How Carolingian Was Bede’s Irish Computus?

One of Bede’s most celebrated works is his On the Reckoning of Time (725). It is a work that is often characterised a schoolbook that sets out how to calculate that pesky movable Easter and to situate that calculation within the crossfire of natural laws, human custom, and divine mystery. It is also partly polemical as it (sometimes very sharply) criticises people, apparently including people within his own monastic environment, who held alternative ideas or who just couldn’t count.

Unusually for an early medieval work, we can not only identify most of the sources, but we think we have some sense of how Bede’s source material was put together. This is because of a manuscript, Oxford Bodleian Library, Bodley 309, from eleventh-century Vendôme. It was once owned by Jacques Sirmond in the seventeenth century and so it is often called the Sirmond Manuscript. Sirmond thought that, although the manuscript was later, the contents were early – a variety of late antique treatises and letters on computus almost all known to Bede. Charles Jones ‘rediscovered’ the manuscript and, in an article in 1937, was able to confirm that the manuscript – or at least items 13-45 (of 54) as he listed them (see below) – pretty much represented all the sources that Bede used.[1] By 1943, when Jones published his first edition of Bede’s works on time, he expanded this to items 1-45 and proclaimed:

By all the tests that I have thus far been able to apply, the Sirmond MS is an accurate, uninterpolated transcript of the computus which Bede created in the same volume with DTR (Bedae opera de temporibus, p. 106).

This was an unfortunately bold statement because, as we shall see, it really isn’t.

Anyway, intrigued by all this, Dáibhí Ó Cróinín had a good look at the manuscript. He noted in 1983 that part of it Jones had labelled vaguely as ‘computistical bits’ (no. 26) mentioned the death of Suibine mac Commáin and gave the present year as 658. This was hard proof that Bede’s computus had an Irish provenance. In a second article, he pointed out items 35 and 36 also contained references to Patrick and ‘Palumbus’ (= ?Columbanus) that might support this further.[2]

None of that establishes that it was the compilation that Bede knew. The texts Bede used were not rare and they were widely available in various combinations across the Latin West. To be pedantic, we should really say that it was compilation like the one Bede probably used.

Doubts do not stop there. Marina Smyth, for example, argued that the dating clause alone does not mean the entire collection was assembled before Bede got it.[3] Some texts could have been copied independently from elsewhere later on. Indeed, closer scrutiny of the manuscript backs this up to an extent. There are clearly some issues with the repetitions and mistakes, which we could probably explain away as later scribes errors, but that takes us away from its accuracy. We definitely cannot include Bede’s own work in the original here, not least because it omits some bits and adds Carolingian historical notes in the chronicle. But most strikingly, item 6a On the Leap Year gives the present year as 789. Suddenly Jones’s ‘tests’ do not look as comprehensive as one might like. We could, however, retreat to his own initial idea that we are only dealing with items 13-45. Ó Cróinín’s argument would not be much diminished by this, even if it remains impossible to prove the exact extent of the 658 layer.

The focus on the Sirmond Manuscript obscures a few other issues. Most importantly, there are four other descendants of the same exemplar. These are noted by Jones (with one exception) and Ó Cróinín but, as they judged the Sirmond Manuscript to be the better textual witness despite its relative age, the others were sidelined. Intriguingly, however, these others are all based around a shared version of the Sirmond texts in which the crucial items 13-45 are split in two (16-26+40 and 27-36), divided by the Carolingian unit of items 3-12 plus some additional material.[4] We could call this the Tours recension because the oldest manuscript – now split between Tours BM 334 + Paris BnF NAL 1612 + Paris BnF NAL 1613 – is from Tours ca 819. Here are the two compared:

Sirmond Tours

[The second oldest, Geneva BdG lat 50, is online here by the way]

Now, it would be easy to argue that this is just a meaningless Carolingian resequencing of the Sirmond Manuscript contents. I am not so sure because I find two features of the Tours sequence intriguing. First, material on Victorius of Aquitaine in this sequence now directly follows the ‘computistical bits’ which are about Victorius of Aquitaine and give the year 658 – so neat that it might represent the original order? And secondly, I find it strange that the letters and argumenta of Dionysius Exiguus (14+18-19), one of THE most authoritative figures in Latin computus, are not in the Tours sequence. Typically, the work of Victorius was ditched for Dionysius in the seventh and eighth century. Within that grand narrative, it makes more sense to me that the Dionysian material wasn’t in the ‘Victorian’ 658 collection leading to someone adding it later, rather than that it was actively omitted after it had become authoritative.

Anyway: I wonder, looking at the comparison above, if the 658 layer was only originally the shorter red block, added to a Carolingian blue block, as well as a second ‘Insular’ collection marked there in green. It’s a thought.

The Sirmond Manuscript is still the better witness to the texts than the Tours recension, even if the sequence of the Tours recension makes me question how authoritative the Sirmond order is. But there is one more thing I find odd. The focus on how much is pre-Bedan or Bedan also obscures the fact that BOTH sequences are followed directly by the Capitulary of 809 – the text of a Carolingian inquest into computistical learning carried out at Charlemagne’s Aachen. One might almost start to wonder if they had a shared exemplar that was very comfortably post-Bedan and continental, with that 809 text an integral rather than accidental part of the common exemplar. There were undoubtedly earlier layers in the collection, but the work has not really been done to establish firmly which parts are early and which are late, and how they might relate to each other.

In many ways, this kind of rethink lacks major consequences. Bede must still have referred to a collection or collections like an ancestor of the Sirmond Manuscript. There is no doubt that some of the material was circulating in Ireland in the seventh century. There is no doubt that Bede engaged with Irish computistical learning. The most important consequence is that we have to think more seriously about the different ways in which these kinds of compilations of early material were shaped and reshaped in Carolingian contexts. This is itself not a new realisation.[5] It is just that, in the race to understand the earliest parts of the compilations, it seems that the later parts and people who made the manuscripts have often been left out.

[1] C. W. Jones, ‘The ‘lost’ Sirmond manuscript of Bede’s Computus’, EHR 52/ 206 (1937), 204-219.

[2] Both articles can be found in his collected essays Early Irish History and Chronology (Dublin, 2003).

[3] M. Smyth, ‘Isidore of Seville and Early Irish Cosmography’, CMCS 14 (1987), 69-102 at 99.

[4] Another important marker of the Tours recension is item 25, the Acta Synodi, is incomplete in that family, but complete in the Sirmond Manuscript.

[5] See K. Springsfeld, Alkuins Einfluß auf die Komputistik zur Zeit Karls Des Grossen (Stuttgart, 2002) and much of the ongoing work of Immo Warntjes and Jacopo Bisagni.