Strasbourg · July to August 1518
What made the poor of Strasbourg dance in the streets for weeks in July and August 1518, how many danced, and did anyone die?
What happened. On 14 or 15 July 1518, in Strasbourg, then a free city of the Holy Roman Empire on the Rhine, a woman began to dance in the street and kept it up for days. The city council had her taken to a shrine of St Vitus, the saint believed both to send this sickness and to cure it, in the hills above Saverne, a small town to the north-west; while she was on the road, others began to dance, about thirty-four within four days. The council restricted drums on 22 July, lodged the dancers in guild halls, sent them to the shrine by wagon on 23 July, wrote to its bishop on 25 July that the physicians called it a natural illness of the heat and the stars, banned all public dancing on 2 August, and voted a hundredweight candle for the saint on 31 August. The council's decree speaks of up to fifty dancing at once; the chronicle of the Imlin family of Strasbourg, written at the time, counts more than two hundred over four weeks.
Why it is a mystery. The cause was not diagnosed in 1518 and is still disputed: the physicians said natural illness; the chroniclers of the next century said a curse, a punishment and fraud; E. Louis Backman, a Swedish physician and historian of religious dance, proposed in 1952 that the dancers were poisoned by ergot, a fungus of rye whose poison causes convulsions and gangrene, and that became the popular explanation; John Waller, a historian of medicine, argued in 2008 for a trance brought on by hunger, fear and belief in the saint's curse, spread by sight; and Gregor Rohmann, a German historian of the Middle Ages, and Robert Bartholomew, a sociologist who studies outbreaks of mass hysteria, read the dancing as a religious rite. The facts are disputed too: the popular story has four hundred dancers, fifteen deaths a day, a first dancer named Frau Troffea (a name found only in Paracelsus, the Swiss physician, who wrote on the sickness about 1526 to 1532) and a council that built a stage and hired musicians and dancers to keep the sick dancing, while the surviving texts give figures from fifty to four hundred (the four hundred rests on one nineteenth-century reading of one word in one manuscript). The first reports of deaths come sixty and a hundred years after the event, from Daniel Specklin, the city architect, writing about 1580, and from Michael Kleinlawel's rhymed chronicle of 1625; the council's own minutes burned with the city library in 1870 and are known only from extracts made in the eighteenth century, though a few originals survive, among them the minutes of the three councillors who questioned the dancers' families on 23 July and the letter to the bishop. The American historian Erik Midelfort (1999) gave the sober account of the sources, and the Strasbourg historians Élisabeth Clementz and Georges Bischoff, in a 2018 exhibition, reduced the outbreak to about fifty dancers with no contemporary death.
How it was judged. Every text that describes the dancing was read in the order it was written, from the council's papers of 1518 to the chronicles of the following century, in the printed editions (above all the medical historian Alfred Martin's study of 1914, which printed the council's papers) and in the page images, with one exception, the account of the cathedral schoolmaster Hieronymus Gebwiler from the early 1520s, known only through later writers; each element of the modern story was traced to the text in which it first appears, the phantom outbreak of 1418 listed in half the modern literature, for instance, to the jurist Johann Schilter, who printed the council's orders under that wrong year in 1698; and the explanations were tested against that record and against three kinds of evidence the earlier writers did not have: the weather of the growing seasons of 1515 to 1518 reconstructed from tree rings and written weather records, the grain prices of those years, and the modern medical case series of outbreaks that spread through schools and villages by suggestion. What tipped the balance was the record of 1518 itself: the texts written that year describe poor people set off by watching others dance or by the sound of a bagpipe, a verdict of natural illness from the physicians, up to fifty afflicted at once and no death, while the four hundred, the daily deaths, the stage and the name Troffea each enter the story later at a date that can be fixed, and ergot fails on eight counts, among them that the rye eaten in July 1518 had grown in the summer of 1517, the third driest of the century at Strasbourg. The verdict is a behavioural epidemic, an outbreak spread by sight and sound rather than by anything eaten or caught, among the poor of a city at the end of two years of dearth (failed harvests and high grain prices), at about 85 percent, with ergot rejected; a document of 1518 recording deaths among the dancers would cut that share to about 70 percent and reopen the organic causes, a poison or a disease.
A behavioural epidemic among the poor of a city at the end of two years of dearth, spread by watching and by music, and given its shape by the belief that St Vitus sends the dance and cures it: about 85 percent that the cause was behavioural rather than a poison or an infection, and ergot rejected. Up to fifty dancing at a time by the council's count, more than two hundred over four weeks by a chronicle written at the time, and no death in any text read from the years 1518 to about 1580.
Every percentage is a judgment rather than a measurement: it depends on the five explanations weighed and on the weight given to each piece of evidence, and the table of what moves it is printed beside the headline. Backman's book of 1952 was not obtained, Waller's book was read only in library search snippets and Rohmann's book of 2013 was not read, so the opposing works are answered through their articles and through other writers.
About 65 percent that it was, more exactly, a psychogenic epidemic, one in which the symptoms pass from person to person by suggestion rather than by any germ or poison (mass psychogenic illness of the motor type, the modern name for such an outbreak of abnormal movement; "a natural illness" in the words of the physicians of 1518), and about 20 percent that the dancing was a culturally scripted rite, the reading of Martin 1914, Bartholomew 2000 and Rohmann 2013; the line between those two is partly vocabulary. Ergot about 5 percent, another organic disease about 5 percent, mostly fraud and begging about 5 percent. Accepting Specklin's and Kleinlawel's late reports of deaths as fact cuts the behavioural share to about 70; the most sceptical case, which also allows an ergot route through old granary stock and reads the physicians' diagnosis as an infection, leaves it at about 55, still leading, with ergot at 18 and another organic disease at 20. In the base case neither ergot nor another organic disease reaches ten percent.
The size of the outbreak, in the texts written at the time: one unnamed woman on Wednesday 14 or Thursday 15 July 1518; about 34 dancers within four days (Imlin chronicle); "uff fünfzig personen", up to fifty, afflicted at once in the council's own decree (its words survive in an eighteenth-century extract from the burned minutes, probably taken from the preamble of the dancing ban of 2 August), a figure for one moment; "more than 200" over four weeks in the Imlin chronicle, the only running total from the period, and nothing in the record contradicts it; "over 100 at once" only in Specklin, about 1580. The council restricted drums on 22 July, sent the sick to St Vitus at the Hohlenstein, a hollow rock above Saverne, on 23 July, told the bishop on 25 July that the physicians called it a natural illness of the stars and the heat and that it was coming to an end, banned public dancing until Michaelmas (29 September) on 2 August, made families and brotherhoods responsible for their sick on 11 August, and voted a hundredweight candle for the saint on 31 August. Nobody is recorded as having died in any of these texts. That is silence in the sources read; it is not proof that nobody died, and no parish or hospital register was searched.
The strongest rival is ergot poisoning, Backman's explanation of 1952 and the popular one. It fails on eight counts: the season (onset before the 1518 harvest), the weather of the crop year (1517 was the third driest summer of the century at Strasbourg in the Old World Drought Atlas, a reconstruction of summer soil moisture across Europe from tree rings, with a dry spring and a warm flowering), the regional form of the disease (the ergotism of Alsace, the region round Strasbourg, was the gangrenous form called St Anthony's fire, which produces burning pain and loss of limbs, and nothing in 1518 mentions burning limbs, gangrene, formication (a feeling of insects crawling under the skin) or cramps), its clinical course (convulsive ergotism keeps its victims in bed with spasms for weeks, and these dancers were on their feet in the streets for days), the way the dancing spread (a boy who watched, a woman who followed a bagpipe), the absence of recorded deaths, the physicians' diagnosis, and the end in August. The argument usually made against ergot, that its alkaloids would have decayed within a year, is unsound (Barger 1931, the monograph by George Barger that is still the standard history of ergot) and is not used: the rye harvest of 1515, which had classic ergot weather, could in principle have reached the poor through the granary, and ergot keeps a few percent because one poisoned person inside a larger behavioural outbreak cannot be excluded.
The first two rows are both behavioural readings and together give the headline figure of about 85 percent in the base case. The table printed beside the headline changes one assumption at a time: accepting the deaths that Specklin and Kleinlawel report sixty and a hundred years after the event cuts the behavioural share to about 70, and the most sceptical case, which changes every assumption at once, leaves it at about 55, with ergot at 18 and another organic disease at 20.
Every text that describes the dancing was read in the order it was written, from the council's papers of July and August 1518 to the chronicles of the following century, in the printed editions (above all the medical historian Alfred Martin's study of 1914, which printed the council's papers) and the page images. The contemporary account of the cathedral schoolmaster Hieronymus Gebwiler was not read and is known through later writers; the council's letter to its bishop of 25 July 1518 was read only in a French translation of 2018; the books of E. Louis Backman (1952) and John Waller (2008) were read only in short passages.
Each element of the modern story was traced to the text in which it first appears: the name Troffea to the Swiss physician Paracelsus; the deaths and the stage to the city architect Daniel Specklin, about 1580; the four hundred dancers to a reading of 1879 of one word in one chronicle. The five explanations were then weighed against that record and against three kinds of evidence the earlier writers lacked: the weather of the growing seasons of 1515 to 1518 reconstructed from tree rings and weather records, the grain prices of those years, and modern medical case series of outbreaks spread through schools and villages by suggestion.
Figure 1 shows the five explanations and the weight given to each: a psychogenic epidemic, in which the symptoms pass from person to person by suggestion rather than by any germ or poison, at 65 percent; the same behaviour as a culturally scripted rite at 20; ergot poisoning at 5; another organic disease at 5; and a few sick people with a crowd of beggars and impostors at 5. Figure 2 shows how the sources reach us: which texts were written within a generation of the event, in which texts deaths first appear, and which links between texts are inferred from shared wording rather than from a documented copy.
The texts written in 1518 describe poor people set off by watching others dance or by the sound of a bagpipe, a verdict of natural illness from the physicians, up to fifty afflicted at once and no death. Ergot, a fungus of rye whose poison causes convulsions and gangrene, fails on eight counts, among them that the rye eaten in July 1518 had grown in the summer of 1517, the third driest of the century at Strasbourg. The verdict is a behavioural epidemic among the poor of a city at the end of two years of dearth (failed harvests and high grain prices), at about 85 percent.
The percentages are judgments that depend on the five explanations weighed and on the weight given to each piece of evidence, and the report prints a table of how they move when one assumption changes. Accepting as fact the deaths reported by Specklin and by Michael Kleinlawel's rhymed chronicle of 1625, sixty and a hundred years after the event, cuts the behavioural share to about 70; the most sceptical case, which also allows ergot a route through old granary stock and reads the physicians' diagnosis as an infection, leaves it at about 55, still leading, with ergot at 18 and another organic disease at 20.
Already established by others. The verdict is the one Élisabeth Clementz and Georges Bischoff reached for the 2018 exhibition "1518, la fièvre de la danse": about fifty dancers, no contemporary death, Troffea an invention of Paracelsus, ergot rejected. They had already questioned how much contemporary testimony lies behind the familiar story, and this report (the full report is linked from this page) builds on their correction. Erik Midelfort (1999) gave the sober English account, one woman about 14 July, thirty-four within four days, the sick sent to Saverne, deaths only in Kleinlawel. John Waller's premise, a city in dearth and fear, is borne out by the climate and price record, and his account of how the dancing spread, by sight, is the one described in the surviving minutes of the three councillors of 23 July.
What historians have said. Alfred Martin (1914) dated the outbreak to 1518, printed the council's papers and disposed of Schilter's 1418. Erik Midelfort (1999) gave the sober English account, with deaths only in Kleinlawel. Élisabeth Clementz and Georges Bischoff, for the 2018 exhibition, reduced the outbreak to "quelque cinquante personnes", found no death in any contemporary document, exposed Troffea as Paracelsus's invention and dismissed ergot. Martin and Midelfort carried Witkowski's 400 reading of the Imlin chronicle, and Waller took it from them. The popular account, from Waller's book of 2008 and articles of 2009 to Wikipedia, has 400 dancers, dozens of deaths, fifteen dying a day and a council that built a stage to keep them dancing. E. Louis Backman (1952) argued for ergot; Gregor Rohmann (2013) and Robert Bartholomew (2000) read the dancing as a religious rite rather than an illness.
The finding agrees with the Strasbourg historians of 2018, with Midelfort and with Martin, and builds on the 2018 exhibition's correction. It shifts the popular account: 400 dancers become up to 50 at a time and more than 200 over four weeks; dozens of deaths become none recorded in the sources read before about 1580; "fifteen dying a day" has one candidate source, Rem's count of fifteen new cases on one day, and its descent into the modern death rate is an inference rather than a traced citation; the council that staged the dancing becomes a council that contained it, Specklin (about 1580) being the only source for the stage. Ergot (Backman 1952) is rejected on eight counts.