This dataset holds hundreds of commemorations, and a day in a calendar establishes less than it looks like it does. Delehaye draws the line twice. A name in a menology, a passionary, a synaxary or a martyrology may represent nothing but literary tradition — a compiler copying a compiler — and such entries have to be set aside before anyone can ask whether a day was actually kept. He gives the sharpest form of it on Procopius: the Eastern martyrology’s notice for him depends on Eusebius’ book, so the entry attests a text and not an observance. What does prove a cult is material and independent — the pilgrims recorded going to a shrine, a basilica an emperor restored, a second shrine attested in another city. And the converse holds: a story’s collapse does not take the day with it. Where a legend is a romance with almost nothing historical in it, the saint’s name, the existence of the shrine and the date of the feast are often all that survives — and they do survive, because converting one saint’s acts into another’s means changing precisely those three and nothing else. ⚠ So the honest reading of a negative verdict: when this project marks a person legendary it is saying what Delehaye says it means — that they are known only by a legend of insufficient authority to prove their existence. That is a statement about the evidence, not about the world, and it is why nothing here is deleted for being legendary.
The Legends of the Saints: an Introduction to Hagiography · ch. VI §V p. 187, ch. V p. 128 and pp. 144–145 (Procopius), ch. IV p. 114, ch. VII p. 216