Epidemiology of the Plague of Athens

Author(s): David M. Morens and Robert J. Littman

Source: Transactions of the American Philological Association (1974-2014), Vol. 122 (1992),
pp. 271-304
Published by: The Johns Hopkins University Press
Stable URL: https://www.jstor.org/stable/284374
Accessed: 27-08-2020 21:50 UTC

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/284374?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and
extend access to Transactions of the American Philological Association (1974-2014)

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Transactions of the American Philological Association 122 (1992) 271-304

Epidemiology of the Plague of Athens

David M. Morens, M.D.

Robert J. Littman
University of Hawaii

...Sometimes...the barbarous rage

Of plague or pestilence, attends mans age,
Which neither force nor arts assuage;
Which cannot be avoided, or withstood,
But drowns, and over-runs with unexpected flood...
Draw back, draw back thy sword, 0 Fate!
Lest thou repent when 'tis too late,
Lest by thy making now so great a waste,
By spending all mankind upon one feast,
Thou starve thyself at last.

-From Thomas Sprat, "The Plague of Athens" (London 1676)

The cause of the celebrated epidemic in Athens (430-425 B. C.) is a subject of

longstanding controversy. For nearly 500 years, scholars have tried
unsuccessfully to identify the infection by comparing clinical features recorded
by Thucydides to those of contemporary diseases.' At least 29 different disease
theories have been advanced by hundreds of scholars. Adding to the
controversy over its etiology was a 1985 publication by Langmuir et al. that
revived an earlier influenza theory by explaining its incompatible clinical
features as complications of toxic shock syndrome.2 The national press
popularized the controversy, including UPI wire service and Newsweek
magazine coverage. Partially in reaction to this controversy, a panel was held
at the annual meeting of the American Philological Association (Society for
Ancient Medicine) where Langmuir presented his case, and various theories
were debated without consensus. At that meeting co-author Morens outlined an

1Over 200 articles and books have been written in this time. Among the best current surveys
are: J. Scarborough, "Thucydides, Greek Medicine, and the Plague at Athens: A Summary of
Possibilities," Episteme 4 (1970) 77-90; J. C. F. Poole and A. J. Holladay, "Thucydides and
the Plague of Athens," CQ 29 (1979) 282-300; and J. Longrigg, "The Great Plague of
Athens," History of Science 18 (1980) 209-225. See most recently R. Sallares, The Ecology of
the Ancient Greek World (London 1991) 221-294, esp. 243-266.
2A. D. Langmuir, T. D. Worthen, J. Solomon, C. G. Ray, E. Petersen, "The Thucydides
Syndrome. A New Hypothesis for the Cause of the Plague of Athens," New Engl. J. Med. 313
(1985) 1027-30; L. Mercier, "Essai d' Interpretation de steriskomenoi de la 'Peste' d'Athenes,"
Bull. Assoc. Guillaume Buds 4th series 2 (1974) 223-6.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
272 David M. Morens and Robert J. Littman

epidemiologic approach to identifying the cause of the Athenian epidemic. In

this paper we suggest that epidemiologic aspects of the Athenian disease strictly
limit possible modes of transmission and thus possible causes, ruling out
influenza and many other previously suggested diseases. To assess the
candidacy of the remaining diseases we suggest a systematic approach that
includes (a) use of mathematical models (theoretical equations that purport to
predict how epidemics will proceed under defined sets of conditions) and (b)
comparison of the actual Athenian epidemic, of mathematically modelled
epidemics of suspected diseases, and of epidemics of known diseases described
and epidemiologically distinguished in well-characterized premodern
populations. We argue that this process constitutes a conceptual framework
from which proposed causes should be viewed, and that the results of this
approach exclude most diseases suggested as having caused the Athenian
epidemic.
Our method tries to identify, or at least limit, candidates for the Athenian
disease by epidemiologic means in two successive stages: first we attempt to
determine the mode of its transmission, and thereby to exclude diseases
transmitted by other, inconsistent means; and secondly we attempt to find a
"best fit" from the remaining candidate diseases, using (a) established
mathematical models for respiratory disease transmission, and (b) data on
premodem epidemics in characterized populations. As part of this process we
examined Thucydides' text for aspects of descriptive epidemiologic data and
synthesized these with information on such factors as climate, population
density, and population growth dynamics, to construct a global picture of the
epidemic according to the approach of modem epidemiologists conducting
investigations of diseases of unknown cause. We compared the language and
phraseology of Thucydides and Hippocrates, his contemporary, to look for
equivocal or inappropriate medical terms or formulaic constructions, and we
scrutinized Thucydides' clinical description for resulting ambiguities. We also
evaluated scholarly publications concerning both the text and the meanings of
Thucydides' description of the epidemic.3 We considered textual matters, such
as those bearing upon the likelihood that Thucydides could have observed what

3These included: H. Haeser, "Die Pest des Thukydides (430 v. Chr)," Historische-
pathologische Untersuchungen. Als Beitrdge zur Geschichte der Volkskrankheiten, vol I,
Chapt. II (Dresden 1839) 32-57; J. Ehlert, De Verborum Copia Thucydides Quaestiones
Selectae (Berlin 1910); W. Nestle, "Hippocratica" Hermes 73 (1938) 1-38; D. L. Page,
"Thucydides' Description of the Great Plague at Athens," CQ 47 (1953) 97-119; A. W.
Gomme, HCT 11 145 ff.; C. Lichtenthaeler, Thucydide et Hippocrate vus par un historien-
medecin (Geneva 1965); G. M. Parassoglou, unpublished diss. (Yale University 1968); A.
Parry, "The Language of Thucydides' Description of the Plague," BICS 16 (1969) 106-118.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 273

he described, sequelae that could have been rare, unrepresentative, or

inaccurately described, the limitations of Thucydides' non-technical vocabulary
in describing medical conditions, and the likelihood that concurrent conditions
could have been misconstrued as being part of the symptomatology of the
epidemic disease. In all cases we placed primary emphasis on epidemiologic
information, drawing secondarily upon clinical information more liable to
error. Using this framework, we attempted to identify the disease by
identification of its mode of transmission, including evaluation of proposed
causes of the epidemic using mathematical models of known candidate diseases,
fitted with time/population data for Athens, and comparison of the epidemic
behavior of candidate diseases to their behaviors in actual premodem
epidemics. For this illustrative purpose we present the simplest of the
mathematical models, that of Maia,4 an early attempt to describe
mathematically the so-called "Reed-Frost theory" of how epidemics behave in
populations. It should be emphasized that our use of this particular
mathematical model should not be construed as an endorsement of its validity.
Definitive epidemic analysis using mathematical modelling is beyond the scope
and intent of this paper. In fact, empirical validation of this and more
sophisticated mathematical models is wanting for most diseases. Since each
observed set of epidemic conditions is unique, validation for a particular
disease in a particular epidemic setting would not necessarily allow
generalization to the model's overall validity. Furthermore, the epidemic in
Athens may have been quite different from epidemics of modem diseases. For
example, high mortality may have affected the human contact rate in ways that
cannot now be suspected. Newer more complicated mathematical models do a
better job of taking into account such factors as the elements of chance and the
changing contact rate. It is not clear, however, if these newer models are more
successful in predicting and describing the course of actual epidemics of highly
contagious diseases. The Maia model seems most appropriate for three reasons:
(1) it is the best known of many models; (2) it is the simplest, and by far the
easiest for non-mathematicians to understand, and (3) it requires fewer
assumptions.

Our approach emphasizes descriptive epidemiology, the characterization

of population health events in terms of "person," "place," and "time," as
described in standard epidemiology textbooks.5 This method, based loosely on

4J. D. 0. Costa Maia, "Some Mathematical Developments on the Epidemic Theory

Formulated by Reed and Frost," Hwnan Bioil. 24 (1952) 167-200.
5For example, J. S. Mausner and S. Kramer, "Descriptive Epidemiology: Person, Place, and
Time," Epidemiology-An Introductory Text (Philadelphia 1985) 118-153.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
274 David M. Morens and Robert J. Littman

the approach of Hippocrates (Airs, Waters, Places), characterizes

behavior in populations, rather than in individuals, as would be t
clinical approach. The epidemiologic approach includes deternination of which
types of individuals are susceptible to the disease (e.g. children vs. adults,
males vs. females), where the disease occurs (e.g. in rural vs. urban locations,
wet vs. dry areas), and when the disease occurs (e.g. winter vs. summer). Such
an approach can characterize individual diseases, and in modem epidemiologic
practice it is used to help identify the agents of epidemics of unknown cause.
Since pre-modern clinical descriptions are by their very nature ambiguous, the
epidemiologic picture may present a better means of identifying and
understanding diseases from the past than dissection of their ostensible clinical
features.

Epidemiologic Interpretation.

Person. Thucydides not only tersely states (2.51) that the disease "carried
away all alike," but implies that it spared no population subgroup, including
citizens, metics, slaves, and refugees. He does not mention clinical differences
in the inbred citizenry, in which first cousin marriages were typical, nor
occupational risk in food-handlers, hostlers, shepherds, or other persons with
animal contact. The army and civilian populations were infected alike, although
there may have been a somewhat lower case fatality rate in the cavalry than in
the hoplite ranks. The cavalry presumably included the wealthier citizens.
Thucydides says in one passage that persons with prior illnesses "all caught the
plague in the end" (49). Alternatively, we cannot rule out the possibility that
exposures associated with treatment of the prior diseases (e.g. exposures to
physicians) were associated with an increased chance of acquiring the disease.
Thucydides specifically links physicians to increased risk, without
distinguishing between the risk of getting infected (attack rate) and the r
dying after becoming ill (case fatality rate). Although Thucydides does not
comment upon the Athenian diet during the sieges, deficiencies in total
calories, protein, or most vitamins would have been unlikely because of grain
warehouses and other storage capability in the city, and an unblockaded port
(Piraeus) with large fleets of naval, commercial, and fishing vessels. But
without fruits or vegetables inside the city walls, and with crops in the
surrounding countryside either bumed or appropriated by the occupying
Spartan army, the Athenian diet would probably have been deficient in
ascorbic acid (vitamin C). Vitamin A deficiency would be less likely unless
some persons were also unable to obtain fish, fish products, fruits or
vegetables.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 275

Place. The origin of the epidemic cannot be determined with certainty.

According to Thucydides (48.1) it was reputed to have originated in Ethiopia,
spread to Egypt and Libya, and spread from there to the territory of the King
of Persia [much of the Middle East]; before reaching Athens (47), "previous
attacks...had been reported from many other places in the neighborhood of
Lemnos and elsewhere." Thucydides states (47) that there was no record of any
disease being so extensive or so destructive to men and that (54.5) Athens was
hardest hit. Livy (4.20-21; 4.25.3-4;.4.30.8-10) describes epidemics in Rome
in 433 and in 428. These may be part of the same pandemic. Plutarch (Per.
35.3) says the disease destroyed not only the Athenians, but also those who had
dealings with their forces. Whatever the epidemic's origin, it was probably
shipbome. Piraeus was the Mediterranean's major port at the time. Premodem
epidemics and pandemics of many diseases (cholera, dengue, plague, smallpox)
were typically spread by ships. Thucydides also says that the disease devastated
Hagnon's naval expedition to Potidaea around July 430 B. C., at the height of
the first epidemic wave. This expedition would have been under sail for about
five days. After sailing, landing, marching and encamping, Hagnon's army
suffered a paralyzing epidemic in Potidaea over a period of about six weeks.
Many of the 3,000 Athenians already there subsequently became infected as
well, although the besieged Potidaeans, reduced even to starvation and
cannibalism, apparently did not. Thucydides also noted directionality of spread:
after beginning in the port area of Piraeus, the disease spread up into the hills
of Athens, separated from Piraeus by a narrow corridor between the Long
Walls that was probably crowded with refugees from the countryside. Though
it was in these refugee camps that the epidemic is said to have flourished,
Thucydides did not distinguish between high attack rate (the epidemic incidence
rate; i.e. the number of new cases per population at risk per time), high case-
fatality rate (proportion of cases who died), or rapidity of epidemic
progression. He linked the explosiveness in the camps, however, to "poor
ventilation," consistent with either airborne spread, poor hygiene and
sanitation, crowding, or any combination of these. His term "ventilation"
probably says more about the (then) prevalent Greek belief in the miasmatic
theory of disease causation than about an actual risk factor. Some authors have
interpreted Thucydides to have meant that crowding was a risk factor for the
disease in Athens. Gomme believes instead that Thucydides meant that once the
epidemic broke out, it seemed all the worse because of the pre-existing
miseries associated with over-crowding (HCT 158). Even so, Thucydides
mentions crowding twice; here, and in his later comment about involvement of
other densely populated towns (54). Potable water was supplied to Piraeus by

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
276 David M. Morens and Robert J. Littman

cisterns and to Athens by numerous wells spaced throughout the city. Athens
also had cisterns, a river-fed aqueduct (from the Ilissus), and several springs,
most of which were said to have had brackish water. The bedrock underlying
Athens has porous limestone in the surface layer, underlaid by slate and marl.
As noted above, grain storage was probable during the Spartan sieges, if not at
other times. A curious item implicit in Thucydides' account of the war, and
mentioned explicitly in one passage, is that despite frequent contact with the
Spartan enemy, with the various allies in many regional cities, with the various
expeditionary enemies, and with numerous allies, among the mainland cities of
Greece only the Athenians were known to have suffered extensively from the
devastating epidemic disease. Despite periodic encirclement of the Athenians by
the Spartans, and frequent contact between the two enemies, the disease
apparently either did not spread immediately beyond the city, or if it did, on
when the Athenians themselves exported it. Thucydides (54.5) says that it did
not enter the Peloponnesus to any extent and its "full force was felt at Athens,
and, after Athens, in the most densely populated of the other places."

Time. The disease was not seasonal: it began soon after the Spartan army laid
siege to Athens in late spring/early summer, probably in early May 430 B. C.,
a time of heat and humidity. Subsequent epidemic waves were noted in the
summer of 428 B. C. and winter of 427-426 B. C. The epidemic apparently
ended after 4 1/2 or 5 years, in winter 426-425 B. C. Just before the first
epidemic wave, the Athenian population would have risen from at least
100,000 persons to about 300,000 or 400,000 persons as refugees streamed
into the city. Gomme cites the base population of Athens as 155,000, composed
of 60,000 citizens, 25,000 metics, and 70,000 slaves. Rostovtzeff believed that
during the sieges the population rose to 315,000, and Major to over 400,000.6
Many scholars appear to agree with the latter figure. The plague continued in
Athens after its appearance in 430 B. C., even throughout the summer of 429
B. C., when Attica was not invaded at all. Neither was Athens under siege
during the epidemic wave of winter 427-426 B. C., and the population would
thus probably have been closer to 100,000, certainly less than 200,000, if
refugees had returned to the countryside. Perhaps the most important set of
clues Thucydides provides is that about the epidemic's duration: it continued
uninterrupted for more than two, and perhaps for as many as four and a half to
five full years (i.e. at the least from spring or early summer 430 B. C. to
summer or fall 428 B. C.). Thucydides notes that it underwent an explosive

6A. W. Gomme, The Population of Athens in the Fifth and Fourth Centuries B.C. (Glasgow
1933) 22, 44; M. Rostovtzeff, The Social and Economic History of the Hellenistic World
(Oxford 1941) 1 95; R. H. Major, Fatal Partners, War and Disease (New York 1941) 9, 13.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 277

rekindling in the winter of 427-426 B. C., continuing for a year after that. In
reporting this latter epidemic wave Thucydides states "...the plague broke out
among the Athenians for a second time. In fact, it had never entirely
stopped...This second outbreak lasted for no less than a year, and the first
outbreak had lasted for two years" (3.87.3). It is uncertain, then, whether at
that point (winter, 427-426 B. C.) the epidemic had gone on continuously for
three and a half full years (spring or summer 430 to winter 427/426 B. C.), or
had actually stopped or had become barely detectable after two years of
continuous activity (428 B. C.), only to have restarted again in the winter of
427/426 B. C., and then prevailed for another year thereafter. There can be
little doubt, however, that it was prevalent for a very long time.

Other Epidemiologic Information. Thucydides clearly described a "virgin

soil" epidemic. The novelty of the disease is supported by Thucydides'
contention that it had not occurred previously, by signs and symptoms
Thucydides claimed were distinctive, and by the rapidity with which it spread
through the population. Population susceptibility is supported by the high
attack rate, and an unvarying course in persons of different age, sex, and
nationality, including an unsparing course in older persons. Although neither
the incubation period, attack rate, or the case-fatality rate are known, in about
July-August 430 B. C. the epidemic killed 26 per cent of an expedition of
4,000 hoplites in approximately 40 days (58). Elsewhere (3.87.3) Thucydides
cites a final Athenian total of 4,400 hoplite deaths (34 per cent), and 300
cavalry deaths (30 per cent). This may be indirect evidence that the first
epidemic wave (which killed 26 per cent of 4,000 hoplites) was by far the
worst: assuming proportional mortality in the rest of the hoplites, 78% of the
eventual 4,400 hoplite deaths would have occurred in the first wave.
Thucydides did not say how many of them were ill and survived, and he would
have had no way of knowing how many had been asymptomatically infected. In
any case, the data suggest an attack rate between 25-100%, and a case-fatality
rate more than 25% but less than 100%, since in the non-expeditionary
population the disease was not invariably fatal. Thucydides himself survived,
and there were apparently enough other survivors to suggest to Thucydides
that survivors were never attacked twice. Thucydides makes several remarks
that might seem to be clues to the mode of transmission, including comments
that physicians were at increased risk, and that they became ill from tending
the sick (the earliest surviving description of what has been cited as a concept
of contagion). He also ambiguously implies the possibility of zoonotic (animal)
disease: "...though there were many dead bodies lying about unburied, the
birds and animals that eat human flesh either did not come near them or, if

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
278 David M. Morens and Robert J. Littman

they did taste the flesh, died of it afterwards. Evidence for this
in the fact that there was a complete disappearance of all birds of prey: they
were not to be seen either round the bodies or anywhere else. But dogs, being
domestic animals, provided the best opportunity of observing this effect of the
plague" (2.50.1-2). Though Thucydides did not claim that anyone had seen ill
dogs or birds, some authorities have cited this passage as proof of en- or
epizootic involvement (low level circulation or outbreak transmission in
animals). His empiric observation, however, is merely absence of birds and
dogs about human corpses; he obviously expected to see them but did not.
Furthermore, the structure of this passage seems to us to imply a perceived
need to explain the unexpected absence of birds and dogs by offering a possible
reason for it (that they became ill too, just as humans did), suggesting that the
observation was not necessarily a fact Thucydides meant to preserve for
posterity. It has even been suggested that Thucydides inserted this passage
purely to bolster his contention that the epidemic disease was a new one; so
new, presumably, that even the dogs and birds were too cautious to approach
the carcasses.7 Epizootic or enzootic disease is thus highly questionable.

The Text of Thucydides. Our analysis leads us to conclude that Thucydides'

description of the epidemic disease is subject to potential error. A number of
difficulties in interpreting Thucydides' text suggest caution in accepting his
medical opinions, especially when they appear to conflict with factual or
epidemiologic information presumably less liable to misinterpretation 2,500
years later. Thucydides' description of the epidemic seems to jumble physical
signs and symptoms, epidemiologic observations, and historical facts. But it is
his accuracy in describing signs, symptoms and clinical features that is the focal
point of disputes about textual credibility. Translations of Greek lay terms into
English medical terms for distinct signs and symptoms may be subject to
considerable inaccuracy. Since there existed no standard or specific medical
vocabulary to describe what he had observed, Thucydides would have applied
"common" words to signs and symptoms, perhaps incompletely aware of the
extent to which multiple meanings or ambiguities might later confound
attempts to link the terms to medical concepts to be developed many centuries
later, and perhaps even unaware of which common terms might have been
favored by Hippocrates or other physicians.
Thucydides may also have chosen words that had special or colloquial
meanings since lost. Littre believed the account to have been written for the

7C. Anglada, E'tude sur les maladies eteintes et les maladies nouvelles pour servir a 1' histoire
des evolutions sgculaires de la pathologie (Paris 1869) 1-50.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 279

"vulgar," and without scientific merit.8 Harrison rues Thucydides'

"[characteristic] over-condensation of style," which she believes "leads us to
convict Thucydides of a real and unavoidable inexactness." She also notes
Thucydides' penchant for inserting "vague addend[a]" to cover points about
which he is uncertain.9 Thucydides, she concludes, "always leaves perhaps
rather much to the intelligence of his readers." Parry not only believed the
account was hampered by a non-technical vocabulary, but claimed that
Thucydides specifically avoided technical terms and wrote the epidemic
account not for purposes of scientific accuracy, but to dramatize the tragic
plight of the Athenians.10 A "medical" term that scholars appear to consider
least ambiguous is phlyktainai. It is often assumed that the epidemic disease was
bullous because Thucydides used a word (phlyktainai) that had been used to
describe spots on a loaf of bread, lesions on the hands of rowers, and other
things. While this may be an excellent description of bullae, it might also be a
description of other lesions for which the Greeks had no specific and unique
medical word. Scholars have used a number of English translations of
"phlyktainai": blains, blebs, blisters, bullae, eruptions, pimples, pustules,
vesicles, and whelks. Hippocrates apparently used "phlyktainai" to describe not
only bullae and vesicles, but pustules, bums, "sweat eruptions," and contact
dermatitis, and others writers have suggested the term includes papules,
plaques, and erythematous lesions as well.1' Shrewsbury argues that while
"phlyktainai" may typically mean "blisters," it is a general and inclusive term
applied to any raised eruption of any sort.12 A typical example can be seen in
the translation of the Hippocratic word sepedon (not used by Thucydides).
Some authors translate it as "gangrene", but Hippocrates used the word to
describe cellulitis and purulent bacterial infection. We thus question with what
assurance any such terms can be equated with modem medical terms, especially
when certain theories of the cause of the epidemic hinge largely on their exact
meanings. The theories suggesting smallpox, syphilis, and typhus, for example,
are built around connection of such terms (e.g. phlyktainai in the case of
smallpox) to modem terms for medical conditions.

8E. Littre, "Deuxieme livre des epidemies," Oevres completes d'Hippocrate, Vol. 5 (Paris
1839) 43-71.
9J. E. Harrison, "The Ancient City, Its Character and Limits" and "The Sanctuaries in
Citadel," Chpts. I and II in Primitive Athens as Described by Thucydides (Cambridge 1906
36, 37-65.
1?Parry (above, n. 3) 106-118.
1 IH. Keil, "The Louse in Greek Antiquity, with Comments on the Diagnosis of the Atheni
Plague as Recorded by Thucydides," Bull. Hist. Med. 25 (1951) 305-23.
12J. F. D. Shrewsbury, "The Plague of Athens," Bull. Hist. Med. 24 (1950) 1-25.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
280 David M. Morens and Robert J. Littman

It should be abundantly clear to any medical person readin

that the signs, symptoms, and features of the epidemic disease
Trying to put them together into a coherent picture of any disease, even a
hypothetical extinct disease, leaves an impression not unlike trying to put
together pieces of different puzzles: something, or perhaps some things, are
clearly wrong. But which ones are they? From the medical perspective, how
confident should we be in fitting together ostensible signs and symptoms when
even Greek linguistic experts fail to agree on, and sometimes strongly disagree
on, the meaning of important terms and phrases? What emphasis should be
placed on Thucydides' signs and symptoms of the Athenian disease when we
cannot decide between bleeding and congestion, pallor and jaundice, blindness
and ocular damage, vesicles and other eruptions, bloody and watery diarrhea,
convulsions and muscle spasms, retching and hiccuping, amnesia and dementia,
lividity and flushing, colic and tenesmus, distal gangrene and some other loss
of, or loss of the use of, the extremities? It is not even clear what killed the
victims: Thucydides states in one passage (49.6) that an "internal fever" was the
cause of death, or else diarrhea. But fever and diarrhea are non-specific
symptoms of illness, not causes of death. Without knowledge of physiology,
microbiology, pathology, or clinical medicine, what criteria would Thucydides
have used for determining the cause of death? Could he have distinguished
what people died with from what they died of? If not, his comments about
deaths from either "internal fever" or diarrhea may be of no more relevance
than to indicate either rapidly progressive courses, or more indolent courses
ending with non-specific and agonal features like diarrhea. When scholars
cannot even agree on whether Thucydides was referring, in one passage (49.6),
to symptoms in the heart, or in the stomach, there may be a need for caution in
any consideration of clinical features. Such problems may not be easily
overcome by clinical or linguistic research efforts. For this reason, too, we
favor the epidemiologic approach. Even so, we must be cautious in interpreting
Thucydides' comments on the epidemiology of the disease. We doubt that
Thucydides would have been able, for example, to distinguish such basic
concepts as virulence and case fatality. We should not, therefore, over-
interpret his words.
Moreover, some of the vocabulary problems that confuse us today may
also have hindered Thucydides, though presumably to a lesser degree.
Whatever trouble Hippocrates may have had with the lack of a precise medical
vocabulary would have been considerably worse for Thucydides, who was not
a physician, and thus would not have had experience observing and describing
diseases. It should also be noted that despite some familiarity with the

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 281

Hippocratic emphasis on prognosis, and with certain Hippocratic terms, the

vocabulary used by Thucydides to describe signs and symptoms of the Athenian
epidemic differs from Hippocrates' descriptions of other diseases. An
additional problem is that because Thucydides had the disease himself, his
account of it may have been colored by subjective experiences. He might, for
example, have stressed features he experienced, while de-emphasizing others
he did not. Gomme suggests that while Thucydides took notes on the epidemic
immediately after his own illness, he fleshed out the description of the
epidemic much later. If so, did he dress up the account with Hippocratic
trimmings? It is also clear that an implicit aim of his brief epidemic account
was to gain credibility with physicians by presenting the disease description in
a way that appealed to contemporary medical thought, and that fit (then)
current medical theories. His assertion that during the epidemic patients
succumbed on the seventh or ninth day, for example, is taken verbatim from
Hippocrates, who espoused the view that certain conditions led to demise on
those specific days (e.g. Aphorisms IV 36; IV 64). Or does the Hippocratic
statement (Aphorisms III 21) on the effect of time of the year on disease, "in
summer.. .we must also expect.. .gangrene of the genitalia," influence
Thucydides' account of loss (of use) of the genitalia due to the disease?
Although Thucydides claimed to be most interested in diagnosis, he tailored a
least a part of his description to the prognosis-oriented expectations of the
Hippocratic school, this despite the fact that he and his contemporaries were
well aware of the prognosis. Thucydides' tendency to describe those disease
features of greatest interest to Hippocratic followers, including formulaic
phrases interspersed in the text, casts some doubt on the validity of the
individual signs, symptoms, and features mentioned.
Finally, the organization of Thucydides' text on the Athenian epidemic
suggests an unsatisfactory synthesis of multiple aims, and leaves the impression,
if examined carefully, that Thucydides was not at all sure what the features of
the disease were, or how to select, present and organize them in narrative
form. After beginning his discussion with several paragraphs of descriptive
epidemiology, he next describes the onset of illness, and then an apparent
chronological course of disease progression: "..It began with.. .The next
symptoms were.. .and before long.. .next the stomach..." (49). This chronology
is then abruptly interrupted by descriptions of apparently different courses
taken by the illness: "Most patients suffered...Whenever it settled...In some
cases ...in others ...many...actually [jumped into wells], the majority ...or else...In
most cases there were...this sometimes ended.. .but sometimes continued..."
(49). Later, Thucydides appears to be describing complications and sequelae,

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
282 David M. Morens and Robert J. Littman

though he would not have had such concepts in mind. He states, for example,
that "...if people survived this critical period, the disease descended...some,
too, went blind...There were some also who.. .suffered from a total loss of
memory..." (49). The description obviously lacks any means of distinguishing
between different courses of the disease, complications, sequelae, and
supervening infections, had any or all of these occurred.
Thucydides also contradicts at least some of his earlier descriptions by
introducing a "head-to-toe" organizational structure that competes with his
initial chronology: "...first settling in the head, went on to affect every part of
the body in turn..." (49). In fact, the competing concepts of chronology and
head-to-toe progression are interwoven unsatisfactorily throughout his
description. In our view, the disease features seem difficult or impossible to fit.
Excluding consideration of atypical courses and complications, sequelae, and
supervening illnesses, it is probably fair to generalize in saying, for example,
that potentially fatal gastrointestinal diseases (e.g. cholera, shigella dysentery)
do not cause the sorts of respiratory symptoms Thucydides describes and, on
the other hand, that fatal respiratory diseases (e.g. influenza) do not cause the
types of gastrointestinal features Thucydides records. In a similar vein, diseases
that make people sneeze normally do not make their genitalia fall off. Previous
critics, grappling with such paradoxes, have sought one of three escapes: (1)
either "the" disease was many different diseases confused as one, or (2) it was
just as Thucydides described it, but is now extinct, or (3) it was one disease
after all, but Thucydides erred in his description of one or more of the
features. We believe the last of these possibilities. On genetic and evolutionary
grounds, we strongly doubt that the disease could have become extinct. While
most microorganisms evolve quickly, they are highly adapted to humans or
other hosts that evolve slowly, thus avoiding the threat of extinction unless the
hosts become extinct themselves. Measles, for example, has apparently been
prevalent for at least 1,000 years, plague for at least 2,000 years, and
poliovirus diseases for at least 3,000 years. Each of these diseases appears not
to have changed at all. Even smallpox has only undergone one recognized
change in clinical form in its 3,000 or more years of prevalence, and this
change (a reduction in virulence) was predictably in the direction of assuring
its survival, rather than hastening its extinction. There is thus little empiric or
theoretical reason to suspect extinction of any human disease except by
purposeful eradication, as with smallpox. Although it is possible that other
concurrent diseases contributed to some of the supposed signs and symptoms,
we also doubt two or more diseases because of the difficulties in explaining a
remarkably concordant time course, because Thucydides himself and the

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 283

Athenians in general apparently believed it was one disease, because it was

confined to the Athenians, and because its reappearance over a five year period
seemed to be associated with the same features.

From the above information we may make general inferences about the
mode of transmission of the Athenian epidemic, eliminate most candidate
diseases from consideration on this basis, and then make additional inferences
about the suitability of diseases that remain. We compared three possible modes
of transmission to the observations of Thucydides: common source, person-to-
person (enteric, inoculation, and aerosol/respiratory), and "reservoir-
associated". The term "reservoir" is taken to mean an animal, insect, or
environmental source in which an infectious agent is maintained when not
infecting humans.

Common-Source Acquisition. Common-source acquisition is not

consistent with epidemic persistence over several years time, or Athenian
exportation to Potidaea. In open populations, common source epidemics are
nearly always foodbome or waterbome. Ergotism has been suggested as a
cause of the epidemic,'3 in part because it causes peripheral gangrene, but there
is no reasonable explanation for such massive and simultaneous contamination
of all grain sources, more frequent (or worse) illness in refugees, lack of
differences in illness by age, and documentation of geographic spread within
the city. Grains would probably have been consumed in less than two to five
years, especially during a siege characterized by unimpeded access of the
Athenian fleet to the many Aegean grain-producing areas. Contamination of
harvests in so many successive years, from such varying sources, is difficult to
imagine. The same grains were also presumably consumed by others who did
not experience epidemic disease, including the Spartans, who had appropriated
the Athenian crops. Similar arguments rule out other foods in the Greek diet,
including other grains, fish, and oils. As noted above, destruction or
appropriation of ascorbic acid sources would probably have placed the
Athenians in a state of vitamin C deficiency, at least temporarily. But the
epidemic began too soon after the probable cut-off of ascorbic acid-containing
fruits and vegetables to implicate scurvy. It first broke out at the beginning of
the Spartan siege (late spring or early summer 430 B. C.), making scurvy an
unlikely cause if, as we assume, it would take at least several weeks, if not
months, for symptoms to develop. Scurvy could have altered or confused the

13A.-P. Read, "Epidemies occasionees par l'usage du seigle ergot'," Traite6 du seigle ergote
dans lequel on examine les causes de cette excroissance veg6tale, les moyens de la prevenir, les
resultats de l'analyse de ces grains, leurs effets sur les animaux, les maladies epid6miques
qu'occasione leur usage, & le traitement qu'elles exigent, Part 3 (Strasbourg 1771) 52-93.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
284 David M. Morens and Robert J. Littman

clinical picture later on, however. Symptoms of scurvy, documented since

biblical times,14 are similar to some of those described by Thucydides (e.g.
fever, oral bleeding, red eyes, hyperkeratotic skin lesions, petechiae, impaired
wound healing, melena, and irritability/agitation) and could have produced
features mistaken as disease complications. Recently, clinically inapparent
vitamin A deficiency has been associated with worsening of severity of measles
and other infectious diseases.15 But despite its association with blindness, skin
lesions, and diarrhea, vitamin A deficiency seems unlikely because of the
probable presence in Athens of vitamin A-containing products, including fish,
during the epidemic.
Waterborne disease is even more difficult to defend, since the epidemic
began in Piraeus, a port whose cisterns were unconnected to other water
sources and unassociated with sewage channels, thus providing no opportunity
for wholesale contamination of potable water. Furthermore, the disease spread
from Piraeus to Athens. With most of the Athenian wells on high ground, and
with a soil underlaid with porous limestone, the direction of groundwater flow
in 430 B. C. would have been, as now, downhill: any contaminant introduced
into the water would have flowed downhill towards Piraeus (or to the north
and west), not uphill towards Athens. There would also seem to be no practical
way (except, conceivably, by avian contamination) for the many hundreds of
Athenian wells to be simultaneously cross-contaminated at the onset of the
epidemic, or to remain contaminated over such a long period of time.
Historically, massive enteric disease epidemics have usually been associated
with sophisticated water and sewage systems, e.g. the London cholera
epidemics of the 1840s and 1850s. Zoonotic diseases that are also associated
with "common source" exposures (e.g. anthrax) are considered below under
the heading of "Reservoir Diseases." Common source acquisition of a single
epidemic disease in 430-425 B. C. can otherwise be ruled out, and with it
etiologic theories for cholera, dysentery, ergotism, shigellosis, scurvy, and
typhoid fever.16

14j. H. Swanson, "Evidence of Scurvy Among Ancient Hebrews" Bull. Hist. Med. 15
(1944) 352-8.
15N. S. Scrimshaw, C. E. Taylor, J. E. Gordon, Interactions of Nutrition and In
WHO Monograph Series. No. 57 (Geneva 1968); G. D. Hussey and M. Klein, "A
Randomized, Controlled Trial of Vitamin A in Children with Severe Measles," New Engl. J
Med. 323 (1990) 160-4; L. Rahmathullah, B. A. Underwood, R. D. Thulasiraj, et al,
"Reduced Mortality Among Children in Southern India Receiving a Small Weekly Dose of
Vitamin A," New Engl. J. Med. 323 (1990) 929-35.
16Cholera: F.-R. Chateaubriand, Mgmoires d'outre-tombe (Paris 1849). Dysentery: F.
Kanngiesser-Neuenburg, "Die seuche des Thukydides (Typhus exanthematicus)," Munchener
Med. Wochenschrift 63 (1916) 1627. Shigellosis: C. G. Ray, "Death in Athens: The epidemic

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 285

Person-To-Person Transmission. Of the three basic subcategories of

person-to-person transmission, enteric ("fecal-oral"), inoculation (venereal
and non-venereal), and respiratory/aerosol, the first two can be ruled out.
Enteric transmission is even less likely than common-source acquisition of
enteric organisms, already discounted. Though organisms transmitted by the
waterbome route are typically also transmitted enterically (e.g. cholera), in
open populations enteric transmission is, for obvious reasons, much less
efficient in bringing microorganisms in rapid contact with large numbers of
people. A purely enteric disease (i.e. one not simultaneously waterbome) could
not possibly spread through tens of thousands of people in a few weeks. The
same can be said of venereal inoculation, even assuming multiple introductions
from returning navies. Thucydides' alleged description of "gangrene" of the
genitalia may, as noted above, actually refer to some other condition. Non-
venereal transmission by inoculation is also unlikely. An inoculum source
widespread enough to cause such an explosive epidemic in over 100,000 people
is difficult to imagine. Most human non-venereal inoculation diseases are
zoonotic, as discussed below. Exclusion of person-to-person enteric and
inoculation diseases eliminates, as noted above, cholera, "dysentery,"
shigellosis, and typhoid fever, and also eliminates syphilis, proposed as a cause
of the Athenian epidemic in the mid 19th century.'7
Consistent with person-to-person aerosol/respiratory transmission is the
explosiveness of the Athenian epidemic (with high attack rate and rapid
spread); the apparent lack of correlation of illness and either age, sex, or
occupation; an association with crowding; and the supposed predominance of
upper respiratory symptoms at illness onset. But most respiratory diseases,
including virtually all that occur in explosive epidemic form, rapidly die out in
closed crowded populations. Universal susceptibility may initially lead to
explosive epidemic progression, but as the epidemic continues, the declining
availability of susceptible persons slows further progression, until the epidemic
ends for lack of new susceptibles: depending upon the ease of disease
transmissibility either all persons have become infected and have, therefore,
died or become permanently immune, or else only some persons have become

of 430 B.C.," Symposium 2 March 1984, Arizona Health Sciences Center. Tape A/7 114,
Media Department, Health Sciences Center Library, Tucson, Arizona. Scurvy: E. H. Smith,
"The Plague of Athens" in S. L. Mitchell, E. Miller, and E. H. Smith, The Medical Repository,
2nd ed., Vol. I (New York 1800) 3-32. Typhoid fever: C. Pfeufer, "Geschichliche Bedeutung
des Scharlachs," Der Scharlach. Sein Wesen und seine Behandlung, mit besonderer
Beruchsichtigung des 1818 zu Bamberg herrschenden Scharlachs (Bamberg 1819) 1-10.
17J. Rosenbaum, Geschichte der Lustseuche im Alterthume (Halle 1845).

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
286 David M. Morens and Robert J. Littman

infected, and the rest are protected by "herd immunity," a population immunity
threshold, characteristic of each disease, that blunts or prevents epidemic
transmission. Respiratory disease epidemics in distinct and crowded
populations thus normally peak and end quickly, progressing with predictable,
even mathematical regularity. But despite a closed population with both limited
immigration and numerical attrition due to war, Thucydides stated
unequivocally that the Athenian epidemic went on for at least two, and
probably four or five years. Furthermore, despite considerable contact
between the two armies, the Spartans appear not to have been affected by the
disease. To invoke a respiratory epidemic of explosive onset, it is necessary to
explain both its persistence over long periods of time and its apparently
imperfect transmissibility to other persons with whom the Athenians
undoubtedly came into contact during the epidemic years.
To understand better the behavior of respiratory diseases in human
populations, it is helpful to consider certain epidemiologic concepts that bear
upon the description of the Athenian epidemic. Since there is variation in both
the incubation periods of diseases, and the percentages of the population
susceptible to diseases at any given time in their epidemic courses, the rate of
disease spread varies also. Incubation periods often tend to approximate the
time from infection until the infected person becomes contagious, which may
in turn approximate the serial generation times-the mean time intervals
between peaks of successive epidemic waves as observed in actual epidemics. If
a disease takes two days to render its victim capable of transmitting to others, it
may spread faster than a similarly infectious disease that takes 10 days to
become contagious. But if, for example, the two day disease infects only 10%
of those exposed to it, while the 10 day disease infects 20%, their rates of
progression in a population of susceptible persons may be difficult to predict
without mathematical formulas. Using these so-called "mathematical models"
we can predict for any contagious disease mathematical patterns of occurrence
over specified time periods under given conditions (e.g. population size and
degree of crowding). We can apply such theoretical patterns of specific
diseases to the Athenian epidemic, as described by Thucydides, to look for
concordance.

To evaluate further respiratory transmission we applied simple versions

of "mathematical models" of known diseases, using information on Athenian
population size, crowding, and susceptibility, to predict the time it would take
for them to die out. Respiratory diseases of short incubation (influenza) and
long incubation (measles, smallpox) are compared below, using the

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 287

straightforward deterministic mathematical model of Maia. The equation used

to generate the predicted epidemic curves is:

Ct+j = St (1 - qCt)
where:

Ct = number of cases of the disease at time t, the beginning of the

epidemic;

Ct+j = number of cases of the disease at time t + 1

t = the chosen time interval; here the serial generation time;
St = the number of susceptible persons at time t

q = i-p
p = the probability of 'adequate contact' (i.e., adequate to cause
infection) between any two individuals per time interval t.

For each disease the following assumptions are made: (1) there is
universal susceptibility at the outset, with infection either killing or conferring
permanent immunity (assumptions about case fatality thus have no bearing on
the epidemic curves, since death and long-lasting immunity are equivalent
barriers to further transmission); (2) the serial generation time is taken to be
4.5 days for influenza, 12 days for smallpox, 14 days for measles, and 19 days
for streptococcal infections such as scarlet fever;18 (3) the total population
figure for Athens is taken to be the minimum of 100,000 as noted above
(during most of the early war years the population probably alternated between
about 100,000-200,000 and 300,000-400,000 as refugees entered and left the
city during the siege'9); and (4) a range of "adequate contact" numbers is
selected to approximate conservative estimates of the frequency of
interpersonal contact under conditions of known severe crowding. Although an
adequate contact figure is difficult to arrive at with precision for any situation,
we can estimate it (only very roughly) from at least two sources: data on
population crowding indices, and from empiric back-calculation using
Thucydides' information about the outbreak in Hagnon's naval expedition.
The area of Athens/Piraeus in 430 B. C., including the sparsely populated
port and storage areas, was about four square miles, suggesting a population
density ranging from 25,000 persons per square mile (about the same as New
York City today), to about 100,000 per square mile (like modem Delhi). These
modem comparison figures, of course, describe populations living in high-

18R. M. Anderson, "Directly Transmitted Viral and Bacterial Infections of Man," T

Population Dynamics of Infectious Diseases: Theory and Applications (New York 1982) 1-
19It has been argued, however, that after the first siege large numbers of persons from t
countryside probably remained in the city year-round (Gomme, HCT).

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
288 David M. Morens and Robert J. Littman

and low-rise buildings, which greatly reduce the degree of crowding. It is also
clear that Athenians were not equally dispersed, leading to pockets of even
greater crowding. Webster estimates that 30% of the square mile of Athens
proper was normally unoccupied.20 Thucydides implies that the narrow passage
between the Long Walls connecting Athens and Piraeus (less than a square
mile) was heavily populated by the refugees, and we may speculate that Piraeus
(about two square miles) was less populated. Even if the minimum estimated
population of 100,000 had been maximally dispersed over the entire land mass
of Athens/Piraeus, each individual would only have occupied a space equivalent
to a square patch of ground considerably less than 35 by 35 feet in size.
Assuming a more probable population of 400,000 would assign each person a
square patch of only 16 feet on a side, about the length of a compact
automobile. But in reality, population dispersion is never maximal, so that
considerably greater crowding than these estimates would be expected. For
example, in crowded populations persons are crammed into houses, rooms, and
public places, while other areas are unoccupied, leaving clusters of densely
crowded areas.21 Indeed, if Xenophon is correct in assuming 10,000 houses in
Athens, there may have been at least 10 persons per average household before,
or as many as 40 during the sieges, the majority of whom were presumably
crowded into the refugee camps. (Estimates of 400,000 Athenians in 10,000
dwellings have been widely accepted.)22 Attica itself, or at least the part
surrounding Athens from which the refugees came, is only thought to have
contained about 250 square miles of land. Another potential problem in
interpretation is that in Athens/Piraeus the epidemic would probably not have
been detected until some time after it actually began, confounding
interpretation of Thucydides' comments on its duration. However, there is no
such difficulty with the Hagnon expeditionary data if, as assumed here, the
expedition would not have sailed from Piraeus with hoplites suffering from the
disease on board. Thus, in estimating adequate contact by this means, no
assumptions need be made about late epidemic recognition. In 4,000
expeditionary hoplites, plus 3,000 hoplites previously encamped in Potidaea

20T. B. L. Webster, "Attica and Its Population," Athenian Culture and Society (Berkeley
1973) 35-57.
2tCf. Webster 40: "The city of Athens was...a jumble of narrow streets with houses, priv
or partly industrial, and shrines closely huddled together...in such a crowded city news spr
fast, and one could not help knowing what one's neighbor was doing...The very crowdin
Athens...made meeting certain."
22R. E. Siegel, "Epidemics and Infectious Diseases at the Time of Hippocrates. Their
Relation to Modem Accounts," Gesnerus 17 (1960) 77-98.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 289

(Phormio and his 1,600 men had already left), the epidemic curve went from
the index case to extinction in about six weeks.

In the Maia equation the calculated value of 'p', which changes as the
epidemic proceeds (i.e. as 't' increases), is a function of two independent
probabilities: the probability that any two individuals, regardless of immunity
status, will contact each other during a serial generation time, and the
probability of disease transmission should contact between an infectious person
and a susceptible person occur. If, for example, during a smallpox epidemic an
average infectious Athenian came "in range" of 25 persons per 12 days, and
was capable of transmitting the infection to only 20% of those who were
susceptible, there would be five "adequate contacts" per 25 susceptible persons
infected every 12 days. Because estimation of the chance of adequate contact is
somewhat speculative, in the illustrations below we present a range of adequate
contacts (20, 10, 5, and 2) that we consider to be characteristic of low to
extremely low transmissibility. A rough independent check on the
reasonableness of these estimates may be found in the form of a "back-
computed" adequate contact number from information Thucydides provided on
Hagnon's naval expedition to Potidaea. However, such an example is illustrative
rather than definitive: we have no reliable information concerning different
epidemic conditions in Athens versus the Potidaean camps. A mathematical
description of a single occurrence may be inadequate to predict what would
happen in similar circumstances or with repeated trials. In Hagnon's
expeditionary outbreak, Thucydides reported fatal disease in 1,050 (26%) of a
4,000 man hoplite force in about six weeks. Application of the Maia model to
this outbreak, back-calculated with a 12 day serial generation time, would
suggest that an adequate contact number of at least 30-40 would have been
required to end the epidemic in six weeks, even assuming that (a) none of the
hoplites had yet become immune two months or so into the epidemic (Figure
1), and that (b) the 3,000 previously encamped hoplites were not involved. If
they were involved, as seems almost certain, the adequate contact number
would have had to have been even greater for the epidemic in all 7,000 men to
extinguish in six weeks. Thus the essential conservatism of our range of
estimates of adequate contact numbers appears to be confirmed. Note that in
our calculations an adequate contact number of 10 for a disease with a 12 day
serial generation time (like smallpox) is equivalent to an adequate contact
number of 4 for a disease with a 4 1/2 day serial generation time (like
influenza). That is, we assume constant rates of contact between people.
The results of mathematical modelling in predicting the duration of
epidemics of various transmissible diseases proposed as causes of the Athenian

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
290 David M. Morens and Robert J. Littnan

epidemic should not be surprising to scholars who have studied actual

epidemics in defined premodem populations. For example, under virtually any
condition of crowding and contact believed to have prevailed in Athens of 430
B. C., influenza would have spread through the entire population and then died
out in about nine weeks (Figure 2). Similarly, epidemic smallpox is predicted
to have extinguished in but a few months (Figures 2 and 3). With 10 adequate
contacts, for example, epidemic smallpox is estimated to last less than four
months, leaving only about five persons untouched and still susceptible (Figure
3). With five adequate contacts, smallpox would be expected to die out in about
five months, leaving only 698 persons still untouched (Figure 3). This is
obviously too few to support subsequent epidemic transmission, even with a
high birthrate in Athens. (The birthrate in ancient Athens is unknown; but even
in the poorest countries today it normally ranges from only 3-5% of the entire
population per year). Since the disease was not solely a pediatric disease in its
later years, however, a large number of adult susceptibles must have existed
later on. Even had smallpox in Athens been a disease of extraordinarily low
transmissibility (two adequate contacts), the mathematical model does not
predict disease persistence in Athens for as long as a year, let alone two to five
years (Figure 3), though in this scenario 20,318 persons would remain
susceptible at the end of the epidemic. Measles, with a slightly longer serial
generation time (14 days) would likewise have rapidly been extinguished in
Athens unless it was of extremely low transmissibility (Figure 2). This is in
accord with the accepted dogma that year-round measles transmission does not
occur in populations under about 300,000-500,000 persons, even with modem
sanitation and lack of crowding.23 Epidemic streptococcal disease (scarlet
fever, erysipelas,24 or both) is more difficult to exclude because of the lack of
information about its behavior in susceptible populations.25 By the time scarlet
fever was distinguished from diphtheria (after Bretonneau's 1826 description

23J. A. Yorke, N. Nathanson, G. Pianigiami, J. Martin, "Seasonality and the Requirements

for Perpetuation and Eradication of Viruses," Am. J. Epidemiol. 109 (1979) 103-23.
24Scarlet fever: F. Schnurrer, Die Krankheiten des Menschengeschlects historisch und
geographisch betrachtet. Chronik der Seuchen, in Verbindung mit den gleichzeitigen Vorgdngen
in der physischen Welt un in der Geschichte der Menschen (Tulbingen 1823). Erysipelas: J.
Malfatti, "Beschreibung eines bosartigen Scharlachfiebers, welches zu Wien im Jahre 1799
unter den Kindbetterinnen geherrscht hat, nebst einigen Bemerkungen," Journal der practischen
Arzneykunde und Wundarzneykunst 4 (1801) 120-152.
25A. R. Katz and D. M. Morens, "Severe Streptococcal Infections in Historical Perspective,"
Clin. Infect. Dis. 14 (1992) 298-307.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 291

of diphtheria,26 and more universally after the 1854-1859 diphtheria

pandemic), it had already become endemic in urban centers of most developed
nations. But with a prolonged mean serial generation time (19 days) due in part
to long-term infectivity (despite its short incubation period), streptococcal
disease is perhaps, on epidemiologic grounds, a candidate for further research,
despite apparent clinical inconsistencies. The mathematical model suggests,
however, that in part because of its long mean serial generation time, epidemic
streptococcal disease could have persisted in Athens longer than measles or
smallpox. Assuming, for example, 10 adequate contacts per 12 days (equivalent
to 16 contacts during the 19 day serial generation time of scarlet fever), the
epidemic would have died out in less than five months (Figure 2). But
streptococcal disease is one of the few infectious diseases that can attack
persons more than once. This would facilitate disease persistence for several
years. However, against identification of this disease is Thucydides' claim that
the disease never attacked twice.

This mathematical model of diseases transmitted solely by the respiratory

route does not adequately explain the Athenian epidemic. Even if they had once
been, for some obscure reason, poorly transmissible, none of these diseases
(e.g. influenza, measles, smallpox) would have persisted two to five years, as
did the Athenian epidemic. Etiologic theories of diseases transmitted purely by
the respiratory route (including some with combined inoculation transmission)
are thus difficult to accept as causes of the Athenian epidemic, among them
influenza and influenza-associated diseases (including Guillain-Barre
syndrome and influenza complicated by staphylococcal toxic shock syndrome),
measles, meningitis, smallpox, staphylococcal diseases (including toxic shock
syndrome), and probably the unidentified "sweating sickness" of 16th century
England and continental Europe.27 The only windows of possibility for
respiratory diseases are provided by either unusual mechanisms for extra-
human persistence, or parallel explosive urban/non-explosive rural
transmission. Both of these possibilities are addressed below.

26P.-F. Bretonneau, Des inflammations speciales du tissu muqueux, et en particulier de la

diphthgrite, ou inflammation pelliculaire, connue sous le nom de croup, d'angine maligne,
d'angine gangreneuse, etc. (Paris 1826).
27Meningitis: J. L. Lefevre-DouviII6, Essais Medico-litteraires sur les Anciens (Paris 1858).
Smallpox: A. Kircher, Scrutinium physicomedicum contagiosae luis quae dicitur pestis (Rome
1658); J. Freind, "To Dr. Richard Mead," Two Epistles to Dr. Richard Mead: The One
Concerning Purging in the Second Fever of the Confluent Small-pox: The Other Concerning
Some Particular Kinds of the Small-pox (London 1729) 1-120; R. J. Littman and M. L.
Littman, "The Athenian Plague: Smallpox," TAPA 100 (1969) 261-75. Sallares (above, n. 1)
244-262. Sweating sickness: J. Caius, A Boke, or Counseill Against the Disease Commonly
Called The Sweate, or Sweatyng Sicknesse (London 1552).

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
292 David M. Morens and Robert J. Littman

Reservoir-Associated Acquisition. The fact that the Athenian epidemic

began explosively and lingered and resurfaced in a closed population over the
course of two to five years without disappearing, and perhaps without greatly
affecting others outside the city, suggests an extra-human mechanism for
persistence of the agent responsible for the disease. Continuous infectious or
non-infectious common source exposures (e.g. a waterbome enteric agent, or a
foodborne source such as ergotism) have already been discounted, leaving
infectious diseases with either animal reservoirs or insect vector reservoirs, or
both. Many such diseases in which the causative organisms have established
zoonotic cycles involving man as an accidental or "dead end" host (e.g.
glanders and rabies), are not otherwise consistent with the epidemic. But for
those in which the organism is co-adapted to humans or other primates (e.g.
yellow fever, dengue, Rift Valley fever), persistence in the vector host for
prolonged periods with or without epidemic or endemic activity is typical. Of
the latter diseases, many produce remarkably explosive epidemic curves,
surpassing in explosivity even those of influenza. Several of these reservoir
diseases, addressed below as possible causes of the Athenian epidemic, meet
two of what we consider Thucydides' most basic criteria: long-term
persistence and explosive/re-emergent potential. "Reservoir-associated" dis
transmission appears to be the only "pure" mode of transmission consistent
with the observations of Thucydides. The principal argument against it is
Thucydides' implication of person-to-person transmission, including his note
that the epidemic raged in the most crowded areas. However, as discussed more
fully below, vectorborne diseases frequently mimic the epidemiology of
explosive respiratory diseases to a remarkable degree, even in their marked
association with crowding.

Discussion.

Epidemiologic analysis of the Athenian epidemic is consistent with an

infectious agent associated with either (a) an animal or insect reservoir, or (b
respiratory transmission combined with a "reservoir-like" mechanism o
persistence. This conclusion is strongly supported by Thucydides'
documentation that the epidemic continued without interruption for two t
years in Athens without noticeably affecting either the surrounding Spar
army or without, presumably, affecting many or most of the hundreds of
thousands of other persons the Athenians would have come into contact with
during the war years, and by the fact that when the Athenians did export the
disease outside the walls of Athens (e.g. to Potidaea) only they appear to have
been affected by it in any great degree. These two possible categories of

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 293

transmission of the agent of the Athenian epidemic are discussed in greater

detail below.

Respiratory Transmission. As noted above, diseases transmitted only by the

respiratory route can be discounted since they do not "over-winter," and
would have spread rapidly in the extraordinarily crowded Athenian population.
Comment on several of the specific disease examples follows. Despite our
caution in interpreting mathematically derived data, it is worthwhile to note
that in modem times the validity of certain mathematical models in predicting
the course of influenza epidemics has been strongly supported by observational
and empirically derived data.28 There is thus reason to expect that the course of
such an epidemic in a natural population, even one occurring more than 2,000
years ago, would be much as the models predict. Rapid extinction in Athens is
also supported by consideration of hundreds of influenza epidemics
documented over the last three centuries, including those reported by
countries, states, counties, and cities. In 1918, for example, influenza spread
through, and died out in Newark, New Jersey (population 435,000) in about 12
weeks. A similar pattem was observed in most American cities of this size in
1918. The 1918 epidemic, the most highly fatal ever recorded, spread through
and died out in the entire United States-with a land mass nearly a million
times larger than that of Athens in 430 B. C., with 10,000 times the population,
and with a crowding index only one two thousandth as much-in about six
months. The Athenian population would probably have had lower standards of
sanitation and hygiene, no awareness of barrier protection or isolation, lower
standards of medical care, and a greater burden of concurrent diseases and
instances of under-nutrition. With respect to the recent theory that influenza
complicated by a toxic shock syndrome-like disease caused the Athenian
epidemic it might also be added that explosive staphylococcal epidemics are not
known to occur. Only about 10-20% of persons can be chronically colonized
with staphylococci; in open populations this apparently proceeds over
prolonged periods. Colonization with some unique strain, e.g. a toxic shock
syndrome toxin-producing strain, would be even less rapid were it to occur.
Invoking influenza/toxic shock syndrome as the cause of the Athenian disease
requires three epidemics: of influenza, of toxin-producing staphylococci, and
of impetigo. A basic tenet of epidemiology is that unknown epidemics are
rarely caused by two diseases. The lack of association of impetigo with toxic

28N. T. J. Bailey, Mathematical Theory of Infectious Diseases and Its Applications (Londo
1975); C. C. Spicer, "The Mathematical Modelling of Influenza Epidemics" Br. Med. Bull. 35
(1979) 23-28.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
294 David M. Morens and Robert J. Littman

shock syndrome and the lack of association of the Athenian epidemic with
pneumonia are additional factors against this possibility.
Although measles is more difficult to exclude on epidemiologic grounds,
mathematical models suggest that it, too, would not have persisted for two to
five years in Athens. This conclusion is also supported by contemporary
empiric observations and accepted mathematical modelling studies; even in
relatively uncrowded populations 300,000-500,000 persons are considered
necessary to sustain measles transmission. As shown in Figure 2, unless it was
less transmissible in a virgin population in 430 B. C. than is observed today,
measles would have extinguished in a matter of months. When in addition we
consider the crowding and sanitary conditions cited above, fitting measles and
the Athenian epidemic becomes increasingly difficult.

Smallpox is the most difficult of the respiratory diseases to exclude, not

the least because it has a long incubation period and is less transmissible than
either measles or influenza. In considering smallpox in our mathematical
model, we assumed a 12 day serial generation time, universal susceptibility, a
range of 2-20 adequate contacts, and a minimal base population/crowding
index of 100,000 persons per four square miles. Under such assumptions the
mathematical model indicates that a smallpox epidemic would die out in less
than 11 weeks given 20 adequate contacts; in 22 weeks given five adequate
contacts; or, at the theoretical extreme, as long as about 11 months given two
adequate contacts. This theoretical extreme we believe to be all but impossible.
The mathematical model thus corroborates historical records documenting that
in "virgin" populations, or in those of low or absent immunity, especially
under crowded conditions, smallpox epidemics are deadly and brief.
Information gathered at the time of the introduction of smallpox into Iceland
1707, for example, supports this view; within the span of a few months
smallpox killed about 18,000 of the country's 50,000 population, which was
spread out over 40,000 square miles.29 Similarly, in Aztec Mexico in 1520
between 3.5 and 15 million of 25-30 million persons, spread out over vast
areas, are estimated to have died in an epidemic of less than six months.30
Other examples of the devastation of smallpox in virgin and in relatively
circumscribed populations abound: virtually all of these epidemics came and
went in a matter of months. The possibility of a long duration smallpox
epidemic in Athens is further confounded by the Potidaean expeditionary

29K. B. Roberts, "Smallpox, An Historic Disease," Occasional Papers in Medical History

No. I (St. Johns, Newfoundland 1978).
30See J. A. Magner, Men of Mexico (Milwaukee 1942); F. F. Cartwright, Disease in History
(London 1972).

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 295

epidemic, in which 1,050 of 4,000 hoplites died in about six weeks. Using the
Maia model to describe the Potidaean outbreak would suggest an adequate
contact number of 30-40 persons by "backwards" calculation. If exposure risks
had been identical in the first Athenian epidemic (in fact, they may have been
even greater, since unlike the Potidaean forces none could have been immune),
a figure of only 20 adequate contacts would predict extinction of smallpox in
less than eleven weeks. Because, however, there are no objective data allowing
fair comparison of exposures during the original Athenian epidemic and the
ensuing Potidaean outbreak, it is difficult to predict whether, and in what
specific ways, crowding might have been worse in Athens than in military
camps in the Potidaean countryside.
A crucial paradox to be confronted in all considerations of respiratory
transmission is how a respiratory disease, explosive enough to devastate a
hoplite expedition in six weeks, and a good part of the Athenian population in
little more than that, can also be sufficiently indolent to linger within
Athens/Piraeus for over two years, and perhaps for as long as five years. In an
attempt to escape this paradox, we must also consider the possibility that the
Athenian epidemic was caused by an explosive respiratory disease also capable
of persistence via either traditional "reservoir" mechanisms, or by indolent
rural transmission. Smallpox fits both of these criteria. Unlike most other
respiratory diseases, viruses in dried smallpox secretions can survive for at
least several months in clothes or bed linen, or even in such inanimate sources
(fomites) as cotton bales,31 suggesting a theoretical means of long term
persistence of the agent beyond the chain of human-to-human transmission in
Athens. So well documented is persistence in fomites that during the French
and Indian wars Lord Jeffrey Amherst, Commander-in-Chief of British forces
in North America, and Colonel Henry Bouquet apparently contrived to defeat
Chief Pontiac's forces by giving them smallpox virus in contaminated
blankets.32

It is also conceivable that after the lifting of the siege, smallpox was
carried back to the countryside and then dispersed into a chain of indolent
person-to-person transmission lasting as long as a year. Indolent smallpox
transmission in nomadic or dispersed groups has been previously documented.
Among North American Indian populations of the Great Lakes/St. Lawrence

31 M. Young, "Small Pox Conveyed by Raw Cotton," Br. Med. J. 1 (1902) 687.
32J. Duffy, "Smallpox and the Indians in the American Colonies," Bull. Hist. Med. 25
(1951) 324-41. An early example of planned biological warfare, Bouquet wrote "I will try to
inoculate the with some blankets that may fall in their hands, and take care not to get
the disease myself."

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
296 David M. Morens and Robert J. Littman

River area, introduction of smallpox in the early 1630s apparently led to

regional transmission that continued for seven years.33 Such indolent
transmission also characterized the final days of smallpox during the
worldwide eradication campaign in the 1970s. It is noteworthy that most such
populations were at least partly immune, as returning populations of Atticans
would surely have been. Shipboard smallpox transmission and survival over
long ocean voyages is also well documented. As noted elsewhere, Thucydides
implied that some other unspecified cities may have had the disease as well. It
is thus conceivable that a combination of migrating Atticans and close allies,
Athenians not initially infected, and susceptible Athenian expeditionary forces
could constitute a critical mass of susceptible sufficient to sustain an epidemic
surge following re-introduction from indolent foci. We believe this possibility
warrants further scrutiny, including study of documented instances of indolent
transmission of smallpox in peri-urban areas, and fitting of the Athenian data
to more sophisticated mathematical models.
Existing data thus suggest to us that although purely respiratory diseases
can be ruled out as causes of the Athenian epidemic, respiratory diseases
capable of persistence in either focal reservoirs, or in fomites, or
simultaneously explosive in crowded populations and capable of indolent
transmission in. dispersed rural populations, should remain under evaluation.
Although most consistent with historically-documented smallpox, other long
incubation diseases such as streptococcal diseases might also be candidates for
further study with respect to mechanisms of persistence in rural and dispersed
populations.

Reservoir transmission. Epidemiologic aspects of the Athenian epidemic

appear to be most consistent with a disease associated with an infectious
reservoir (an insect or animal vector). Of these, certain zoonotic diseases
proposed as causes of the Athenian epidemic can be ruled out because human
infection is accidental ("dead end") in the course of zoonotic transmission, and
thus irrelevant to epizooticity. Among these are glanders, leptospirosis, rabies,
and tularemia.34 Psittacosis, which has not to our knowledge been previously
suggested as a cause of the epidemic, can also be excluded on this basis.

33D. R. Hopkins, "A Destroying Angel," Princes and Peasants. Smallpox in History
(Chicago 1983) 234-94.
34Glanders: C. H. Eby and H. D. Evjen, "The Plague at Athens: A New Oar in Mud
Water," J. Hist. Med. Allied Sci. 17 (1962) 258-63. Tularemia: J. A. H. Wylie and H.
Stubbs, "The Plague of Athens: 430-428 B.C. Epidemic and Epizootic," CQ 33 (1983) 6

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 297

Considered here under reservoir diseases, rather than under diseases of

inoculation, anthrax35 is an interesting possibility, primarily because by a
combination of inhalation, inoculation, and ingestion, anthrax could be
considered highly consistent with the clinical picture described by Thucydides.
Not only are dogs affected by anthrax, but the organisms may be carried by
birds, which are capable of contaminating water supplies. On clinical and
zoonotic grounds anthrax is difficult to refute. But large-scale anthrax
epidemics are not known to occur. What set of extraordinary circumstances
might lead to epidemic anthrax is difficult to imagine. Secondly, there is no
easily identified reservoir. Athenian sheep and cattle had been sent to Euboea,
probably leaving only dogs, rats, mice, and birds within the city. An epidemic
so massive and yet so chronic would have required a highly ubiquitous
infectious source not only continually present, but exportable to the
expeditionary forces. Spore-containing animal hides probably do not meet
these criteria. Anthrax is thus difficult to reconcile with epidemiologic
observations.

Other reservoir diseases suggested to have caused the Athenian epidemic

include various insect-bome diseases, whether or not associated with zoonotic
reservoir hosts, including malaria, plague, typhus, and various arboviral
diseases, for example dengue, yellow fever, and Rift Valley fever.36 All have
been associated with explosive epidemics, and all except Rift Valley fever are
closely linked to conditions of war, refugees, and overcrowding. Descriptive
epidemiologic aspects of certain of the arboviral diseases, including dengue and
Rift Valley fever, are probably more consistent with the Athenian epidemic
than are epidemiologic aspects of any other disease. These arboviral diseases
may not only be persistent, but re-emergent and explosive, satisfying the major
observational criteria of Thucydides. The explosive behavior of Rift Valley
fever has previously been noted. The explosive behavior of dengue has been
repeatedly documented in both crowded and uncrowded situations, in times of

35F. Jahn, "Beitrage zur Geschichte Carbunkel-Krankheiten mit Ausschluss der Pest," Janus
Zeitschriftfiir Geschichte und Literatur der Medicin (Breslau 1846) 369-414.
36Malaria: W. H. S. Jones, Malaria and Greek History (Manchester 1909) 33-40; Plague: F.
Ranchin, "L'histoire de la Peste," Opuscules ou traictez medecine divers et curieux en m6decine
(Lyons 1640); A. Hirsch, "Beulenpest," Handbuch der historisch-geographischen Pathologie.
erste Abteilung. Infektionskrankheiten (Erlangen 1859) 192-214; Typhus: Scarborough (above,
n. 1) Arboviral diseases: in discussing arboviral diseases we refer to them generically rather
than individually; comparisons to extant diseases would be misleading if, as appears likely,
significant evolutionary changes in these agents had occurred over more than two millennia;
Dengue: J.-P. Beteau, La Peste d'Athenes (430-428 av. J.-C.) preface par M. le Docteur L.
Tanon (Paris 1934). Yellow fever: Smith (n. 17 above); Rift Valley fever: D. M. Morens and
M. C. Chu, "The Plague of Athens," New Engl. J. Med. 314 (1986) 855.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
298 David M. Morens and Robert J. Littman

war and of peace. For example, numerous explosive epidemics were recorded
in the Pacific during World War II. Eighty per cent of U.S. Army troops
stationed in Northern Territory and Queensland, Australia, got dengue in
March-May 1942.37 Because of troop rotations, the incidence rate in Espiritti
Santo was 1,713 cases per 1,000 average troop strength in the month of April.
The corresponding annual figure for Saipan in 1944 was 3,560 per 1,000. In
peacetime, a 1977 Puerto Rican dengue epidemic swept through three million
persons, with substantial population dispersion (only one thousandth as
crowded as Athens in 430 B. C.), in five months.38 Study of a 200-worker
cohort during that epidemic revealed higher secondary attack rates in families
(secondary attack rates are traditional indicators of respiratory spread) for
mosquito-transmitted dengue than for respiratory-transmitted influenza, which
was also epidemic at the time. Even in an open population of Puerto Rico
(Bayamon municipio), dengue attack rates were paradoxically more highly
correlated with family size than were influenza attack rates.
Also associated with dengue and other insect-borne diseases are door-to-
door progression within neighborhoods, and increased risks for physicians and
other persons caring for the ill. Had Thucydides observed the 1977 Puerto
Rican epidemic of dengue and influenza cited above, he would probably have
noted dengue's marked association with crowding, and perhaps even with
infection of health care workers, before he noted the same for influenza, which
is among the most explosive diseases transmitted by the respiratory route. In
addition to a strong association with crowding, many arboviral diseases are
also linked to water storage, particularly storage in cisterns (widespread in
Piraeus) and in urns (widespread in Athens and Piraeus), where certain
disease-causing peri-domestic mosquitoes selectively lay their eggs. In modern
day Bangkok, explosive dengue epidemics may run through five million
persons, spread out over hundreds of square miles in a period as short as two
months, only to linger indefinitely, producing cases year-round, and to re-
emerge periodically. Crowded areas with substantial water storage are usually
stricken first and most dramatically, affecting multiple family members per
house. The reservoir mode of transmission is consistent with the two to five
year duration of the Athenian epidemic, with its explosiveness and its

370. R. McCoy, "Dengue. Epidemiologic Considerations," in J. B. Coates, E. C. Hoff, P

M. Hoff, Medical Department, United States Army (eds), Preventative Medicine in World W
II, Volume VII: Communicable Diseases. Arthropodborne Diseases Other Than Malari
(Washington, D.C. 1964) IV 29-40.
38D. M. Morens, J. G. Rigau-Perez, R. H. L6pez-Correa, et al., "Dengue in Puerto Rico,
1977: Public health response to characterize and control an epidemic of multiple serotypes,"
Am. J. Trop. Med. Hyg. 35 (1986) 197-211.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 299

association with crowding and contact exposure; it is the only obvious means of
explaining why neither the Spartans nor the many other people with whom the
Athenians came into contact over the five year period acquired the disease that
was devastating tens of thousands of Athenians.

A serious problem in identifying any of the nearly countless vectorborne

viral diseases, however, is that they evolve so rapidly: infectious agents existing
2,500 years ago would likely have undergone significant evolutionary change
in the intervening centuries, and thus might not be recognizable today.
Because, however, such evolution is usually most marked at the level of the
genome, and least dramatic at the level of epidemiologic "behavior" (with
changes in clinical appearance intermediate in degree), inferences about an
arboviral cause of the Athenian epidemic that reflect epidemiologic
observations are probably less liable to inaccuracies resulting from secular
changes in the agent. Other than the arboviral diseases, malaria, plague, and
typhus, have also repeatedly been associated with sudden explosive epidemics,
followed by persistence of the causative agents over sufficiently prolonged
periods of time to seed recurrent endemic or epidemic disease. Such other
zoonotic viral agents as hantaviruses, arenaviruses, and (presumably)
filoviruses also present clinical/epidemiological similarities, but because there
is less available information about most of them, they are not considered
further in this discussion.

Malaria is the least consistent of the remaining vectorborne diseases

because of seasonal incompatibility and problems with epidemic explosiveness,
particularly in closed populations such as Hagnon's expedition. Furthermore,
because malaria was known to and recognized by Hippocrates and
contemporary Athenian physicians, it is unlikely that Thucydides would have
confused it with an epidemic disease considered to be novel. Finally, the
clinical picture described by Thucydides is less consistent with malaria than ar
most other proposed vectorbome diseases.

Though consistent with the epidemiology of the Athenian epidemic,

plague appears to be less consistent with the clinical description. Thucydides
did not describe buboes, and deaths from pneumonic plague should have
occurred much more rapidly (one to three days) than the disease Thucydides
described.

Typhus, the classic explosive epidemic disease associated with wars, is

consistent with the description of Thucydides, including the common
occurrence of (actual) gangrene of the extremities, and of blindness. It may
obviously persist in crowded populations for prolonged periods, and is also
"persistent" in infected survivors who may, after long periods, have

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
300 David M. Morens and Robert J. Littman

recrudescences during which they again become infectious.39 The chief

argument against typhus is the ostensible presence of bullae in the Athenian
epidemic. However, as noted above, it is not certain that Thucydides actually
described bullae. In any case, vesiculobullous lesions have been described
occasionally in typhus (MacArthur has claimed they are not uncommon in
some epidemics,40 although most experts would probably consider them rare),
and their occurrence in rickettsialpox (an agent related to the agent of epidemic
typhus) suggests that an ancestral rickettsial agent could have caused the
Athenian epidemic. Examination of data from many typhus epidemics
associated with wars and refugee situations suggests that the Athenian epidem
was typical of typhus in many other respects.
In summary, our epidemiologic evaluation of the Athenian epidemic
excludes all common source diseases and most respiratory diseases. By a
process of exclusion, the cause of the Athenian epidemic can be limited to
either a reservoir disease (zoonotic or vectorborne), or one of the few
respiratory diseases also associated with an unusual means of persistence: eithe
environmental/fomite persistence, or adaptation to indolent transmission amon
dispersed rural populations. We suggest that the diseases in the first category
include typhus, arboviral diseases, and plague, and in the second category
smallpox. Both measles and explosive streptococcal disease appear to be less
likely candidates, but historical and modelling research may serve to further
characterize their suitability.
Our systematic approach to identifying the Athenian epidemic emphasizes
descriptive epidemiologic methods, use of mathematical models, and fittings of
the diseases' epidemiologic behavior to epidemics of known diseases
documented in pre-modem times. We have de-emphasized reliance on clinica
symptoms in favor of the disease epidemiology because pre-modem
descriptions, which lack detailed information on serology and accurate
accounts of rashes and other clinical features, will always retain a high degre
of uncertainty. Although the framework presented here is both conceptual a
preliminary, we suggest that there is much to be learned by comparing
Athenian epidemic to well-documented premodern epidemics of candidate
diseases, such as those we cite. We believe that our approach has already
limited the possible causes of the Athenian epidemic to a plausible few. Of
these, typhus and smallpox may share the most clinical similarities with the

39For example, C. R. Green, D. Fishbein, I. Gleiberman, "Brill-Zinsser: Still With Us,"

JAMA 264 (1990) 1811-1812 report recrudescent typhus in a World War II concentration camp
survivor more than 40 years after the infection.
40W. P. MacArthur, "The Athenian Plague: A Medical Note," CQ 48 (1954) 171-174.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 301

Athenian epidemic. Since there exists an enormous body of empirically derived

data on these candidate epidemic diseases in premodem times, we suggest as an
avenue for future studies attempts to develop refined and situation-specific
mathematical models for them using available historical information on their
behavior in defined populations of varying sizes, crowding indices, immunities,
and other relevant characteristics. In this manner, the limited number of
remaining candidate diseases can be compared to the epidemic situation in
Athens, in an attempt to find a "best fit". Such efforts will require an
interdisciplinary approach that draws upon the talents of physicians,
epidemiologists, biostatisticians, classicists, historians, and other scholars.41

41The authors acknowledge the help of Sally Drake, Hamilton Library, University of Hawaii
in locating rare manuscripts and Gertraud Maskarinec, M.D., M.P.H. for research assistance.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
302 David M. Morens and Robert J. Littman

= 10 Adequate Contacts per 12 Days

- - - = 20 Adequate Contacts per 12 Days
= 30Adequate Contacts per 12 Days
o,0oo = 40 Adequate Contacts per 12 Days

,~30)

(40) '/ \ (0
1,0001

I t y~~~~~~~~~~~~~~~~~~~

10 ~ ~ ~ :

5 10 15 20 25 30 35 40 45 50 55 60 65 70
_ ~D A YS AF T ER E PI DE M IC ON SE T

Figure 1. Theoretical epidemic curve of a transmissible respir

generation time (e.g. smallpox) in Hagnon's naval expedition to
population of 4,000 persons, according to the deterministic mat
separate curves are computed using adequate contact numbers of

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
 Epidemiology of the Plague of Athens 303

= INFLUENZA
= SMALLPOX
-- -----= MEASLES
= STREPTOCOCCAL DISEASES

100,000
.MEASLES
SMALLPOX M S

INFLUENZA k ,

I, l ', ,STREPTOCOCCAL
I \ I.. ~~~~~~DISEASES

10,000 \ I

u~~~~~~

0~~~~~~~~~~

iL 1

0 1 2 3 40 5 0 7 0 0 101012 3 4 5

DAYS AFTER EPIDEMIC ONSET

Figure 2. Theoretical epidemic curves of influenza A, smallpox, measles, and streptococcal

disease in a closed population of 100,000 susceptible persons, according to the mathematical
model of Maia, assuming an adequate contact number of 10.

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms
304 David M. Morens and Robert J. Littman

- . l-- - = 2 Adequate Contacts

. - 5 Adequate Contacts
- 10 Adequate Contacts
100,M00 -20 Adequate Contacts

(20)
,(10)

j>5)

10,000 I

3.To1a00
I~~~~~~ r I /

10 25 50 7 0 2 5 7 0 2 5 7 0 2 5 7

persons, according to the mathematical model of Maia. The c

numbers of 2, 5, 10, and 20 persons per serial generation time

This content downloaded from 206.253.207.235 on Thu, 27 Aug 2020 21:50:13 UTC
All use subject to https://about.jstor.org/terms