Based on an analysis of the 60 000 spellings found in the Ramses corpus, we show that the encoding scheme that is used for hieroglyphic texts, known as the Manuel de Codage (1988), is problematic for the development of text corpora in general. We consequently (1) argue in favour of basic principles that should be followed when encoding hieroglyphic texts in the framework of corpus projects, and (2) suggest guidelines for a new encoding scheme, especially with respect to the structure of the sign-list.