The place of smallpox in the history of the ancient Mediterranean is rarely contested. Although Enlightenment-era physicians and medical historians debated whether Greco-Roman populations knew the disease,Footnote 1 most who commented on smallpox's history before 1850 found little reason to think it was anywhere to be found in the Mediterranean region in antiquity.Footnote 2 Yet, gradually, from the mid-19th c., well-known ancient plagues, like the Athenian and Antonine plagues,Footnote 3 and less-discussed epidemics, like those encountered in Aëtius of Amida's medical compendium and dated to Trajan's reign, and in Eusebius's Church History and fixable to 310–11,Footnote 4 were identified as smallpox. The Cyprianic and Justinianic plagues also emerged, on occasion, in scholarship as potential outbreaks of smallpox,Footnote 5 though those diagnoses no longer hold favor.Footnote 6 In this paper, we synthesize recent literature in the fields of paleogenomics and evolutionary biology that casts doubt on smallpox's antiquity and, thereby, on retrospective smallpox diagnoses of ancient plagues. We draw particular attention to the Antonine plague, as a smallpox diagnosis has been assigned some significance in histories of that pandemic.
Retrospectively diagnosing ancient plagues is no straightforward matter.Footnote 7 Without paleogenomic support, which remains hard to obtain for those disease outbreaks for which we have written sources,Footnote 8 no modern diagnosis of a disease reported in an ancient text is unequivocal.Footnote 9 Yet, disease identifications, no matter how disputable, have carried significant weight in histories of ancient disease, shaping our efforts to understand the demographic and economic impact, and determinants, of poor health in antiquity. A prominent example is the smallpox diagnosis of the Antonine plague. Although opinion on the diagnosis of that plague has begun to pivot in recent scholarship, as discussed below, no other ancient disease outbreak is as commonly or confidently identified as smallpox as is the 2nd-c. pandemic.Footnote 10
Reference to smallpox in scholarship on the Antonine plague, which appears to have commenced in the reign of Marcus Aurelius,Footnote 11 is often fleeting.Footnote 12 That said, the identification of the Antonine plague as smallpox has been used in a variety of ways in modern histories of the pandemic. Most significantly, the diagnosis has been employed to support the idea that the 2nd-c. outbreak sickened and killed widely, as then contemporary (and particularly non-contemporaryFootnote 13) ancient authors claim.Footnote 14 The diagnosis has, as such, shored up the thinking that the multiple and varied references we possess to epidemic disease in the 160s–180s, the bulk of which make no reference to symptoms and can therefore be difficult to connect to one another, refer to a singular pandemic-scale disease outbreak, as opposed to multiple discrete epidemics.Footnote 15
Some uses of the diagnosis are in opposition. Scholars convinced that there is no evidence for smallpox in the Mediterranean region in the centuries or decades leading up to the Antonine plague have employed the diagnosis to argue that the smallpox of the 2nd-c. pandemic would have been especially devastating, as the population, without previous exposure, would have been immunologically naïve.Footnote 16 Conversely, scholars convinced that the Roman world was already acquainted with smallpox have proposed that the outbreak, while demographically significant because it was smallpox, was not as demographically significant as it could have been had the population not been previously exposed.Footnote 17 A smallpox diagnosis of the Antonine plague has additionally informed speculation on the pandemic's case fatality rateFootnote 18 and on the population cohorts worst affected.Footnote 19 It has also influenced conjecture on the pandemic's geographical originsFootnote 20 and association with climate change,Footnote 21 not to mention its seasonality and ecology.Footnote 22 The diagnosis has even underpinned ideas about the outbreak's duration, periodicity, and relation to the 3rd-c. Cyprianic plague.Footnote 23 Considering all of this, it is not surprising that the smallpox diagnosis of the 2nd-c. pandemic has been regarded as “potentially a matter of great significance.”Footnote 24 Indeed, the diagnosis has amassed considerable weight.
That a smallpox identification can support these arguments rests on the knowledge we have of smallpox's more recent history, as well as on the idea that the pathogen that causes smallpox existed in antiquity and then manifested and recognizably behaved as smallpox did in the centuries leading up to its eradication in the 1970s.Footnote 25 Smallpox is caused by the variola virus (VARV), a highly infectious orthopoxvirus that is host specific (it only infects and causes disease in humans), transmitted primarily via respiratory secretions, and capable of spreading rapidly.Footnote 26 VARV causes severe disease, though its presentation can vary. A noncontagious incubation period averaging 10–12 days results in fever, body aches, and malaise before a spotted rash forms in the mouth and subsequently spreads to the face, extremities, and appendages before the trunk. A few days later, the spots fill with a cloudy, dense fluid, eventually becoming hard and pustular, “like peas under the skin.”Footnote 27 Five or so days later, the pustules start to crust and scab. For three weeks, from the appearance of the rash to the flaking of the scabs, the sufferer is infectious, though infectivity varies over that period.Footnote 28
In naïve populations, VARV is thought to have caused large-scale outbreaks resulting in significant excess mortality.Footnote 29 Without prior infection, or widespread variolation or vaccination, the disease could be highly prevalent, with case fatality rates reaching 25–30 percent (or higher).Footnote 30 That said, considerable variation in VARV prevalence and mortality between regions, settlements, and subpopulations is to be expected in outbreaks, owing to an array of cultural, demographic, and ecological factors, such as population density and distribution, mobility, housing, and preexisting disease burden. In populations where VARV was endemic,Footnote 31 and the adults had some immunity, VARV could still apply near-constant demographic pressure, claiming the lives of many infants and children on an annual basis and, in certain circumstances, irrupting as epidemics.Footnote 32
This is, in large part, how smallpox manifested in recent centuries. For generations, historians, both armchair and professional, as well as physicians, have looked for this smallpox in antiquity. But can we superimpose this smallpox onto the distant past? What of this smallpox should we expect to be able to identify in ancient sources? Our ability to overlay our modern knowledge of smallpox onto the ancient world and diagnose ancient plagues as novel emergences of VARV,Footnote 33 or surmise that VARV was endemic in large Roman-era cities and regions densely populated in antiquity,Footnote 34 requires careful consideration of the available evidence, most notably newly emergent data from paleogenomics.
Evidence for smallpox in antiquity is primarily textual. Apart from the quasi-pox-like rashes visible on some New Kingdom Egyptian mummiesFootnote 35 (which though often mentioned in the literature have yet to yield VARV DNAFootnote 36), reports of osteomyelitis variolosaFootnote 37 (a nonspecific bony lesion associated with VARV infection which remains very rarely attested in the archaeological record), or the great age of variolation (and insufflation) purported for several regions of Asia in epidemiologically orientalist writing on smallpox's past,Footnote 38 what we have are infrequent references to: i) smallpox-like disease, such as Thucydides's small ulcers and pustules,Footnote 39 Galen's ulcerated and scarring exanthem,Footnote 40 Eusebius's severe carbuncles,Footnote 41 Ge Hong's disfiguring epidemic pustules,Footnote 42 and Gregory of Tours's malignant pustules and vesicles;Footnote 43 and ii) sequelae associated with recovery from smallpox, mainly pox-scarred skin and blindness.Footnote 44 Although more evidence of symptoms and sequelae may yet be recovered,Footnote 45 textual references have been and will continue to be incapable of providing the evidentiary basis needed to definitively confirm or deny the presence of smallpox in antiquity. Other data, however, novel and largely independent of archaeological and textual indications of ancient smallpox,Footnote 46 now strongly suggest that the disease we know as smallpox in fact did not exist in antiquity,Footnote 47 but rather has a more recent evolutionary history.
To date, paleogenomics have both solidified some old thinking about ancient diseaseFootnote 48 and upended some traditional narratives.Footnote 49 As our understanding of the evolutionary history and historical geography of pathogenic disease advances with paleogenomics, historians and archaeologists must remain abreast of developments and, as necessary, reconsider our histories of ancient disease. The ability to detect, capture, and sequence remnants of pathogens continues to improve, but robust results in the subfield of pathogen paleogenomics have been standard now for over a decade.Footnote 50 Although bacterial pathogens causing acute disease, like Yersinia pestis (the cause of plague), have been more often reportedFootnote 51 than double- or single-stranded DNA viruses,Footnote 52 or viruses containing highly fragile RNA,Footnote 53 all have now been identified owing to the refinement and ever-increasing sensitivity of paleogenomic methods.
There are, presently, a handful of VARV paleogenomes that have been drafted using decades-to-centuries-old human remains from archaeological contexts and medical archives.Footnote 54 These paleogenomes, some of which are only partial (or poorly phylogenetically resolved),Footnote 55 date to the 17th–20th c. and provide unequivocal evidence for VARV before the modern era. They can be studied alongside the ~45 genome-scale sequences obtained from samples taken in the mid-20th c.Footnote 56
Recent phylogenetic analyses using modern 20th-c. genomes coupled with the reconstructed paleogenomes have refined estimates of VARV's evolutionary history. Inferences of these analyses are historically significant. They show, for instance, that the two prominent groups of VARV strains eradicated in the 20th c., variola major (Clade I) and variola minor (Clade II/alastrim), shared a recent common ancestor some 200 or so years ago, potentially the result of a vaccination-induced population bottleneck.Footnote 57 In addition, these analyses have emphasized the recentness of the evolutionary history of variola minor, which caused a less-acute variety of smallpox but granted immunity to more virulent varieties of the disease.Footnote 58 Although a scarcity of VARV paleogenomes continues to hamper our understanding of VARV's evolutionary history, we have also learned, most importantly in relation to ancient pandemics, that VARV associated with smallpox appears to be centuries old not millennia old.
The inactivation of genes (rendering them non-functioning) in orthopoxviruses has been associated with both host specificity and virulence. All inactivated genes found in 17th- and 18th-c. VARV paleogenomes are also found in all sequenced 20th-c. strains. We can be sure, in other words, that the VARV paleogenomes which have been drafted to date caused smallpox.Footnote 59 This contrasts sharply with recently identified “sister lineages” of VARV – the four oldest orthopoxvirus paleogenomes yet recovered.Footnote 60 These early 7th–late 10th-c. genomesFootnote 61 are genetically distinct from the VARV we know to cause smallpox, so much so that when they were reported in 2020 they were identified as “aVARV” (“ancient VARV”), as opposed to VARV (or “mVARV” for “modern VARV”).Footnote 62 These strains, while ancestral to the reconstructed 17th–20th-c. VARV paleogenomes, contain a different pattern of inactivated genes than the strains associated with clinical smallpox, including some genes known to be associated with virulence.Footnote 63 Clinically and epidemiologically, therefore, the disease the millennium-old aVARV strains caused is likely to have differed, perhaps considerably, from smallpox as we know it.Footnote 64 Evolutionary analyses of, and genetic relationships between, orthopoxvirus sequences now reconstructed from ancient human remains and medical archives have taught us that it is improbable that a variola virus like the VARV associated with the disease we know as smallpox (“mVARV”) existed in any similar sense in antiquity. As sister lineages, aVARV should not be considered the direct ancestor of mVARV, and it must be underscored that our knowledge of any shared ancestors, intermediary forms, and historical host distributions remains unclear and likely to be clarified only by paleogenomics.
Pathogens evolve, and as they do their host specificity, transmission mechanics, and infectivity, as well as the disease they cause can, among other things, change. DNA viruses, like VARV, may not evolve as fast as RNA ones, but our ongoing experience with the pathogen SARS-CoV-2, the cause of the disease COVID-19, has made clear the ability of viruses to mutate in meaningful ways.Footnote 65 As the evolutionary history of quickly evolving viruses begins to come into focus, our accounts of their past must remain alert to the limits of our knowledge and sensitive to the weight of the terms we use to describe disease caused by pathogens not suffered and studied in the modern era.
As nomenclature may mislead, to be clear, the recently detected VARV “sister lineages” are not the VARV that caused smallpox. The aVARV paleogenomes neither evidence the discovery of smallpox in ancient or Late-Antique human remains nor raise the possibility of the existence or circulation of smallpox in antiquity.Footnote 66 Rather, aVARV is an orthopoxvirus, now presumably extinct, to which humans were susceptible. Whether it was zoonotic, rodent-borne, or dependent on a nonhuman reservoir, as speculated,Footnote 67 remains uncertain,Footnote 68 but that it has been discovered in early medieval northern Europe, where there were no population centers or regions with a population density even close to approaching the size needed to maintain smallpox, indicates how epidemiologically different aVARV could have been. Its degree of virulence and clinical manifestation are likewise unclear.Footnote 69
What we know for certain is that the sequences of VARV so far obtained from 17th-c. and later samples are remarkably similar to the VARV we understand causes smallpox, so much so that they have been collectively referred to as mVARV. On the other hand, aVARV sequences are more distantly related but, barring intervention and subdivision from the International Committee for Taxonomy of Viruses, must be considered of a single species and taxonomic unit with mVARV. Nevertheless, the genomic content of aVARV DNA, and the inferences for its virulence, and the uncertainty of its host specificity and/or reservoirs, suggest that its expression and behavior are very likely not that of smallpox but rather something for which we do not have a name. These semantics are important. Indiscriminate use of “smallpox” confuses matters. Recent scholarship notably continues to use “smallpox” in its discussion of the Antonine plague and aVARV. This loose usage has a long tradition: over 175 years ago, Haeser suggested smallpox symptoms may have differed in antiquity.Footnote 70 Yet, if the symptoms and epidemiology differed, the diagnosis loses its value, interpretive and heuristic, and becomes a hazard, as diagnoses, certainly ones of smallpox, carry considerable cultural, demographic and epidemiological weight.
Importantly, the aVARV sequences help us to pinpoint a minimum date range for the window during which VARV emerged as the virulent and highly infectious pathogen later described as the etiological agent of smallpox. These “time to the most recent common ancestor” (tMRCA) analyses are Bayesian analyses that attempt to estimate a date that reflects the point in time when aVARV and mVARV shared a common ancestor. Current analyses date this theoretical ancestor – with large confidence intervals – to the 4th c. CE.Footnote 71 Crucially, we do not know whether this theoretical ancestor resembled mVARV or was capable of causing a disease we would recognize as smallpox. A smallpox-causing VARV, therefore, could have emerged in the 4th c. or after the 4th c., not by the 4th c.Footnote 72 Molecular-clock analyses can only estimate the evolutionary history of sampled pathogen diversity, however, and on the basis of all available reconstructed mVARV and aVARV sequences, we must presently conclude that a VARV capable of causing smallpox could have emerged any time between the 4th c. and 16th c. CE. The time frame suggested by all available sequence data (paleogenomes and genomes), suggests a tMRCA closer to the early modern period than the Late-Antique one.Footnote 73 As such, smallpox's antiquity is very much in doubt and its Late Antiquity is in doubt too.Footnote 74
That smallpox, as known to modern biomedical science, first appeared as recently as 500 years or so ago is not set in stone. We remain in the midst of a paleogenomic revolution, and many more VARV and related orthopoxvirus sequences will yet be drafted from old soft tissues, teeth and bones, guaranteeing that our understanding of VARV's evolutionary history will continue to evolve. However, it must currently be regarded as unlikely that the emergence of a VARV that we understand to have caused smallpox will be pushed back to antiquity or Late Antiquity.
Without smallpox – pandemic, epidemic, or endemic – ancient Mediterranean populations should seem healthier to us.Footnote 75 So should societies in other world regions the disease is likewise thought to have long afflicted.Footnote 76 We are only beginning to grasp the pathogenic load suffered in antiquity anywhere, but the implications of a smallpox-less ancient world are many. For instance, several areas long stigmatized as ancient “cradles” of the disease based on exceptionally weak or nonexistent evidence,Footnote 77 and speculatively othered as sources of ancient Mediterranean smallpox epidemics,Footnote 78 can no longer carry that burden. In the absence of smallpox diagnoses, ancient disease outbreaks will also seem less familiar and understandable. Not being able to employ smallpox, what we know of its symptomatology or epidemiology, in our arguments regarding the demographic import, chronology, origins, or periodicity of ancient epidemics and pandemics will be a challenge to overcome. This is especially true for the Antonine plague, on account of the many roles a smallpox diagnosis has been assigned in our histories of that pandemic.
One option to confront this challenge, albeit one not recommended, would be to assume that plagues previously diagnosed as smallpox were in fact aVARV, or another VARV ancestor, which happened to manifest and behave a lot like VARV.Footnote 79 Although one might propose that the written evidence we have for ancient smallpox-like plagues (Galen, Eusebius, Ge Hong, etc.) corroborates this thinking, the idea is problematic. We know so very little about VARV ancestors – their ecology, epidemiology, or symptomatology – that such a diagnosis is not only unsubstantiated and unhelpful, but also misleading, as the continuity this proposal suggests between a VARV ancestor and VARV is wholly conjectural and not supported by available evidence. Another option would be to reject diagnosis, to adopt agnosticism and interpret the primary evidence on its own, without the superimposition of modern science.
Whether or not one pursues this second option, they should remain alert to the aspects of our histories of ancient plagues that have been built or buttressed with a smallpox diagnosis. Concerning the Antonine plague, we cannot simply abandon the diagnosis from our histories and move on. We will have to reckon with how the loss of the diagnosis will cause our accounts of that pandemic to change – our confidence regarding its origins, periodicity, and breadth will weaken, and a few ideas present in our histories will have to be given up, like questions concerning the 2nd-c. plague's demographic profile and density dependence.
The Mediterranean antiquity of smallpox gained favor in academic writing from the mid-19th c., when the disease was, in many world regions, in steep decline. On the verge of VARV's eradication, smallpox identifications of ancient plagues gained currency. Commentators on smallpox's past who lived when and where the disease was prevalent, to the contrary, found no place for it in the Greco-Roman past. Paleogenomic data and evolutionary biological analyses now lend support to that thinking and indicate it is time to eradicate smallpox from our histories of the ancient world and ancient plagues from our histories of smallpox.