This paper presents a new manuscript of part of the Historia Augusta from Erlangen, which vindicates a more than century-old hypothesis by E. Patzig: that the 1489 Venice edition of the work is textually valuable. On this basis, and building on the recent work of R. Modonutti, I present five new passages that are not printed in modern editions of the HA, six lacunose passages restored, and propose that the lost Murbach manuscript is the source. Armed with this new evidence, I re-examine the question of the great lacuna between the Lives of Maximus and Balbinus and the Lives of the Two Valerians, showing that it is a codicological — and not authorial — feature.