This paper gives a brief survey of the Saarbrücken project on Old Icelandic legal texts, sponsored by the German Research Society, within the Special Research Area “Computer linguistics.” The project's main points of interest are (1) producing adequate machine-readable versions and parsed indices of all legal texts in Old Icelandic, (2) graphemic studies of legal manuscripts, and (3) studies of the distribution and valence of the verbs in those texts. A proposal for encoding Old Norse/Old Icelandic demonstrates how texts of different standards (normalized, diplomatic, graphetic) can be encoded as compatibly as possible. The description of a combined normalization-lemmatization process reveals that even little normalization in a diplomatic text will save much manual parsing.