The Collation and Textual Criticism of Icelandic Manuscripts (l): Collation

This article describes programs (collectively named COLLATE) developed for the simultaneous machine collation of forty-four manuscripts of two Old Norse poems. Preparatory programs dealt with transcription of the manuscripts into electronic form, and with normalisation of their spelling. This yielded a complete frequency table of all variant spellings found in the manuscripts The collation itself was designed to cope with widelyvarying texts, with marginalia and with readings tagged as dubious Each reading in each ms was compared in turn with a notional master, and all variants there found in all mss were then tabulated against one another Critical here were two special functions which identified scribal variants and sense variants The accuracy so achieved was estimated at above 99%, including transcription errors. There is no theoretical limit to the number of mss and the length of text COLLATE could process A second article discusses techniques developed to analyse COLLATE's output