooding could have occurred during the early 370s, and thus may, alongside other social factors, have served as a destabilizing influence on the Gothic polities, particularly if they proved incapable of mitigating the impacts of severe flood events. In that event, flooding could have potentially played an important but indirect role in shaping subsequent Gothic and Roman interactions in the late fourth and early fifth centuries. While more research is needed to explore this possibility, these dendroclimatological proxy data do at least offer the tantalizing prospect that severe flooding in what is now Romania, Moldova, and Ukraine may have helped to alter the respective histories of both the Goths and the Roman Empire. [End Page 368]
We would like to express our gratitude to the Research Center for Anatolian Civilizations at Koç University, where we had the opportunity as Junior Residential Fellows (2013–2014) to begin our work on this project. We would also like to thank Ulf Büntgen for generously providing us with the Western and Central European dendroclimatological proxy data that appear in this paper, and Ian Jones for his assistance in developing the map. Furthermore, we are grateful to the two anonymous readers and Noel Lenski, whose comments and suggestions were exceedingly helpful; all remaining errors are our own.
2. Amm. 31.3.1–4.
3. Amm. 31.3.5.
6. For example, Amm. 27.5.1, 27.5.9.
7. Them. Or. 8.113; Zos. 4.10.
8. The earliest-attested Gothic foray into the Roman Empire occurred in 238 ce, when Goths attacked the city of Histria, situated on the Black Sea coast near the mouth of the Danube in modern Romania. The Goths continued to raid the Roman territories of Dacia, Moesia, and Thrace into the early 250s, and they even managed to kill the emperor Decius in battle in 251. They also launched a series of seaborne invasions against Asia Minor between the mid-250s and 270s. On this period of conflict, see Salamon 1971; Wolfram 1988, 43–56; Drinkwater 2005, 28–66; Kulikowski 2007, 14–33; Mallan and Davenport 2015.
12. Amm. 26.6.11; although there are compelling reasons to doubt the veracity of the Gothic plan to attack (see Wolfram 1988, 66–67), Libanius (Or. 12.78) interestingly claims that a Gothic embassy in 363 conveyed their unhappiness with the terms of their treaty, and Julian essentially replied that only war could alter it.
14. While Ammianus records that several Gothic kings made the decision (Amm. 26.10.3), Eun. fr. 37 and Zos. 4.7.2 attribute it to the "Scythian chieftain" (that is, Athanaric). Given Julian's earlier dismissal, the Goths may have viewed Procopius as a potentially more amenable emperor than Valens; see Chauvot 1998, 194. Although the circumstances and ownership of a hoard containing nearly 3,000 silver coins found near Caracal (modern Romania) remains unknown, the presence of thirty coins of Procopius in it may suggest that the Goths were also paid for backing the usurper; see Ştirbu 1980, 152–53; Kulikowski 2007, 115.
15. Ammianus reports that 3,000 Gothic soldiers were sent to Procopius, which is certainly more believable than Zosimus's claim of 10,000.
16. Zos. 4.10.1–2; according to Zosimus, Athanaric also unsuccessfully petitioned Valens to return the soldiers.
17. Amm. 27.5.1–2.
18. Amm. 27.5.1–2; Zos. 4.10.3–4; it is also possible that Valens was manipulated into war with the Goths by his generals, who shared the same aggressive foreign policy as Valens's brother Valentinian; see Chauvot 1998, 193–95.
19. The best treatment of Valens's First Gothic War remains Lenski 2002, 116–52; additional studies of note include Nagl 1948, 2106–11; Thompson 1966, 17–20; Chrysos 1972, 94–108; Zahariade 1983; Wolfram 1988, 64–69; Chauvot 1998, 190–99; Wanke 1990, 73–110; Gutmann 1991, 112–32; Heather 1991, 115–21; Heather and Matthews 1991, 17–26; Seager 1999, 599–601; Heather 2006, 72–76; Kulikowski 2007, 114–18.
20. On the chronology of the campaign in 367, see Zahariade 1983. Although Ammianus implies that the Romans crossed in early spring, Valens was still in Marcianople on 30 May, as attested at CTh 11.17.1. For this reason, the invasion probably did not begin until early June. Zahariade 1983, 59 suggests that Valens waited this long because either the Danube was swollen enough to delay the crossing or Victor, whom the emperor had tasked with the initial embassy to the Goths (Amm.27.5.1), had not yet returned.
21. Amm. 27.5.3; the montes Serrorum to which the Goths fled were clearly the Carpathian mountains, and perhaps the southern portion known as the Transylvanian Alps. On the probable loss of the element of surprise for the Roman offensive, see Wanke 1990, 96.
22. On the return to Marcianople, Valens passed through Durostorum on 25 September, as attested at CTh 10.1.11, 12.6.14.
23. Amm. 27.5.2–4; see also Zos. 4.10–11.
25. Themistius arrived in the Roman camp at Marcianople and delivered Oration 8 to Valens on 28 March 368, see Vanderspoel 1995, 168. Even though the orator served as the mouthpiece for the imperial court, this oration ought to be read as a warning to the emperor about his Gothic campaign, including its financial burden on Roman taxpayers, see especially Or. 8.114–15 with Lenski 2002, 129.
27. Amm. 27.5.5: ingredi terras hostiles pari alacritate conatus fusius Danubii gurgitibus evagatis impeditus.
30. Amm. 27.5.6. But it is unclear whether the Romans ventured far enough to encounter the Greuthungi in their own territory (see Lenski 2002, 132) or the Greuthungi came west to aid their Tervingi brethren (thus Wolfram 1988, 68).
31. The scale of the "leviora certamina," mentioned at Amm. 27.5.6, is not altogether clear, although Them. Or. 10 implies that there was not a decisive battle. The nature of these engagements is further problematized by the fact that Ammianus's language is frequently inspired by Livy, who notably often qualifies certamen with levis (levia certamina: 27.12; 28.14; 37.18; levi certamine: 22.21; 24.36; 38.30; levique certamine: 1.10; levibus certaminibus: 7.32; 32.18; levium certaminum: 22.12). We would like to thank Philip Rance for drawing our attention to the likely influence of Livy here.
32. Themistius's silence concerning a decisive battle is especially telling because, if there had been one, he would have certainly known about it as an eyewitness to the final stages of the war, and it would have offered irresistible material to bolster his efforts in selling the war as a Roman victory in Or. 10.
33. Amm. 27.5.6.
34. Amm. 27.5.7; see also Them. Or. 10.133.
35. Amm. 27.5.7. Lenski rightly argues that Ammianus's "fear of the enemy" refers not to the Gothic fear of the emperor's presence (which is asserted by Thompson 1966, 19; Seager 1999, 600), but rather the Roman fear of the growing Persian threat (Lenski 2002, 133–34). This reading is especially compelling given that the Persians had ousted the pro-Roman rulers of Armenia and eastern Georgia in 367/368; see also Heather 2006, 73.
36. Amm. 27.5.9–10. There has been some debate over whether the treaty was finalized in late summer/early autumn 369 or early 370, but the former is more likely given Themistius's eyewitness remark about how hot it was at the time (Or. 10.134); see Wanke 1990, 106; Lenski 2002, 133 n. 101.
37. Ammianus was largely uninterested in the details of the treaty, but see Them. Or. 10.133–36 and Zos. 4.11.4.
38. Them. Or. 10.140.
39. See n. 35 above.
42. Them. Or. 10.135.
57. These same tree ring proxy data have also recently been utilized in a multi-proxy reconstruction of flooding in Hungary in 1242 ce, which likely helped to halt a Mongol invasion in that year (see Büntgen and Di Cosmo 2016).
68. It is also worth noting that Ammianus exhibits a discernible tendency to omit events when they do not interest him: for example, the spread of Christianity and the revolts of the Isaurians and Saracens. See Rubin 1981, 45–50; Lenski 1995, 59–65.
70. Gothic material culture is typically identified as the Sîntana de Mureş-Černjachov material culture complex, which stretches from the Danube to the Crimea. For a general survey of the Sîntana de Mureş-Černjachov culture, see Heather and Matthews 1991, 47–95; for studies that connect this material culture with the Goths, see Kazanski 1991; Heather 1996, 9–93. Poulter 2007, among others, argues that there is no connection between this material culture and the Goths; see also the critiques of the ethnic ascription approach in archaeology in Gillett 2002.
80. Thompson 1966, 100–102; Kulikowski 2007, 117. The lasting effects of widespread devastation caused by flooding may have also played a role in the subsequent Gothic civil war that fragmented the Tervingi confederacy shortly thereafter, on which see Soc. 4.33.1–3 with Rubin 1981, 41–49; Lenski 1995.