The textual history of the Historia Brittonum is well known to be complex to the point that it is all but impossible to determine what the original text contained. Some forty manuscripts are known to exist, not all of equal weight in reconstructing the text and not all of independent value, as some are clearly copies of extant manuscripts. The work was also quoted by several Anglo-Norman historians and even the French encyclopaedist Lambert of St-Omer, who may have had access to manuscripts no longer extant. Most recent editions, though, have used British Library Harleian MS 3859, of c 1100, as their base, with commentators often stating that it is the ‘best’ text, although they are generally reticent about their reasons for regarding it as such. The principal reason appears to be that it is the fullest text without the clearly interpolated passages of the pseudo-Nennian recension: it is a member of the only recension to contain the genealogies of Anglo-Saxon kings and the so-called ‘Northern History’. Since the preface attributing it to ‘Nennius’ – which is found only in another recension of manuscripts – specifically lists Saxon genealogies as among the materials he has heaped together, this has been seen as supporting the primacy of the Harleian text.
This attitude to the text is inconsistent with the contents of individual recensions, as the preface is not contained in any examples of the Harleian recension and appears only in two manuscripts containing interpolations from what may be termed the pseudo-Nennius recension. It is only copies derived ultimately from these two manuscripts that contain the attribution to Nennius: nowhere does a copy of the original pseudo-Nennius recension survive. To make matters worse, its preface appears in two quite different versions as well as in abbreviated form in the Irish translation known as Lebor Breatnach. It is now tolerably certain that this so-called Nennian preface is a later addition, in which case it cannot be taken as a guide to the contents of the archetype. Indeed, the presence of a genealogy of Hengest in §31, which was present in the archetype, may have given the author of the forged preface ample reason to include Saxon genealogies among his supposed sources. Another possibility is that as the pseudo-Nennius recension was a derivative of an ancestor of the Harleian, the Saxon genealogies had already been added, which is what ‘Euben’s’ abridgement of the so-called Northern History appears to indicate.
Moreover, it is also clear that the Harleian recension and its close relatives, the pseudo-Gildas, the Sawley and the pseudo-Nennius recensions diverge widely from the two others, the Edmundine and the unique Chartres text. It can be demonstrated that although an early redactor of the text that led to the Edmundine recension made significant verbal changes to the original, frequently to improve the style and sense, he nevertheless often retained readings that were superior to the Harleian recension and its relatives. Its chronological preface, which is a replacement for the calculations that enable the archetype of all the other recensions with the exception of the Chartres to be dated to the fourth regnal year of Merfyn Frych, King of Gwynedd, 828×9, date it to the fifth year of the English king Eadmund, AD 976. Nevertheless, it retains the computus of §16, which clearly dates the archetype of all but the Chartres recension to 828×9.
Importantly, though, the Chartres recension not only lacks the computus of §16 but also contains a rambling passage towards the end of §31, which seems to indicate that it should be dated to some point after the mid-eighth century (sicut libine abas iae in ripum ciuitate inuenit uel reperit, ‘as Slébine, Abbot of Iona (752-767) came across or discovered in the city of Ripon’). In other words, the passage dating the Historia Brittonum to 828×9 is secondary and must date the archetype of the remaining branches containing the Vatican, Harleian, pseudo-Gildas, pseudo-Nennius and Sawley recensions.
However, a cladistic analysis of contents and name forms demonstrates that the Vatican and Chartres recensions are more closely related to each other than to any of the others; the Harleian, pseudo-Gildas, pseudo-Nennius and Sawley recensions bear the mark of a redactor later than 828×9, which the computus at the end of §16 appears to date to 859. Dumville’s exposition of the muddled chronology of the Historia, which points out the author’s confusion between anno domini and anno passionis dating, completely avoids discussion of this date, which cannot be an error for 829 either on palaeographic grounds or on the basis of the confusions between AD and AP. This means that we should take the evidence of the Edmundine and Chartres recensions a great deal more seriously than has generally been done, and although Heinrich Zimmer suggested a similar conclusion only to reject it, it has generally been ignored. There are therefore two broad groups of texts: that including the Harleian may be termed the Computistical because it contains the calculation of decemnovenal (nineteen-year) Easter cycles dating from 859 that is lacking in the Chartres and Edmundine recensions and the other the Silvian because of its inclusion of an alternate genealogy of Silvius in §10, lost in the edition of 859. Previous editors have concentrated on the evidence of Computistical family instead of incorporating the occasionally superior evidence of the Silvian.