Contemporary Problems for Ancient Disasters
Epistemological road-blocks have emerged for three major strands of ancient disaster research: 1) problems of causation emerging from ‘neo-catastrophist’ or ‘declensionist’ perspectives, 2) issues of equifinality from the archaeological record, and 3) problems of source-reliability that emerge from quantifications of primary source texts, which imperfectly match onto the scientific and archaeological record. In this essay, I will briefly survey these three epistemological concerns.
Correlation vs. Causation in Neocatastrophism and Declensionism
Neocatastrophism is a mode of historical explanation that explains complex historical processes with punctuated, short-term disasters like earthquakes, volcanoes, plagues, droughts or famine. Ultimately the origins of catastrophism may be found in Renaissance accounts of the Biblical Deluge, the Flood of Noah (Gen. 6:9 – 9:17).
An article by Morhange and Marriner published in 2013 drew attention to the modern origins of neocatastrophism in the environmental activism of the 1960s. Morhange and Marriner pointed to the fertile ground that catastrophic explanations have found in mainstream press and scholarship because both are hard-pressed for readers. Spectacular words—‘fear’, ‘catastrophic’, ‘terror’, ‘extinction’, ‘collapse’—are sprinkled liberally throughout neocatastrophist accounts in order to drive up readership and communicate relevance in our particular 20th or 21st century moment, which is so characterized by (perceptions of) perpetual crisis.
Closely related is ‘declensionism’, which in the context of environmental history adopts earth and climate events as prime movers in narratives of civilizational collapse. Roni Ellenblum’s 2012 book, Collapse of the Eastern Mediterranean: Climate Change and the Decline of the East, 950-1072 is a prime example of this neocatastrophist or declensionist tendency, which tends to take historical accounts of catastrophe at face value, and either ignores or over-interprets scientific evidence for climate change. Ellenblum leaves the reader with an overly neat and tidy story of climate-driven collapse: unfortunately writing history from the reconciliation of archaeological, textual, or scientific evidence for climate and catastrophe is considerably more difficult than Ellenblum seems to suggest (See now the reviews Burke and Frankopan, or another consideration of the same period by Xoplaki et al. 2015.) At the heart of Ellenblum’s book and its critiques is an old problem: correlation versus causation.
In early February 2016, a very carefully argued article in Nature by Ulf Büntgen, in association with the PAGES 2k Consortium, drew attention to Swiss and Russian tree-rings as evidence for major volcanic eruptions in the sixth century (536, 540, and 547 CE). The 536 CE eruption is recorded in Byzantine and Chinese sources as the Dust Veil Event (DVE), which induced long-term cooling trends throughout Europe and Asia. The authors identify the effected period that began with the DVE / 536 eruption and which persisted until 660 CE as the Late Antique Little Ice Age (LALIA). This period also witnessed the movements of steppe peoples from Central Asia into Europe throughout the 500s and 600s, the outbreak of the Justinianic plague throughout the Mediterranean in 541, political upheaval in China concluding with the establishment of the Sui dynasty in 589, and the collapse of the Sasanian empire vs. the rise of the Umayyad caliphate after the 620s. As stressed by the authors of this research, but not the popular press, volcanic-induced cooling during LALIA is correlated with but did not necessarily cause transformative changes in Mediterranean culture during this period.
Büntgen et alia emphatically acknowledge that multifactor analyses are required to understand long-term historical processes like global cooling vs. the transformation of the Mediterranean and the Roman world: environmental change is just one factor among many. The authors obstinately resist reductionist approaches and nondeterministic explanations of history based on climate: “any hypothesis of a causal nexus…requires caution”. New data from Swiss and Russian tree-rings are, perhaps, ultimately less important than the demonstration of how historians and scientists can effectively communicate with one another and the reading public about causation, complexity, and evidentiary limits.
Journalistic accounts of Büntgen et alia’s research were – no surprise – less measured and cautious. Media translations of the Büntgen article ranged widely in the truthfulness (or truthiness) of their translations. Typical were headlines like the Western Daily Press’s “Mini ice-age caused by volcanoes destroyed the Roman Empire - research has found” or the Daily Mail’s “Did Climate Change Cause the Collapse of the Eastern Roman Empire?”.
Both are neocatastrophist click-bait. The Daily Mail article begins with a dramatic statement of causation: “The eastern Roman Empire was brought to its knees after a century of cold weather led to famine and plague across the ancient world, according to new research.” The researcher is quoted advising restraint only at the very end of the article, after lots of ads and some heavy scrolling. Journalists covering Büntgen et alia’s 2016 article were overwhelmingly myopic for researcher concerns about causation and climate change in the ancient Mediterranean (pace the Washington Post and Phys.org).
Journalistic simplifications may at least partially be attributed to the fact that the experts themselves are not quite sure how to evaluate causation from the evidence, to write history from climate science in a way that rigorously acknowledges the complexities of both.
To quote John Haldon’s response to Büntgen et al. from the same publication: “It is recognized that societies do not always respond with the ‘right’ answer to crises, and the reasons for their failure can be located as much in the constitution and contents of the dominant belief system and its inner logic, as in other more concretely visible factors. Our rationality may not be theirs, even in the most existentially threatening circumstances. … We may not have the data to reveal causal mechanisms in detail.”
Diagnosing Catastrophe: the Archaeology of Equifinality
While Büntgen et alia’s article is focused on global-scale, long-term phenomena, there are similar problems of causation at local scales. Namely: what do famines, floods, tsunamis, or earthquakes look like from an archaeological perspective, at the scale of a city or region? With this question in mind, researchers have identified universal material signatures for disasters and climate change in the archaeological record that can facilitate comparison.
The fundamental challenge for archaeologists diagnosing disasters on the ground is equifinality: a variety of natural or anthropogenic causes can result in the same chaotic, high-energy destruction contexts unearthed in excavations, damage or repairs to buildings, and mass burials. Was a destruction deposit the result of invasion, epidemic, revolt, an earthquake, or a cow kicking over an oil lamp?
Mass burials illustrate the problem of equifinality and the potential for its resolution with scientific advances. Victims of non-plague catastrophes – of famine or drought, waterborne epidemics, earthquakes, war, and etc. – are probably under-represented in the archaeological literature, where sixth- to eighth-century and later medieval mass burials are often related to the Bubonic plague. For instance, the Mamilla Caves in Jerusalem contained hundreds of sixth-seventh century burials: Kloner 2001 linked these burials to the plague outbreak of 541 CE, while Avni 2010 connected the Mamilla mass-burial to the Persian Sack of Jerusalem in 614. Recent advances in the biomolecular identification and diagnosis of Yersinia pestis offer the opportunity to confirm or deny such hypotheses: Haensch et al. 2010 published a since-replicated method for identifying Y. pestis-specific proteins in ancient DNA extracted from dental pulp and bone material. While genetic testing of ancient remains from mass burials is a laborious process, McCormick 2015 took important steps in that direction with an exhaustive database and preliminary analysis of mass burials from the late antique Mediterranean, in anticipation of systematic biomolecular testing in the years to come. Even so, where there are no human remains, but only grave inscriptions that attest to date-clustered mass death, the ambiguity of equifinality will remain.
Earthquake archaeologists (archaeoseismologists) have spoken to equifinality by identifying characteristically earthquake-induced forms of structural damage to buildings in order to distinguish earthquakes from other types of catastrophe in the archaeological record. Earthquakes damage buildings with vertical ground deformation and lateral transient shaking. Clear signs of seismic activity that can be differentiated from other forms of destruction include pavement buckling, column falls, dropped keystones, folded steps, and penetrative fractures (e.g. Galadini et al. 2006 and Rodríguez-Pascua et al. 2010). Thus we might more critically distinguish whether earthquakes or lack of maintenance brought down a Roman apartment block; if Christian zealots or earthquakes destroyed a Roman temple during late antiquity, for example.
An interpretive problem remains, even with earthquake diagnostics in hand: it can be difficult to differentiate historical catastrophes with identical criteria from an urban or landscape perspective. How are we to know, for instance, if structural damage sustained at the North Bath / Church in Phrygian Hierapolis was caused by an earthquake c. 360 or c. 680 CE? Equifinality thus remains a challenge for the diagnostics of archaeoseismology, because different earthquakes can result in identical forms of seismically-induced structural damage.
Infrastructural or social responses to water scarcity have also been assessed (Scarborough 1991, Richard 2012). Equifinality is a problem here too, however: whether water-infrastructure adaptations were driven by climate change or social pressures like population growth can be difficult to determine (Izdebski et al. 2015).
Quantifying History: the Textual Limits
The last trend in ancient disaster studies that we might consider is quantitative assessments of disaster frequency via primary source texts. Catalogues of earthquakes, climate events, and various other natural disasters based on primary sources have been produced over the last two decades. The most recent and compendious catastrophe catalogue is the geodatabase assembled by the DARMC team at Harvard, led by Michael McCormick. The DARMC geodatabase – a publicly available Excel spreadsheet of famines, crop failures, earthquakes, droughts and floods attested in historical sources between 0-800 CE – was compiled from primary sources and earlier catalogues by Ambraseys (2009), Guidoboni (1994/2005), and Telelis (2004 and 2007). The DARMC geodatabase of disasters was interpreted in a widely cited article by McCormick et al. 2012 in the Journal of Interdisciplinary History, which sought to compare and confirm scientific paleoclimate data with the textual record for Roman antiquity. McCormick et al. 2012’s figure 6b demonstrates the acute risks of the textual record, however: coverage provided by Roman historians is spotty, inconsistent, and weighted by the ideological concerns of authors.
For instance, during the last quarter of the fifth century in the Eastern Empire, the primary source records assembled by McCormick et alia 2012 yielded no catastrophes whatsoever: this is improbable. The spikes of attested disasters that follow in the sixth and later seventh centuries, on the other hand, should not be associated with any absolute increase in the frequency of catastrophes, but rather with the widespread apocalyptic sentiments that pervaded the Eastern empire during a period of military setbacks, territorial losses, and broader cultural dislocation (for which see Magdalino 1993 or Haldon 1990). Further, the preeminence of disaster notices for Constantinople and Antioch in late antique sources reflects the interests and priorities of contemporary authors, like Malalas and Theophanes, rather than any veritable clustering of events. The assembled textual evidence is thus quite suspect, despite the appearance of quantitative integrity when formatted into a bar graph. Textual attestations will never equate to disasters in a 1:1 relationship chronologically: precisely what the quantification of an incomplete data-set means is, in this regard, up for debate.
Causation and correlation, equifinality, and the evidentiary limits of quantification are all considerable roadblocks for the future of ancient disaster studies. It may be hoped that Princeton’s Climate Change and History initiative and the PAGES 2k Consortium will suggest ways forward in the years to come. In the meantime, the issues outlined here should continue to concern readers – both interested scholars from other fields and laypeople – as they encounter the deluge of new disaster research in journals and its dissemination in the popular press.
Jordan Pickett (UPenn AAMW PhD '15) is a post-doctoral research fellow and lecturer at the University of Michigan, in association with the Ancient World Program and the Kelsey Museum of Archaeology. Jordan's research focuses on the social and environmental history of Roman and early Medieval cities in the Eastern Mediterranean