Scholarly music editions concern with a special kind of text. This text offers all the problems known from literary texts: There are substitutions, additions, alterations, cancellations and concurring variants in different sources. But unlike in the literature sciences, the individual signs that establish the musical text are far from being standardized during the last 400 years. The meaning of a given sign heavily depends on its author and the area and time he lived. Therefore, an editor of scholarly music editions always has to take the context of his sources into consideration.

For the encoding of musical scores this implies a strictly independent view at the material. As David Halperin says, “our task is to encode the written scores, not the (sounding) music” [Selfridge-Field (ed.): Beyond MIDI, p. 574]. Of course a scholarly edition has to pay attention to the sound of music at a certain stage, but first the individual characters have to be analyzed detached from their apparent meaning (which is the sound intended by the author). So a music encoding format with an editorial scope has to offer separate views at the graphical surface and the music contained within the sources. The majority of existing music notations formats does not concern with these issues, whereas the TEI currently lacks support for encoding music notation.

The Edirom project aims at the development of tools for the preparation and inspection of digital historio-critical music editions. These editions cover large sections of european music history and therefore are very demanding to a possible file format, which should cover all issues in an appropriate manner. The newly established TEI SIG Music “seeks to […] incorporate encoding for music into […] TEI documents”. During my talk I will try to illustrate the requirements for such an encoding from the perspective of a musicologist resp. music editor.

The intended usage of the encoded material within TEI documents turns out to be a crucial question here, as the generation of a mere printout entails completely different demands than an analysis. Furthermore, the delimitation of encoding and facsimile seems to be very important. Due to the inherent problem of assigning meaning to symbols, these symbols may be encoded in a very particularized and sophisticated way. But sometimes a facsimile might prove to be the better solution, as the knowledge gained by bezier curves of handwritten ties and slurs tends to be rather small. Therefore it seems to be very important to find a well-reasoned balance between mightiness and limitation of a music encoding scheme for editorial purposes. The multitude of music encoding formats meanwhile abandoned enforces a very close examination of the different requirements for music encoding within the TEI. After all, the quality of music encoding should meet the quality of text encoding already achieved.