The transcription of a medieval manuscript is inconceivable without a certain “fidelity”… Certainly, but fidelity to what? Should a “faithful” transcriber today be obliged to reproduce in every detail the spatial arrangement of the source he is working on? Once considered, such a program now seems problematic. Our transcriptions are computer files in text form that need to be writable by humans and readable by machines. If we wanted to mimic the two-dimensional space of the folio as accurately as possible, we would have to overload its transcription with a myriad of annotations about the specific shape of the graphic elements, their size, their spacing, their absolute or relative position, and so on.

If you have the ambition to encode musical data, the first idea that comes to mind is to use one of the software packages that have become widely available in recent years. However, a project team that manages to choose between these competing products would still be at the beginning of its problems: while it is true that these software programs have acquired some credentials in the art of harmoniously arranging graphical objects in a two-dimensional space, they remain largely ineffective in the art of collecting data.