The data used are of two kinds: lexical items and proper names. They have been dealt with differently. For lexical items we need to know how spellings are distributed across the lexicon (lexical frequency) and how frequently they occur in running text (text frequency). For names, this is hardly practicable. Correspondences peculiar to names and patterns of variation found in names were examined by closely sampling the large number of names given in L P D and E P D , but without any quantification.