The foundation of Wendy Davies's rehabilitation of the LL charters is her analysis of the witness lists. The great majority of the charters in LL have witness lists – one of the main justifications for calling them ‘charters’ rather than ‘memoranda’. In nearly all of them the clergy and the laity are listed separately, as in the Llandeilo Fawr charters in the Lichfield Gospels and the Llancarfan charters attached to the Vita Cadoci. On the basis of recurrent names and patronymics Davies arranged most of the witness lists in three approximately chronological sequences on pp. 35–7, 41–53, and 59–69 of The Llandaff Charters. Sequence i contains lists from charters 72a to 122 (not 74) and 160–166, Sequence ii contains lists from charters 140 to 159b and 174b–211b (not 199bii), and Sequence iii contains lists from charters 168 to 174a and 212–274, plus 74 and 199bii. The probable chronological order within these broad limits differs considerably from the paginal order in LL, as one would expect from the compilers’ chronological uncertainties, discussed in Chapter 3 above. Each of the three Sequences is self-contained, with no persons being securely identifiable in more than one Sequence. (Davies's tentative identification of Athruis rex Guenti regionis in charter 165 of Sequence i with the Athrwys ap Meurig of Sequence ii is uncertain.) For ease of reference I shall number the charters in Davies's proposed Sequences as i.1–19, ii.1–64, and iii.1–61; for a concordance see Appendix I at the end of this book.

The tables in The Llandaff Charters are fuller, more logical, and more consistent than those offered by earlier scholars, which they supersede in nearly all respects. Users may regret, however, that Davies did not follow Anscombe's practice of indicating the order of attestations in each witness list, since this can indicate probable equations of persons with the same name, order of precedence or kinship of witnesses, and close or suspicious relationships between particular charters.

An ideal arrangement of charters should show a changing order of precedence between witnesses A, B, C, etc. over a chronological sequence a, b, c, etc.:

Needless to say, so ideal a progression is not to be found even in reliable Anglo- Saxon cartularies.