For most of its European life the Madrid Codex was two manuscripts with two names. One portion, the Troano, was described in Madrid in 1866 and bought by the Spanish state in 1888. The other, the Cortesianus, was acquired by the Museo Arqueologico Nacional in 1872 from a collector who said it had come from Extremadura. Leon de Rosny recognised in 1880 that the two were halves of a single book. They have been kept together since, and the compound name Tro-Cortesianus preserves the accident.
How the book left Yucatan is not recorded. The claim attached to the Cortesianus half, that a companion of Cortes brought it back to Spain, is a dealer's provenance and not a document, and the atlas states it as such rather than repeating it as history. The Madrid Codex therefore has a well attested modern history and an unrecoverable early one, which is the ordinary condition of the surviving Maya books rather than a peculiarity of this one.
What the split cost is worth naming, because it is the kind of damage only a catalogue can do. For fourteen years the two halves were studied as separate manuscripts. Any reading of an almanac that ran across the join was wrong by construction, and no amount of care by the individual scholars could have corrected it, because the error was in the shelf and not in the argument. Recognising the two as one book was a custodial act before it was a scholarly one.