In 1983 Richard Stothers and Michael Rampino of NASA published a list of all ancient volcanic eruptions known from Mediterranean historical sources. Their list included a persistent dust veil or dry fog which darkened the sky for about a year in AD 536--37, bringing about cold, drought and food shortage in the Mediterranean area or, as it has since been claimed, all over the northern hemisphere. Especially following two popular books devoted to the dust veil by David Keys and Mike Baillie, it has been acclaimed as the worst climatic disaster in recorded history. In the most wide-ranging scenarios, the year 536 is seen as a watershed moment between the ancient and modern worlds, bringing about economic decline, population movements, political unrest, and ultimately the collapse of civilizations.

In a lengthy article written in 2004 and just to be published in the Byzantinist journal Dumbarton Oaks Papers' 2005 issue*, I have gone through all the available physical and written evidence for the 536 event. The inevitable conclusion from the ancient literary sources is that the historical impact of the cloud must have been extremely limited. On the other hand, some assumptions about the cloud's physical nature that have hitherto been taken for granted should be re-examined. In the following, I give a brief summary of my paper.

Physical evidence for the 536 event is derived from two main sources: tree rings and acid layers in Greenland ice. The tree rings show 536 and the following ten years as a period of very slow growth for Scandinavian pines, North European oaks and several North American species. However, the contours of a sudden catastrophe cannot be directly read from the tree ring evidence. In many series, the drop in 536 is followed by a recovery in 537--38 and then again by an even more serious plunge. In most cases, the worst years are around 540, and in Siberia 543. In southern Chile, the trough is in 540, while in Argentina there was dramatic growth reduction only after 540, with a minimum in 548. In Tasmania, the tree growth declined between 546 and 552.

Thus, although the year 536 was certainly a very bad growing season in many parts of the world, it is situated in a decade-long downturn in the climate of the Northern Hemisphere and is separated from the really worst seasons by 3--7 years. Moreover, and perhaps even more seriously, in the Scandinavian pines as in the oaks and North American trees, it is possible to see a long-term growth decline during the early part of the sixth century which is matched by an equally slow rise in the average growth during the second half of the century. This would place the years around 540 as the lowest point in a slow climatic cycle. While it does not disprove a climatic anomaly in 536, all this nevertheless suggests that the link between the dark cloud and tree growth is not as straightforward as might be wished. The dendrochronological maxim "trees do not lie" may be true, but neither do they seem to provide unequivocal answers to the questions which historians would like to pose to them.

Historical eruptions are usually attested as acid layers in Greenland ice. In the previously published studies, all the relevant sections of the Greenland ice cores for the mid-sixth century have been either missing, flawed or poorly dated. Recently, Danish scholars have reported that a major eruption can be dated to the early spring of 528. It is unclear whether it might be possible to redate the whole sequence of ice layers by a few years, matching the new attested eruption with the 536 event. Any conclusions therefore must remain tentative, but so far we have to admit that no acid layer sufficient for a major volcanic eruption has been confirmed around 536. That is why the cloud has been attributed to the impact of a comet. This hypothesis is not confirmed by any direct evidence either.

Archaeological evidence does not help us assess the consequences of possible crop failures around 536. Recent archaeological work serves to stress the need for a regional approach: economic and demographic developments may differ in neighboring regions. The whole western part of the Roman empire was in clear decline already in the fifth century. The Persian devastations in northern Syria, combined with recurrent earthquakes and epidemics, would probably suffice to explain any sixth-century economic decline in the Byzantine Near East.

The results of my inquiry into the written sources are relatively straightforward: although the cloud occasioned confusion and crop failure at the time it was seen, its effects did not last long after it had dissipated. Compared with almost all other contemporary civilizations around the world, the circumstances in the Mediterranean area are extremely well documented. The literary sources which record the darkness of 536/7 all seem to consider it a temporary misfortune. Among the innumerable earthquakes, droughts, plagues, swarms of locusts, and slaughters which are listed by the historians of this time, the dark cloud was not counted as a particular catastrophe. Shortage of food was a recurrent phenomenon in the ancient world, and people were used to it, however intense the short-term suffering might be.

For example, two Italian sources, Cassiodorus and the Liber Pontificalis, attest continuing problems with the harvest in 537, which is not surprising if the fog persisted until the summer. Immediate effects of the event are not reported after that. The historian Procopius for his part does not mention the crop failures of 536/7. He says that outside besieged Rome the Goths were also starving, but he rather seems to give the credit for it to a successful Byzantine naval blockade. In contrast, the historian describes at great length a terrible famine in Italy in 539. However, he is quite explicit that it was due to the fields being left uncultivated because of the war. A little later he returns to the subject of food shortage among the Goths, again insinuating that the lack of supplies was a logistic problem. He does not give a hint that climatic conditions might have been blamed for continual bad harvests.

Though these sources leave no doubt that a mysterious fog was seen in an area which extended at least from Italy to Asia Minor and caused bad harvests there for one or two years, they all seem to treat it as a temporary bad omen, not as the beginning of a long period of unfavorable climatic conditions. Of course, the writers might not have noted a slight drop in average temperatures, and might perhaps not have cared to record a change in prevailing winds or precipitation. However, if the direct consequences of such underlying factors for agriculture had been grave enough to undermine the economic well-being of the empire, we would expect somewhat more attention being paid to them by contemporary writers.

Thus, the combined force of the available evidence irresistibly shows that, whatever happened around 536, its historical implications remained very limited, at least in the Mediterranean area. On the other hand, the sources report interesting, though sometimes conflicting, details of the fog. Although the haze has been called a dry fog or dust veil ever since 1984, a passage from the eyewitness antiquarian writer John Lydus which has hitherto been neglected rather suggests that the fog was damp. This is not decisive because it can reasonably be claimed that Lydus may not have been able to observe its actual composition. However, he also asserts that the fog was seen only in Europe, and it is more difficult to discredit this report out of hand. It would be in clear contrast to the common scholarly assumption that the cloud was a global or at least a hemispherical phenomenon. Remarkably, all the other literary sources mention the fog only for an area around Italy and Asia Minor.

Cold and drought are attested in other parts of the world but not the persistent fog. Chinese sources record that the star Canopus was not seen at the spring and fall equinoxes in 536. Although this might be taken to refer to reduced atmospheric transparency (as many scholars have assumed), it seems a rather understated way to describe a darkness which continued for a year. It is especially odd if it was the factor which caused summer frosts, drought and widespread famine, duly recorded in Chinese historical works between 535 (sic) and 538. At least two possibilities emerge: either the Chinese did not mention the fog because opaque skies are not unusual in northern China due to the frequent desert storms there, or the fog was tropospheric and localized in the Mediterranean area. While zonal winds would have spread a stratospheric fog over the northern latitudes within a few weeks or months, a tropospheric fog (volcanic or not) might very well have attenuated before reaching China. The problem remains that no tropospheric fog of such duration has been observed in historical times.

However, if we accept the possibility that the fog may have been seen in northern China though it was not clearly recorded, it might also be possible to explain Lydus' account in a different way. All those areas for which the fog is securely attested (Italy, Constantinople) lie above 35 degrees of northern latitude, perhaps even above 40 degrees, depending on how we interpret Procopius' report. The same is true of northern Mesopotamia (ca. 37° N). In contrast, those areas further east which Lydus claims did not witness the fog (Persia, India) all lie below 40 or even 35 degrees northern latitude, and this also applies to most of China. Thus, we might actually have a cloud which could be seen only at latitudes north of the Mediterranean and in the very north of China. Such a rather abrupt and globally uniform cutoff latitude falling between 30 and 40 degrees has been observed for stratospheric aerosol veils stemming from large eruptions of northern volcanoes, notably Lakagigar (Iceland, 1783), Ksudach (Kamchatka, 1907) and Katmai (Alaska, 1912). For example, the dust cloud from Katmai was seen and measured at Bassour, Algeria (36° N), at Simla, India (31° N) and at two US observatories (34-36° N), but not at Helwan, Egypt (30° N).

If we interpret Lydus' text in this manner, disregarding his report of the moist fog and assuming that the missing or misdated acid layers in the ice cores can be explained somehow, it would add a new dimension to the volcano hypothesis. It would actually support the suggestion made by Richard Stothers that the mystery cloud derived from a far northern volcano, and not from a tropical one like Rabaul (New Guinea), Krakatau (Indonesia) or El Chichón (Mexico), which have been earlier suspects. The observed decline of tree growth in South America in the 540s might seem to be at odds with this. However, it has not yet been established whether a high-latitude eruption could have global climatic effects. The issue is currently debated.

We cannot check the scientific accuracy of Lydus' reports. They may mislead us, but at the very least they invite us to re-examine the scientific evidence for the event. It remains true that the Greenland ice cores have so far produced little proof of volcanic activity around 536, and that the tree rings are surprisingly ambiguous about climatic variation in different parts of the world between 535--552. Two main alternatives emerge. The dark cloud may have originated from a northern volcano, being visible only at latitudes north of the Mediterranean, or the fog may have been locally more restricted, perhaps damp, originating from a totally unknown source. As a tropospheric fog of such duration would be quite exceptional, the first alternative perhaps seems at present more likely. Further ice cores may prove or disprove it in the future. However, for those who are as of yet not convinced by the volcano hypothesis, the second alternative might appear worth serious consideration.

* Antti Arjava: The Mystery Cloud of 536 CE in the Mediterranean Sources, Dumbarton Oaks Papers 59 (2005), forthcoming in Dec. 2006.