The experts are fighting over whether the painting perished in 1529, 1634 or 1669, but the fact remains that is is gone. It's therefore natural to ask whether the books we have are faithful witnesses to the dance, as it once looked on the cemetery wall of St. Innocents' cemetery.
None of the printed books or manuscripts mention St. Innocents and maybe "Danse Macabre" just denotes a literary genre? Maybe the text that we know is as different from the mural in St. Innocents, as — for instance — the dance of death in Lübeck is from the dance in Berlin?
As a contrast, take Basel: When Frölich first published the text of the dance of death in Basel, he titled the book, »Lobspruch An die Hochloblich unnd Weitberümpte Statt Basel […]«. Merian called his book, »Todten-Tantz, Wie derselbe in der weitberümbten Statt Basel […]«, and ever since then all publishers of the local dance of death have remembered to include the word "Basel" in the title. In contrast the French books and manuscripts don't indicate that the Danse Macabre should be from Paris. Not even the books that were printed in Troyes, Rouen, Genève and Lyon.
Among the first to address this question were Lincy & Tisserand. They proposed two arguments for why we can trust the text:
les fresques se conservent assez longtemps, et celles des
Innocents, qui, selon M. Édouard Fournier, existaient encore au xviie siècle, devaient être
assez conservées soixante ans après leur exécution, pour être reproduites, comme on l'a fait ,
par un simple trait.
Guyot Marchant avait donc les sujets sous les yeux;
(Lincy & Tisserand, Paris et ses historiens, 1867, page 289)
Their first argument was that frescoes last for a long time and that Guy Marchant therefore would have had the mural "under the eyes" 60 years later.
Unfortunately this doesn't prove anything. To use Basel as an example again, the family Mechel also had the dance of death in Basel "under the eyes" through almost a century, while selling their mixture of Basel and Holbein with some Bern and a little of Frölich added. The publisher of Der Todendantz had the same mural "under the eyes", when he published his text with lots of deviations in two editions. And both Gross & Tonjola had the mural "under the eyes" when they reprinted the text in countless editions and still included a greeting to Frölich's imaginary satyr.
de plus, le manuscrit
du fonds Saint-Victor n° 1122, dont sa première édition est la reproduction textuelle, porte
cette mention significative: Est la Dance Macabre, prout habetur apud Sanctum Innocentium; et
un autre manuscrit (pupitre T. T. n° 12), cité par M. Paul Lacroix, donne à peu près le
même titre : Dictamina choree Macabre, prout sunt apud Innocentes, Parisius.
(Lincy & Tisserand, Paris et ses historiens, 1867, page 289)
The second argument is that according to Paul Lacroix there are two manuscripts, which themselves state that they reproduce the text from St. Innocents.
The problem is that this is not true. Those claims stem from the abbey's catalogue and all they tell us is that the abbot in 1514 thought that the text in the manuscripts was "such as" the one at St. Innocents. Maybe this abbot would also think that the dance in Lübeck was "such as" the dance in Berlin?
On a sidenote, the manuscript that Lincy & Tisserand mentions, n° 1122, is the catalogue and doesn't contain any Danse Macabre-text. See much more about the manuscripts from Saint-Victor.
Another scholar who had addressed the question already in 1862 was Langlois. He too had listed two arguments for his "fermement" belief that the old books from Paris and Troyes reproduce the mural from the ossuary of St. Innocents in Paris:
Nous croyons fermement que les éditions primitives de la Danse
Macabre et les éditions consécutives de Troyes ne font que reproduire
la peinture et les vers du charnier des Innocents de Paris.
Nos raisons pour arriver à cette induction sont que, dans ces volumes,
on retrouve le sujet des trois Morts et des trois Vifs, et celui
de l'Homme Noir, qui faisaient partie de cette peinture (voyez
p.111 et 128 );
(Eustache-Hyacinthe Langlois, Essai historique, philosophique et pittoresque sur les Danses des morts, 1852, Vol. 1, page 197)
The first argument is that the text was published together with the legend of the three living and the three dead and with the black man.
This argument seems very flimsy. It is true that since 1408 there had been a sculpture of the three living and the three dead at the entrance to the church, donated by Jean, Duc de Berry, who was counting on being interred there himself. But the legend is much older than the dances of death, and can be found in lots of manuscripts. As an example, the manuscript to the right has three different versions of the legend (here, here and here) without having any Danse Macabre, and conversely Guy Marchant's first edition of la Danse Macabre did not feature the three living and the three dead.
The black man is even worse. He normally only appears in the so-called shepherd's calendars where he is a frequent guest. The absence of black men from the books with la Danse Macabre would according to Langlois' logic rather prove that the text is not from St. Innocents.
Langlois' second argument is considerably better:
et qu'en outre les vers qui accompagnent ces différentes
éditions présentent des analogies évidentes de tous points
avec ceux du moine Lydgate, qui a traduit en anglais les vers français
de la Danse des Innocents de Paris, comme il l'apprend luimême
dans l'introduction de son poème, publié dans le Monasticon Anglicanum
de Dugdale. London, 1682 (vol. III, p. 367-374).
(Eustache-Hyacinthe Langlois, Essai historique, philosophique et pittoresque sur les Danses des morts, 1852, Vol. 1, pages 197-198)
We have in fact a contemporary witness, John Lydgate, who a few years after the mural had been executed (maybe as early as in 1426), translated it into English. Lydgate took great liberties with the text, e.g. he added five new dancers, but there is no doubt that the text he worked with was the same as the one we know from Guy Marchants first edition of la Danse Macabre.
Lydgate gives us all the details that the French publishers omit: The text is from France: »Ther of frensshe clerkes […] Owte of the frensshe / Macabrees daunce«, »Owte of the frensshe / I drowe hit«; more specifically Paris: »And fro Paris / to Inglond hit sent«, »whiche that at Parise«; it was at a place with ecclesiasticals: »Ther of frensshe clerkes«; it was a mural: »depicte / ones on a walle«; and in case we are still in doubt, the dance was at St. Innocents: »The whiche daunce / at seint Innocentis«
The odd thing is that in spite of the fame and popularity of the painting there's only one contemporary description of it.
This book doesn't have a name, but has been catalogued as »Recueil d'épitaphes, formé par Pierre Clairambault, en partie avec des débris du cabinet de Gaignières. I-V«. In other words: A collection of epitaphs from all possible churches of Paris, which partly consists of the "débris" from the 2,886 books of the Cabinet de Gaignières.
The book is incredibly disorderly and impossible to navigate. The index at the start apparently describes some other books; at some places the handwriting is beautiful, sometimes illegible; and many pages are blank. Gallica adds to the confusion by scanning some pages several times (e.g. pages 299, 237 and 239) while other pages are missing (e.g. page 233).
The description of the epitaphs in the Charnier des Lingères starts abruptly after three blank pages — with arcade number 2 — without telling what part of which church is being described. After arcade no. 28 the text continues by describing arcade no. 2, but still without telling what part of what church we are now discussing.
But here follows the description of the 17th arcade:
Icy commence la Dance macabee [sic] qui dure 10 . arcades
The dance took up 10 arcades with six 8-lined verses in each arcade, and 8 verses in the four last. This means there would have been 68 verses like the "soixante hvict hvictains" in the title of the book to the left.
There's a slight problem here for the penultimate verse, the one starting, »Bon y fait penser soir et main«, was added later. It appears neither in the manuscripts nor in Guy Marchants first edition of la Danse Macabre.
A much bigger problem is that the 68 verses include the usurer's customer, the poor man. This means that there would have been 5 or 7 verses in this arcade. The same thing happens around the hermit, where an extra cadaver gets the last word, which would also have resulted in 7 verses.
Thus we can't make the equation 100% true, but since the 68 verses after all is a fairly accurate number, and the author even quotes the first verse: »O creature raisonnable […]«, we must say that the description fits.
Coupled with Lydgate's testimony there is therefore no doubt that the text of Guy Marchant's first edition is the same as the one that was once painted on the cemetery wall of St. Innocents.
Finally we should not forget that the text itself states that it originates from a painting: »Hec pictura decus […]«.