In principle, a few of the strong earthquakes (I0 >= 8/9, M >= 5.8) that affected Italy in the past may still be missing from parametric catalogues or be listed there as lesser events, their actual strength unrealized. This seems a reasonable enough inference, given that some strong earthquakes were listed by catalogues quite by chance, from information drawn, mainly or even solely, from a single source. Had this source been destroyed before catalogue compilers were able to consider it, or had they for any reason overlooked it, the earthquake it recorded could also have been missed or underestimated. This paper examines the two most peculiar Italian cases of «single-source earthquakes» (1561 «Vallo di Diano?»; 1639 «Amatrice?»). Is all relevant information on each event really tied up in a single source? And if so, why? Finally, are these cases unique or do they share any common features that could, by occurring elsewhere, act as markers for situations where forgotten earthquakes could still lurk undetected?