You are on page 1of 40

immo warntjes

The Computus Cottonianus of


AD 689: A computistical
formulary written for
Willibrords Frisian Mission

Abstract
The Anglo-Saxon mission to Frisia led by Willibrord must have brought fundamental texts to the Continent, not only for preaching the Gospels and for liturgical services, but also for teaching converts and for educating future priests.
Still, not a single computistical text has been associated with Willibrords mission, which is particularly surprising when considering the fact that the Easter
controversy had been a central episode in Willibrords life before (and presumably even during) his missionary life. This paper presents the case that the Computus Cottonianus of AD 689 (London, British Library, Cotton Caligula A
XV, 73r80r) was compiled in Rath Melsigi under the auspices of Ecgberht for
Willibrords mission on the basis of its sources and comparable material in Willibrords calendar, the date of composition, the provenance of the manuscript,
and the reception of the work.
Keywords
Frisian mission, Rath Melsigi, Echternach, Willibrord, Ecgberht, Wilfrid, Computus Digbaeanus of AD 675, Computus Cottonianus of AD 689, Computus
Rhenanus of AD 775.

Introduction
The manuscript known as Cotton Caligula A XV, kept in the British
Library in London, is, for various reasons, one of the most celebrated
computistical manuscripts. It consists of two independent parts, commonly labelled A and B. The second part, B, of English provenance and
dated to the eleventh (pre-Conquest) century, is one of the principal
sources for Old English computistica, since it not only contains imporThe Easter Controversy of Late Antiquity and the Early Middle Ages, ed. by Immo Warntjes
and Dibh Crinn, Studia Traditionis Theologiae, 10 (Turnhout, 2011), pp. 173212.
BREPOLSHPUBLISHERS10.1484/M.STT-EB.1.100734

immo warntjes

tant fragments of lfrics De temporibus anni, but also various Old English computistical bits.1 The main section of the first part, A, however,
which is the section that concerns us here, is of eighth-century origin.2
The exact date of its compilation is highly disputed: A dating clause on
folio 107r suggests AD 743, but then this might only reflect the inclusion of an earlier tract in a later manuscript, especially when considering
the fact that earlier dating clauses can be found in the same manuscript.3
An analysis of this dating clause certainly supports this theory. It reads:
In Christi nomine. Incipit cyclus per indictionem XImam. Et anno
quota fuerit luna Kalendis Ianuarii? Prima, et dies dominicus festi paschalis XVIII Kalendas Maii luna XV. Et quotus annus est ab incarnatione domini nostri Ihesu Christi? DCCXLIII, et recapitulatio Victurino
CLXXXIIII annus est, et primus annus Childerici regis Francorum cum
consulibus suis Carlemanno et Pipp\h/ino.4
In the name of Christ. The cycle begins, through the 11th indiction.
And which lunar age occurred on 1 January in this year? The first, and
Easter Sunday (was) on 14 April, luna 15. And which year is this from
the incarnation of our Lord Jesus Christ? The 748th, and in the Victorian recapitulation (i.e. the second revolution of the Victorian Easter
table) it is the 184th year, and the first year of Childeric, the king of the
Franks, with his consuls Carloman and Pippin.

There are three possible reasons for the composition of this passage.
First, the phrase incipit cyclus may suggest that this passage introduces an
Easter table (apparently starting with what Bede considers the first year
of the cyclus lunaris;5 or was it supposed to introduce only the 19 years
1
For this section of Cotton Caligula A XV see especially Henel (1942), xxixxiv;
Ker (1957), 1736; Liuzza (2001), 2156; Chardonnens (2007), 50912; and also Planta (1802), 456; Lowe in CLA 2, 19; Gneuss (2001), 74 (no. 411); Blake (2009), 145.
2
Part A, which extends to f 117, contains two ninth-century insertions, fols. 39
41 (Cyprian) and fols. 6572 (Computus); cf. CLA 2, 19; Bischoff (2004), 106.
3
For the question of which material in Cotton Caligula A XV can be ascribed to
AD 743 cf. also James Palmers contribution in the present volume.
4
This passage is also transcribed in Krusch (1884), 139 and transcribed and translated pp. 2189 of the present volume, where also a facsimile can be found.
5
Bede, De temporum ratione 56 (ed. by Jones (1943), 2768, trans. by Wallis
(1999), 13941). As is apparent from the Dionysiac Easter tables and especially Dionysius Epistola ad Bonifatium et Bonum (Krusch (1938), 856), the first year of the
cyclus lunaris matches the fourth year of the cyclus decemnovenalis exactly. Contrary to
Dionysius, however, Bede defines the years of the cyclus lunaris as stretching from luna
1 of the January lunation to luna 29 of the December one. Bede, very suggestively, starts

The Computus Cottonianus of AD 689

of the cyclus lunaris starting with this year?); in this case, however, one
would expect an Easter table (possibly consisting only of the 19 years of
the cyclus lunaris) to follow, which is not the case.6 Second, the year in
question, AD 743, was a problematic one concerning the date of Easter
Sunday. In Francia, the Dionysiac reckoning was not unanimously accepted at this stage; in fact, the Victorian reckoning appears to have still
enjoyed wide popularity. For this year, Victorius had listed his notorious
double dates for Easter Sunday, namely 14 April, luna 15, and 21 April,
luna 22.7 The Dionysiac date, for its part, agreed with the first of these
two dates.8 The Cotton passage in question refers both to Dionysius
annus domini and to the Victorian table after recording only one date as
lawful for the celebration of Easter in this year, namely the one shared by
both reckonings. Accordingly, this passage may have been part of a tract
advising Frankish Christians following Victorian Easter tables with double dates to celebrate Easter on 14 April rather than the alternative, 21
April;9 or, it may simply have placed Dionysiac authority over Frankish
the cyclus lunaris with the year in which the beginning of the January lunation coincided
with 1 January (i.e. luna 1 on 1 January). In this, however, Bede starts that cycle a year
early when compared to Dionysius practice, at least from an epactal point of view: The
epact of 22 March falling in the first year of the Bedan cyclus lunaris is 22, which is
the epact for the third year of the Dionysiac cyclus decemnovenalis and therefore of the
19th year of the Dionysiac cyclus lunaris. As Dionysius lunar year stretches from luna
15 of the Easter lunation to the Easter full moon of the following year, the first year of
the Bedan cyclus lunaris coincides with the fourth year of the Dionysiac cyclus decemnovenalis and therefore the first year of the Dionysiac cyclus lunaris for 8 lunations
(from luna 15 of the Easter lunation on 14 April to luna 29 of the December lunation
on 20 December). Cf. especially the discussions in Jones (1943), 388; Pillonel-Wyrsch
(2004), 3413; Warntjes (2010), 345; and the too often neglected commentary by van
der Hagen (1734), 3156, 396401.
6
The table of fols. 110r117v of this manuscript is certainly independent from
this dating clause, not only because it does not physically follow the dating clause in the
manuscript, but more importantly because it starts with AD 703; for this table see Jones
(1938), 2045; Warntjes (2010), 311.
7
Krusch (1938), 36. Among the Victorian Easter tables listing double dates for
this year, representative for some Frankish churches may be, besides the famous Gotha
table, the Victorian Easter table for AD 700771 in Paris, Bibliothque Nationale, Lat.
4860, 148r and the closely related Vatican, Biblioteca Apostolica, Reg. Lat. 586, 10r, as
the provenance of the older, ninth-century Paris MS is Reichenau (for this MS see especially Mommsen (1861), 57483; idem in MGH Auct. ant. 9, 3635; Jones (1939), 128;
Borst (2001), 2268; idem (2006), 2678; Springsfeld (2002), 989; for the Victorian
Easter table of this MS see Mommsen in MGH Auct. ant. 9, 674 and Warntjes (2010),
LXXXIVLXXXV).
8
See the tables in CCSL 123C, 555 and Wallis (1999), 397.
9
That such guidance was needed is illustrated by Gregory of Tours account of the
Easter of AD 590 in his Historiarum libri decem X 23 (ed. by Krusch in MGH SS rer.

immo warntjes

churches;10 the focus of this passage, however, is not on the Easter date.
Third, the phrase primus annus Childerici regis Francorum may suggest
that this dating clause was part of a text commemorating the accession of
Childeric III; again, however, the focus of this passage is not on the date
of the accession of Childeric.
In any of these three cases, the passage itself certainly does not appear in its original context in the Cotton manuscript, since the immediately preceding tract deals with the increase of moonlight per day,11 the
immediately following tract lists the calendrical characteristics of each
Julian calendar month, with both texts being, in terms of content, in no
way related to the dating clause. Thus, the dating clause appears to have
been part of a longer tract or table originally, and is here transmitted
outside of its original context in a compilation of excerpts, suggesting
that only this dating clause can confidently be dated to AD 743, but no
other part of the manuscript. Indeed, Lowe dates section A of Cotton
Caligula A XV to the latter half of the eighth century, arguing that it
was copied from an exemplar written in AD 743.12 Even if, then, from
the second half of the eighth century, Cotton Caligula A XV certainly
is one of the oldest computistical manuscripts that have survived. As
for the provenance of this manuscript, which is as disputed as its date,
Merov. 1,1, 5145): Dubietas paschae fuit ob hoc, quod in cyclum Victuri luna XV. pascham scripsit fieri. Sed ne christiani ut Iudei sub hac luna haec solemnia celebrent, addidit:
Latini autem luna XXII. Ob hoc multi in Galliis XV. luna celebraverunt, nos autem XXII.
Inquesivimus tamen studiosae sed fontes Hispaniae, quae divinitus implentur, in nostrum
paschae repleti sunt.
10
Note in this context that the Victorian Easter table attached as chapter 16 to the
Frankish Dial. Burg. of AD 727 only records the Latin date of 21 April, luna 22 (Borst
(2006), 371). Likewise does the one in Leiden, Universiteitsbibliotheek, Scaliger 28, 7r;
as the provenance of this early ninth-century manuscript is Flavigny and as the Easter
table proper contains Frankish annals, it is very likely that this table had been in use in
Francia in the eighth century (for the MS cf. Krusch (1880), 20910; Lindsay (1915),
459; Jones (1939), 119; Bischoff (2004), 48; for the Victorian Easter table of this MS see
especially Mommsen in MGH Auct. ant. 9, 673 and Warntjes (2010), LXXXIV).
11
Interestingly enough, however, both this tract (which is transcribed under omission of the following incipit in Warntjes (2010a), 104) and the dating clause start with
similar phrases: In nomine Dei summi incipit [...] vs In Christi nomine incipit [...].
12
Lowe in CLA 2, 19 (no. 183) states that this part of the MS is saec. VIII, copied from an exemplar written AD 743, the year mentioned on fol. 107. Thompson, in his
earlier Catalogue of ancient manuscripts in the British Museum II, 66 believes (according
to Jones) that the latest dating clause of the MS (i.e. AD 743) agrees with the date of
composition. Lindsay (1915), 461 is more cautious, stating written apparently in France
in 743 [unless transcribed from an original of that date]. Cf. also Jones (1939), 120;
Gneuss (2001), 61 (no. 311); Bischoff (2004), 106; Gomz Pallars (1987), 22; idem
(1994), 201.

The Computus Cottonianus of AD 689

a northern-eastern French scriptorium with Insular connections is the


most widely accepted opinion.13
Now, Cotton Caligula A XV holds such a prominent place in modern scholarship because of its early date and the material it contains: For
many letters originating in the late antique and early medieval Easter
controversy, it transmits one of the best, certainly the oldest, surviving
witness.14 Concerning other computistical texts, like the ps-Cyprian
Computus of AD 243 and the curious Easter table at the end of this part
of the manuscript, it transmits the only known copy.15 This latter fact
also holds true for the text that is at the centre of the present article, the
Computus Cottonianus of AD 689. A modern hand added the incipit
Cassiodorus de computo paschali at the top of the first page of the text (f
73r; cf. Plate 1 in Appendix I), an attribution that fortunately failed to
misguide modern scholars as the text itself contains, already on its first
page, dating clauses for AD 688, i.e. roughly 100 years after Cassiodorus
death.16
The first scholar to have studied this text in detail was the Wittenberg professor Wilhelm Jan in 1718 for his still unsurpassed edition of
Dionysius Exiguus computistical works. When editing the Dionysiac
Argumenta of AD 525 from a rather inappropriate manuscript (Oxford,
Bodleian Library, Digby 63), he regularly cited parallel passages from the
Computus Cottonianus.17 To be sure, however, both the Digby and the
13
CLA 2, 19 (no. 183): written probably in North-east France, in a centre with
Insular connexions. Cf. also Bischoff (2004), 106; Jones (1939), 120; Gomz Pallars
(1994), 21; Gneuss (2001), 61 (no. 311).
14
For the corpus of computistical letters and tracts included in this MS cf. p 183
below, under section 3.
15
The ps-Cyprian text is known through two copies, of which the better one in a
Rheims MS is now lost, leaving the Cotton Caligula A XV copy the only surviving manuscript witness; an edition of this text can be found in CSEL 3,3, 24871, a translation
in Ogg (1955), 117; a new edition is currently being prepared by Alden Mosshammer.
For the Easter table cf. n 6.
16
Krusch (1884), 1134 refers to Cotton Caligula A XV as the only manuscript
known to him to contain the Computus attributed to Cassiodorus. The Cotton text,
however, is datable to AD 688/9, the work ascribed to Cassiodorus to AD 562. Lehmann (1912), repr. (1959), 478 clarified that the Computus Cottonianus of AD 689
is not a later recension of the Computus paschalis of AD 562; his own description of
the Cotton text, however, is problematic in itself, since it suggests that this computus
consists of the 16 argumenta published under Dionysius name by Jan (i.e. that the Computus Cottonianus of AD 689 contains the exact same argumenta as the Computus Digbaeanus of AD 675), which is not correct.
17
Jan (1718), 7994 (for the use of the Cotton MS see also ibidem, 513); Jans
entire book on Dionysius computistica is repr. in PL 67, 453520 (with the Argumenta

immo warntjes

Cotton text must be regarded as independent treatises, ultimately based


on the Dionysiac Argumenta, but considerably extending the original
Dionysiac corpus. In the end, Dionysius set a trend with his Argumenta,
which present, as far as we know, the earliest computistical formulary
written in Latin. Latin computistical formularies started to flourish (or
rather mushroom) to an exceptional degree from the eighth century onwards; before the end of the seventh century, however, only five computistical formularies are known, namely Dionysius original Argumenta of
AD 525, its slightly extended version (transmitted in Group B manuscripts), the Computus paschalis of AD 562 attributed to Cassiodorus,
the Computus Digbaeanus of AD 675 just mentioned, and the Computus
Cottonianus of AD 689 under discussion here.18 I have dealt with the development of computistical formularies written in Latin from Dionysius
to the Computus Digbaeanus of AD 675 at the first conference on the
science of Computus two years ago, so that it is hoped that this study of
the text immediately following in that development will round off the
picture of Latin computistical formularies composed before the start of
excessive compilation in the eighth century.19
Before the 20th century, then, modern scholars principally studied
the Computus Cottonianus only for its relation to Dionysius and Cassiodorus formularies, arriving at the conclusion that this text was not
a simple adaption to AD 688/9 of either Dionysius or Cassiodorus
published on cols. 497508, the use of the Cotton MS on cols. 47981). Kruschs edition of the Dionysiac Argumenta in (1938), 7581, is inferior to Jans, not least because
Krusch did not consult the Computus Cottonianus as Jan had done.
18
There still is no reliable edition of Dionysius original Argumenta, which must
be based on the manuscripts transmitting the extended version (classified as Group B
in Warntjes (2010a), 647); the Cassiodorian computus is edited by Lehmann (1912),
525; the Computus Digbaeanus is edited under Dionysius name by Jan and Krusch
(cf. previous note); the first 2/3 of the Computus Cottonianus are edited by Gomz Pallars (1994), 2231, repr. (1999) 1009 (note that the edition in the 1994 article is here
preferred to the 1999 reprint, as the apparatus appears strictly at the bottom of the page
only in the former, while it breaks the Latin text in the latter). Only very recently I have
realized that the computistical argumenta for AD 695 in Munich, Bayerische Staatsbibliothek, Clm 14725, 14rv belong to a cohesive formulary extending folios 12r14v
(an article on this text will be published shortly). A critical edition of all these texts is in
preparation; only such an edition will provide conclusive evidence for the relationship
between them, while it will, at the same time, unveil the development of computistical
formularies written in Latin between AD 525 and 700.
19
Warntjes (2010a). Again, note should be taken in this context of the formulary
of AD 695 in Clm 14725 referred to in the previous note; this text, which appears to
have also originated in the circle of Willibrord (like the Computus Cottonianus, as will
be proven in the following), provides important insights into the development of certain
argumenta, especially of Argumentum XIV.

The Computus Cottonianus of AD 689

text. If this text, then, represents an original work of AD 689 (obviously


dependent on the Dionysiac and ps-Dionysiac argumenta), by whom,
or rather where was it compiled? Since the groundbreaking studies of
Charles W. Jones on Bedan and pre-Bedan computistica in the first half
of the 20th century, part A of the Cotton manuscript, and with it the
text in question here, was generally placed in an Insular context.20 A new
perspective on the Computus Cottonianus was then introduced by Alfred Cordoliani in 1942, repeating his arguments in a 1958 publication.
Influenced by his recent work on Spanish manuscripts, he argued for a
Spanish origin of this text. Cordoliani, a codicologist by training and
profession, was one of the most prolific modern commentators on computistical manuscripts, with his list of articles on this subject extending
over more than 50 titles.21 Unfortunately, he had never really found a
way of classifying computistical tracts, and as a textual scholar without
scientific background, he trusted incipits of texts more than their content, a practice most unsuitable for computistical treatises.22 As a result,
many of his identifications are unreliable, and the same holds true for
the text in question here. Having studied the codex Nouvelle acquisition
latine 2169 of the Bibliothque Nationale in Paris, a ninth-century manuscript written in a Visigothic hand, he arrived at the conclusion that
the tract on folios 5v9v showed considerable parallels to the Computus Cottonianus. Unfortunately, however, Cordoliani had never seen the
Cotton manuscript in person; he simply worked from Lehmanns rather
meagre (and wrong, as pointed out in note 16 above) description of that
text. The formulary in the Paris manuscript contains an algorithm for
calculating the Hispanic era; because of the alleged parallels between
this and the Cotton text, Cordoliani assumed that the latter may also
include this era, which would certainly point to a Spanish origin of the
Computus Cottonianus. Yet, as long as he could not examine the Cotton
manuscript in person to verify this hypothesis of the Spanish era featuring in the Cotton text, he conceded that doubts must remain about this

Jones (1937), 208; idem (1943), 112.


Cf. the (incomplete) list of Cordolianis articles in CCSL, Clavis patristica
pseudepigraphorum medii aevi 3A, 20912. For Cordolianis career and work cf. Borst
(1998), 157.
22
A noteworthy exception, or possibly rather a late realisation of the inappropriateness of his earlier approach, is Cordoliani (1961a), 17680, where he lists argumenta
by topic rather than incipits; this is the example that should be followed in the future.
20
21

immo warntjes

theory.23 In fact, the Hispanic era is nowhere to be found in the Cotton


manuscript, neither in the text that concerns us here, nor in any other
part of the codex. Cordoliani then returned to the topic 16 years later. In
this publication, he exclusively referred to the aforementioned section in
the Paris MS as Computus Cottonianus as if it was identical with the relevant section in Cotton Caligula A XV, a most inappropriate and confusing choice of terminology. This Paris text he described and compared
in detail with the Dionysiac (and ps-Dionysiac) Argumenta, as well as
the Computus paschalis attributed to Cassiodorus. Yet, he apparently had
still not studied the Cotton manuscript.24
In the end, what Cordoliani should have done (but never did) is a
detailed comparison between the relevant sections of London, British
Library, Cotton Caligula A XV and Paris, Bibliothque Nationale, Nouvelle acquisition latine 2169. He would then have realized that he was
mistaken in his assumption that the Cotton and the Paris texts are closely related (save for the trivial fact that many of the formulae of both texts
are ultimately based on the Dionysiac Argumenta, as is the case with all
of the other numerous computistical formularies that have survived),
and that therefore conclusions drawn about the Paris text of AD 817 (!)
have no bearings on the Cotton text of AD 689. Still, the quintessence
of Cordolianis articles, particularly due to his misleading terminology
and the misguided title of especially his second publication on the topic
(Textes de comput espagnol du VIIe sicle. Le Computus Cottonianus),
was the (completely unfounded) theory that the text in Cotton Caligula
A XV ascribed by a modern hand in that manuscript to Cassiodorus and
containing various dating clauses for AD 668/9 was, in fact, of Spanish
origin.
It appears that it was precisely this theory that triggered Joan Gmez
Pallarss interest in the Cotton manuscript. His initial study, a 1986 Barcelona Ph.D. thesis entitled Estudis sobre el Computus Cottonianus,25 was
subsequently followed by the publication of his main results in various
articles. Gomz Pallars did what Cordoliani should have done before
publishing his unfounded theory, namely a detailed comparison of the
Cotton text with the known early ninth-century Spanish computistical
23
Cordoliani (1942). For Lehmanns problematic description of the Cotton text
cf. n 16.
24
Cordoliani (1958).
25
As Leofranc Holford-Strevens has kindly informed me, Gomz Pallarss Ph.D.
thesis is now available online at: http://www.tesisenxarxa.net/TDX-0428108-155819/
index.html.

The Computus Cottonianus of AD 689

manuscripts. His thorough analysis, then, led to the main result that the
Computus Cottonianus was textually not close enough to the early ninthcentury Spanish texts to conclude for a direct dependency between
them.26 By establishing this fact that the Computus Cottonianus was disconnected from Spanish computistica, the only argument forwarded by
Cordoliani for a Spanish origin of the Cotton text was proven wrong. In
fact, his detailed study convinced Gomz Pallars that Joness placement
of Part A of the Cotton manuscript, and with it of the Computus Cottonianus of AD 689, in an Insular context was to be endorsed.27
Yet, the Insular context of the eighth-century Cotton manuscript
does not necessarily imply an Insular origin of the text datable to AD
688/9; like the Paschal letters incorporated in this codex, it may only
have been the transmission of this text, not its origin that was Insular.
A detailed study of the contents of the Computus Cottonianus is needed
to draw conclusions about its origin. The present paper, then, provides
such an analysis, suggesting that the Computus Cottonianus was written
for Willibrords Frisian mission of AD 690 in the Anglo-Saxon monastery of Rath Melsigi (or in whichever Anglo-Saxon centre in Ireland this
undertaking was prepared) under the auspices of Ecgberht.

Definition of the Computus Cottonianus of AD 689


First of all, the Computus Cottonianus of AD 689 needs to be defined.
Gomz Pallars, the only scholar who has worked intensively on this
manuscript, and especially the section in question, regularly refers to the
Computus Cottonianus as comprising folios 73r to 77r of Cotton Caligula A XV.28 There can hardly be any doubt about the beginning of this
text, since it starts on a new quire and the immediately preceding folio
(72v) is, in fact, blank, as Plate 1 in Appendix I illustrates.
The end of this text is less easy to establish, and the reasons for setting
it on folio 77r appear, principally, to be three-fold: First, the calendrical formulae end here, with the following tracts being of an explanatory
Gomz Pallars (1987), 2141, especially 2931; idem (1989), 5762, especially
602.
27
Gomz Pallars (1987), 32; idem (1994), 99.
28
Since Cordoliani had never studied the Cotton MS, he did not comment on the
extent of this formulary. Gomz Pallars (sometimes mistakenly giving the impression
that he relied on Cordoliani in this question) defines this text as covering folios 73v77r
in (1987), 267; (1989), 57, 601; (1994), 99; he also edits this part in (1994), 1019.
26

immo warntjes

rather than an algorithmic nature. Second, the comparison of the Cotton formulae with the Computus Digbaeanus of AD 675 (i.e. the Argumenta published under Dionysius name) has shown that the two texts
are somehow related; accordingly, it is a natural reflex to set an end for
the Computus Cottonianus at a point when the material of this related
formulary ends; in fact, the last of the argumenta shared by both texts,
the pseudo-Dionysiac Argumentum XIII, can be found on folio 76v
(with this overlooking the fact that 2 of Argumenta XVI and III can
be found later in the manuscript, on folio 79rv). Third, none of the argumenta before folio 77r have the character of an independent treatise;
on folio 77r, then, occurs the text known as Suggestio Bonifati primiceri,
a very short letter of the primicerius Boniface to the pope, explaining
the Easter data for the year AD 526;29 Boniface had been the addressee
of Dionysius Exiguus famous letter outlining the details of the 19-year
cycle, and he therefore appears to have been appointed by the pope with
the task of establishing the correct date for the celebration of Easter of
AD 526; for this year, Victorius had calculated luna 22 for Easter Sunday, which was considered uncanonical by the Alexandrians.30
Consequently, the crucial question concerning the extent of the
Computus Cottonianus is whether or not it appears likely that this Suggestio was an integral part of a computistical formulary. Since the lunar
age for Easter Sunday of the year AD 526 is calculated in detail in the
Suggestio on the basis of the Dionysiac Argumentum IX, the character of
this short tract (though explaining the Easter data of only a single year)
is very much that of a formula.31 Hence, in terms of content there is no
obvious reason why the Suggestio should not have been just one of many
argumenta included in a formulary. In the case of the Computus Cottonianus, the palaeographic evidence suggests, in fact, that the Suggestio is
an integral part of this formulary rather than marking its end: There is
no indication on folio 77r that would suggest a break between two texts,
as Plate 2 in Appendix I illustrates (at least not as long as every single
formula is considered as an independent tract).
On the contrary, the more general composition of the computistical half of Part A of Cotton Caligula A XV reveals that the Computus
This text is edited by Krusch (1926), 568.
The Epistola Dionysii ad Bonifatium et Bonum is ed. by Krusch (1938), 826; for
the Victorian and Dionysiac data for Easter Sunday of this year see Krusch (1938), 51
and 69 respectively.
31
Cf. Dionysius Argumentum IX (ed. Krusch (1938), 77); for the differences between the Suggestio and this argumentum cf. Krusch (1926), 56.
29
30

The Computus Cottonianus of AD 689

Cottonianus extends to folio 80r, as only there is a break between two


sections (or rather texts) evident. The computistica of Part A of Cotton
Caligula A XV can be divided into the following sections:
1) fols. 65r72r: A formulary of unknown date (since it does not
contain an explicit dating clause), defined by the beginning of
the computistical part of the manuscript and the blank folio 72v,
clearly marking the end of a section; Lowe, in fact, argues that this
quire (4 double leaves) is a ninth-century addition.32
2) fols. 73r80r: The Computus Cottonianus of AD 689; the beginning is clearly defined by the preceding blank folio 72v and the
beginning of a new quire on folio 73r; from folio 73r there is no
evident break in the text until folio 80r, of which roughly the last
fifth was originally left blank, with this space being filled by a later
hand with an interesting sequence of Victorian epacts (cf. Plate 3
in Appendix I); note that in the 1802 Catalogue of Cotton manuscripts in the British Museum, Planta also treats this section as one
single text.33
3) fols. 80v105v: Compilation of Paschal letters and tracts in the
following order: Acta synodi Caesareae, recension B (De ordinatione feriarum); Disputatio Morini; Epistola Dionysii ad Bonifatium et Bonum; De sollemnitatibus; Epistola Proterii; Epistola
Pascasini; Epistola Cyrilli; excerpts from Gaudentius De pascha;
ps-Cyprian Computus of AD 243.
4) fols. 106r109v: Computistical formulae, tables and diagrams.
5) fols. 110r117v: A very interesting table (or rather text), still unpublished, apparently analyzing which data of the Dionysiac 95year Easter table have to be recalculated after the expiration of
this period; since the first year discussed is AD 703, it may be assumed that the text itself was compiled in the years preceding; it
may be noteworthy in this respect that the Easter table originally
brought by Willibrord to the Continent consisted only of the period AD 684702 (cf. n 56), so that the analysis outlined in the
text in question here may, in fact, reflect the need of Willibrords
circle to extent the table ending in AD 702.34

CLA 2, 19.
Planta (1802), 45.
34
For this table, cf. n 6.
32
33

immo warntjes

The Computus Cottonianus of AD 689: A Product of the Circle of Ecgberht and Willibrord
Having defined the extent of the Computus Cottonianus of AD 689 as
comprising folios 73r80r of Cotton Caligula A XV, we can now turn
to the question of the origin of this text. The analysis will focus on the
following criteria, each discussed in turn: the sources and content of this
text, its date of composition, its transmission, and finally its reception.
Sources and Content
As mentioned in the introduction, only four computistical formularies
pre-dating the Computus Cottonianus are known at present, namely the
original corpus of the Dionysiac Argumenta, its slightly extended version
(transmited in Group B manuscripts; it is therefore termed the Group B
corpus in the following), the Computus paschalis of AD 562 (an adaption
of the Dionysiac Argumenta to AD 562), and the Computus Digbaeanus
of AD 675 (an extension of the Group B corpus). Since the author of the
Cotton text adapted the Dionysiac Argumenta to his annus praesens,35
and since otherwise these argumenta show little variation, it is difficult
to establish if he worked from any of these four texts directly, or from a
now lost recension. Argumenta III and IV appear in their original Dionysiac form (except for the adaption to AD 688, of course) in the Computus Cottonianus, i.e. they do not include the ps-Dionysiac additions
transmitted in the Group B corpus and the Computus Digbaeanus;36 this
suggests that the author of the Cotton text may have worked directly
from the original Dionysiac corpus. On the other hand, he also included
material that was not part of the original corpus, but which was transmitted in the Group B corpus as well as the Computus Digbaeanus, like
the ps-Dionysiac Argumenta VII and XIV, while the addition to Argumentum III appears later in his text.37 Therefore, at this stage, it is impossible to be more specific than to argue that this author worked from one,
35
A full inventary of the contents of the Computus Cottonianus of AD 689 can be
found in Appendix III.
36
Argumentum III can be found on f 73r, Argumentum IV on f 73v of London,
British Library, Cotton Caligula A XV (ed. in Gomz Pallars (1994), 234; cf. the
commentary in Jan (1718), 824). For the additions cf. especially Warntjes (2010a), 46
8, 824; more generally for the Dionysiac and ps-Dionysiac Argumenta in the Computus
Cottonianus ibidem, 745.
37
Argumentum VII on f 73v, Argumentum XIV on fols. 74r75r (ed. in Gomz
Pallars (1994), 247; for Argumentum XIV cf. Jan (1718), 902), 2 of Argumentum
III on f 79rv (that this paragraph occurred later in this MS is not mentioned by Gomz

The Computus Cottonianus of AD 689

possibly even two recensions of the Dionysiac Argumenta. One of these


recensions, however, appears to have been the Computus Digbaeanus or a
closely related text. This is evident from the fact that the Computus Cottonianus contains the second part of the ps-Dionysiac Argumentum XVI
which is first attested in the Computus Digbaeanus of AD 675.38 These
two texts also transmit the ps-Dionysiac Argumenta XIXIII, which are
otherwise only known in one earlier version, to be found in the Fragmentum Nanciacense; most conclusively in this respect, the Computus
Cottonianus and the Computus Digbaeanus share the same version (with
some variation) of Argumentum XIII, which differs markedly from the
one in the Nancy fragment.39 Moreover, if the argumenta of the Computus Digbaeanus are compared in detail with those of the Computus Cottonianus, many more correspondences reveal themselves. It may be sufficient here to draw attention to the parallels in Argumentum I outlined
in Appendix II, which demonstrates that the argumenta of the Cotton
text are in many instances closer to the Computus Digbaeanus than to
the Group B corpus of the Dionysiac Argumenta, which is the format in
which these Argumenta circulated in the Latin West (note especially the
omission of the phrase qua argumentum integrum possit custodiri XV non
adsumas, sed ad sumam primam unum semper adicias, ut puta in both the
Digby and the Cotton computus).
Now, the important aspect for the present study is the fact the Computus Cottonianus is closely related, in whatever form, to the Computus
Digbaeanus. This suggests that both texts originate from a similar, if not
the same, intellectual milieu. Concerning the Computus Digbaeanus,
it has been argued that the place of its composition probably is to be

Pallars or especially Jan (1718), 83); for these cf. especially Warntjes (2010a), 468,
513, 824, 912.
38
2 of Argumentum XVI on f 79r (see Jan (1718), 94). Note that the palaeographical evidence of Oxford, Bodleian Library, Digby 63, 78v may suggest that Argumentum
XVI constituted a tract independent of the rest of the Computus Digbaeanus; for this
argumentum see especially Warntjes (2010a), 45, 53, 924.
39
Argumenta XIXII on f 75v, Argumentum XIII on f 76rv (ed. in Gomz Pallars (1994), 2730; cf. the commentary in Jan (1718), 8890, where Argumentum XII
is transcribed in full from this MS, while its variants are given for the other two argumenta); for these three argumenta cf. especially Warntjes (2010a), 5051, 8991, 95.
Argumentum XIII is edited from the Fragmentum Nanciacense in Warntjes (2010a),
967, from the Computus Digbaeanus in Jan (1718), 90 and Krusch (1938), 78, from the
Computus Cottonianus in Gomz Pallars (1994), 2930; for the differences between
them cf. the detailed discussion in Warntjes (2010a), 97105. For the Fragmentum Nanciacense see especially Warntjes (2010a), 6972 (description), 1101 (facsimile).

immo warntjes

identified with an Anglo-Saxon monastery in Ireland.40 Accordingly,


due to the close relation between these two texts, the same appears to be
true for the Computus Cottonianus.
From this perspective, then, a connection of this formulary with the
circle of Willibrord seems justified: The most reliable information concerning Willibrords career comes from Bede, who wrote his Historia
ecclesiastica gentis Anglorum at a time when Willibrord was still alive,
and who may have been very well informed about Willibrords mission
through Bishop Acca of Hexham.41 Valuable additions to Bedes account
can be gathered from Alcuins Vita Willibrordi, which, however, has to
be read with caution not only because of its chronological distance from
the events described, but also because of its partisan character, with Alcuin probably being a relative of Willibrord and composing this text for
the Echternach abbot Beornrad.42 Still, this should not trouble us here,
since we are only interested in the basic framework of Willibrords life.
Alcuin relates that Willibrord entered the monastery of Ripon as a boy
and stayed there until the age of 20, when he went to Ireland for the sake
of studying.43 The same author argues that Willibrord stayed in Ireland
for 12 years and left for Frisia in the 33rd year of his life, and this mission,

Warntjes (2010a), 925.


Bede, Historia ecclesiastica V 11 (Plummer (1896), i 303): Ipse autem Uilbrord,
cognomento Clemens, adhuc superest, longa iam uenerabilis aetate, utpote tricesimum et
sextum in episcopatu habens annum, et post multiplices militiae caelestis agones ad praemia remunerationis supernae tota mente suspirans. In chapter III 13 (Plummer (1896), i
1524), Bede relates that Acca of Hexham spent some time with Willibrord on his way
to Rome and that on this occasion the missionary told many stories about miracles performed in the name of St Oswald, which Acca then in turn passed on to Bede; from this
account especially Kirby (19656), 349; Fritze (1965), 2356; idem (1971), 10910;
and others suppose that Acca was Bedes principal source also for Willibrords mission in
general; more cautious Levison (1908), 3345. A general overview of the sources (and
their limits) for Willibrords continental activity is magisterially presented by Levison
(1940a).
42
That Alcuin belonged to Willibrords wider family has been concluded from
Alcuins own statement in the praefatio of his Vita Willibrordi (MGH SS rer. Merov.
7, 114; note that the praefatio is omitted in Veyrard-Cosme 2003 edition), in which he
argues that he presides, at the time of composition, over a foundation of Willibrords
father by right of legitimate succession; cf. Dmmler (1893), 545; Levison in MGH
SS rer. Merov. 7, 94; Angenendt (1973), 97; von Padberg (1997), 21. The dedication
to Beornrad can also be found in the praefatio (MGH SS rer. Merov. 7, 113; omitted in
Veyrard-Cosme (2003)). For the date of Alcuins Vita Willibrordi see Levison in MGH
SS rer. Merov. 7, 945.
43
Alcuin, Vita Willibrordi 34 (MGH SS rer. Merov. 7, 1178; Veyrard-Cosme
(2003), 40, 42).
40
41

The Computus Cottonianus of AD 689

as will be outlined presently, took place in AD 690.44 Accordingly, if Alcuins Vita Willibrordi is reliable in this chronological data, Willibrord
turned 20, the age at which he left Ripon, in AD 678.45
This information has numerous implications. Ripon itself holds a
prominent place in the seventh-century Easter controversy. The Deiran
king Alhfrith, sub-king to his father Oswiu, founded that monastery in
the AD 650s and gave it to followers of Irish traditions. Shortly afterwards, Alhfrith converted to Roman customs, apparently impressed by
the imposing figure of Wilfrid or Benedict Biscop. In the course of this
process, Alhfrith expelled the monks clinging to Irish traditions from
Ripon and made Wilfrid himself abbot of that monastery.46 From this
time, then, Ripon must be regarded as a Romanist stronghold, and there
can hardly be any doubt that Willibrord was educated an ardent Romanist (especially in terms of the method for calculating Easter) being an
alumnus of Wilfrid.47 In the aftermath of the Synod of Whitby, Wilfrid
44
Alcuin, Vita Willibrordi 45 (MGH SS rer. Merov. 7, 119; Veyrard-Cosme
(2003), 44). For the date of Willibrords mission see pp. 1923 below.
45
Doubt about the accuracy of this information may arise from the fact that the
age ascribed to Willlibrord when starting his missionary activity corresponds suspiciously with the age traditionally assigned to Christ at the time of his crucifixion. More generally for Alcuins use of hagiographical features and his remodelling of historical facts see
Fritze (1971), 13240; Schferdiek (1994), 1835; the most detailed study of Alcuins
hagiographical style and his models is Veyrard-Cosme (2003), 142415, where the passage in question is discussed pp. 3501.
46
For Ripon and the relation between Alhfrith and Wilfrith see Bede, Historia
ecclesiastica III 25, V 19 (Plummer (1896), i 1823, 325); idem, Vita Cuthberti 78 (Colgrave (1940), 174, 176, 180); Stephen of Ripon, Vita Wilfridi 78 (Colgrave (1927),
149). For the relation between Alhfrith and Benedict Biscop see Bede, Vita beatorum
abbatum 2 (Plummer (1896), i 365).
47
In the older Vita Wilfridi by Stephen of Ripon (cap. 26), Willibrord is called
Wilfrids filius, Inbripis nutrius (Colgrave (1927), 52). A good, balanced discussion of
Wilfrids Romanism can be found in Mayr-Harting (1986), 26. For Wilfriths career
from the perspective of the Easter controversy see especially Harrison (1976), 625,
725; his influence on Willibrord in this matter is also stressed ibidem, 734, while
already Hauck (1887), 405; Levison (1940), 315; idem (1946), 55 have more generally
argued that Willibrord had been brought up and educated in a fervently Roman environment (a view subsequently challenged on various grounds especially in the context
of Benedictine monasticism, church organisation, and missionary ideas and ideals , not
so, however, in respect of the method for calculating Easter; cf. especially Fritze (1969),
Angenendt (1973), 85104, van Berkum (19789), and also Weiler (1989), 835; when
reassessing Willibrords Roman influences, Angenendt (1990), 1920 omits the Easter
question; cf. also n 52 below); Flaskamp (1929), 10 even goes so far as to call Willibrord
a second Wilfrith, but he is also one of the first to note that Willibrords monastic organisation owed plenty to Irish customs (p 18); for the close relation between Wilfrith
and Willibrord see also Wampach (192930), i 189. Just very recently, Crinn

immo warntjes

became bishop of York, but was deposed by King Ecgfrith in AD 678.


In fact, he was not only deposed, but driven out of Northumbria and he
did not return before the death of the king.48 Wilfrids expulsion must
have had wider than only personal consequences. His students certainly
faced a difficult situation with their mentor falling out of favour with
the reigning king. It may therefore not have been a coincidence that Willibrords departure corresponded, chronologically, with Wilfrids deposition.49 Yet, while Wilfrid spent the following years on the Continent,
Willibrord joined Ecgberhts community in Ireland, to be identified
with Rath Melsigi.50
Accordingly, Willibrord was educated by the person who apparently
first introduced the Dionysiac reckoning into England (at least the first
whose efforts showed far-reaching success) and who became its most ardent propagator in England in the second half of the seventh century,
not the least as the main advocate of the Roman cause at Whitby; Willibrord then became the student of the greatest champion of that reckoning in Ireland, his fellow Northumbrian Ecgberht, the very man who
later converted the last and most passionate stronghold of Irish customs,
Iona, to the Dionysiac system.51 This certainly implies that computistics
played an immensely important role in Willibrords education and that
(2007), 21 n 92 has advanced the suggestion that Wilfrid and Willibrord may even have
been related.
48
Bede, Historia ecclesiastica III 28, IV 2, 3, 12 (here the date of Wilfrids expulsion), 13, V 19, 24 (here also the date of Wilfrids expulsion) (Plummer (1896), i 1945,
2056, 22832, 3257, 3545); Stephen of Ripon, Vita Wilfridi 1115, 24 (Colgrave
(1927), 2233, 4851). For Wilfriths career in general see especially Poole (1919);
Duckett (1967), 101214; Farmer (1974); Kirby (1974); Isenberg (1978); Mayr-Harting (1972), 10713, 12947; idem (1988).
49
Cf. Hauck (1887), 4056; Wampach (192930), i 224; Levison (1940), 315
6; idem (1946), 55. Van Berkum (19789), 38591, and then Weiler (1989), 86 following his lead, question this view on the basis that the sources do not explicitly connect
the two departures; these two Dutch scholars rather suggest the opposite chronology of
events by arguing that Willibrord probably left Ripon before Wilfriths departure.
50
Alcuin, Vita Willibrordi 4 (MGH SS rer. Merov. 7, 1189; Veyrard-Cosme
(2003), 42). For Rath Melsigi see especially Crinn (1984).
51
For Wilfrid being taught in Roman customs, including the calculation of Easter,
by Archdeacon Boniface in Rome in the AD 650s see Stephen of Ripon, Vita Wilfridi
5 (Colgrave (1927), 12); Bede, Historia ecclesiastica V 19 (Plummer (1896), i 324). For
him advocating Roman methods for calculating Easter in England see Stephen of Ripon,
Vita Wilfridi 7, 10 (Colgrave (1927), 147, 203); Bede, Historia ecclesiastica III 25,
V 19 (Plummer (1896), i 1819, 325, 330). For Ecgberhts role in the Irish Easter controversy see Bede, Historia ecclesiastica III 4, 27, V 9, 22, 24 (Plummer (1896), i 1345,
1914, 2967, 3468, 356); idem, Chronica maiora 586 (MGH Auct. ant. 13, 319).

The Computus Cottonianus of AD 689

he himself must have been very well acquainted with the Dionysiac reckoning.52 More conclusively, Willibrord studied for more than ten years
in an Anglo-Saxon monastery in Ireland, the very milieu which probably saw the production of one of the principal sources for the Computus Cottonianus, namely the Computus Digbaeanus of AD 675. When
he was sent over to the Continent to set up a missionary base in Frisia,
there can hardly be any doubt that his mentor Ecgberht, a fervent follower of Dionysius, provided him with a formulary containing the most
basic rules of the Dionysiac reckoning, like the Computus Cottnoiaus of
AD 689; this argument is even more plausible when considering the fact
that the Frankish church still followed the Victorian reckoning, so that
texts teaching Dionysiac truth could not be acquired in or close to the
area of Willibrords activity. Judging from the sources and the historical
background, then, it appears absolutely probable that the Computus Cottonianus was composed under the auspices of Ecgberht for Willibrords
Frisian mission.
This hypothesis is further strengthened by one of the original passages of the Computus Cottonianus. On folios 78v79r, the author of
this text introduces, for the first time, a comparison of the lengths of the
lunar months (lunations) according to different customs, headed Quattuor differentiae lunae, which became fairly popular in eighth-century
Frankish computistics.53 In this comparison, the sequence of lunations
52
Note that the question of the Easter reckoning followed by Willibrord lies outside of the debate about Anglo-Saxon versus Irish influence on Willibrords continental activity, fairly recently summarized by Eugne Hone (2000): Both in Northumbria
and in Ireland Willibrord studied only under staunch supporters of Dionysius and in the
Easter controversies of both regions as well as later on the Continent he certainly will have
promoted the Dionysiac reckoning, not least by use or introduction of the Computus Cottonianius; so in this question, he was Roman through and through; cf. also n 47 above.
53
London, British Library, Cotton Caligula A XV, 78v79r: IIII deferentiae lunae: Victorini XXX; Anatolius et laterculus alius XXX, alius XXVIIII; apud Ebraeos XXVIIII et semis; apud Aegyptus XXVIIII. For the eighth- and early ninth-century reception
of this passage cf. Bobbio Computus 15 (PL 129, 1287); Lib. ann. 58 (Borst (2006),
7534); and also the Computus Rhenanus to be discussed below (C 187r; W 171r). Anatolius name is, in fact, spelt correctly only in very few of the manuscripts used by Borst
for the edition of this passage of Lib. ann. of AD 793, a fact that led Springsfeld (2002),
3667 to the wrong conclusion that the spelling Anatolius was a later correction; the
earliest witness, from a textual as well as from a manuscript point of view, the Computus
Cottonianus, clearly reads Anatolius; this evidence was not known to Springsfeld, and
neither was the fact that Anatolius and the latercus applied the same sequence of lunations. The laterculus of this passage was first identified with an 84-year Easter table by
Krusch (1910), 233 n 5, without any further specification; the table referred to is, in
fact, the 84 (14)-year Easter table followed by the Irish (i.e. the latercus); for the correlation/dependency between Anatolius and the latercus in the sequence of lunations and a

immo warntjes

of the latercus (here laterculus, i.e. the 84-year Easter reckoning unanimously used by the Irish until the early seventh, in the northern part,
particularly in Iona, until the early eighth century) is referred to as being identical with the one applied by Anatolius. Anatolius writings were
cited as authoritative by both parties involved in the Synod of Whitby,
in particular by Wilfrid.54 Wilfrid, being the main advocate of the Roman party, would not only have known the details of the most important texts on the paschal question, but also the technicalities underlying
the system supported by his adversaries. It would not surprise, then, if a
student of Wilfrid included such details in his computistical work.
Most intriguingly, however, another original passage of the Computus Cottonianus can be directly connected to Willibrord. Among the
Cotton formulae occurs an algorithm for the calculation of the weekday on the Calends (first day) of each month from the concurrentes (the
weekday of 24 March) of the year in question and a certain regular for
each month (which is to be identified as the weekday-difference between
24 March and the Calends of the respective month). The Computus Cottonianus is the first text to outline this algorithm, but it soon became
extremely popular, as it features in the Bobbio Computus, Bedes De
temporum ratione of AD 725, the Frankish Dial. Neustr. of AD 737 and
Lect. comp. of AD 760, and many other texts and manuscripts. Now, in
most texts and manuscripts the list of regulars starts with the one for
March; almost all remaining texts and manuscripts, including Bedes De
temporum ratione, begin this list with the January regular.55 In the Computus Cottonianus, however, the first regular is the one for October. According to Borsts edition of Lect. comp., the passage in question found
its way into (at least) 38 manuscripts; only one of these has the same
order as the one given in the Computus Cottonianus, namely Munich,
Bayerische Staatsbibliothek, Clm 14456; this manuscript is famous for
containing the only known copy of the Munich Computus, an Irish
computistical textbook of AD 719, and this may strengthen the general
thorough discussion of this passage see Warntjes (2007), 413, especially n 33. Note that
this passage in particular reveals that the Computus Cottonianus must have been written
on the British isles, if further evidence was needed for this general fact.
54
Bede, Historia ecclesiastica III 25 (Plummer 1896, i 1867).
55
January: Bede, De temporum ratione 21 ( Jones (1943), 222), which is cited by
Rabanus Maurus, De computo 73 (ed. by Stevens in CCCM 44, 289). March: Bobbio
Computus 3 (PL 129, 1282); Dial. Neustr. 8 (Borst (2006), 389); Lect. comp. I 3 (Borst
(2006), 5467); Lib. comp. II 15B (Borst (2006), 1165); Lib. calc. 16A (Borst (2006),
1396); Pacificus of Verona, Computus 68 (Meersseman and Adda (1966), 75); various
calendars (cf. Borst, (2001), 16403). Both: Computus Rhenanus (C 189v; W 173r).

The Computus Cottonianus of AD 689

argument of an Insular origin of the Cotton text.56 More revealing,


however, is the fact that among the very few argumenta accompanying
Willibrords Easter table and its continuations occurs this list of regulars
starting with October.57 Though the phrasing and some numbers show
considerable variations,58 the very rare fact that the list of both the Cotton text and Willibrords Easter table begin with October alone suggests
that both texts derive from the same intellectual milieu, namely the circle of Willibrord.
The Date
One of the most suggestive arguments for the ascription of the Computus Cottonianus to the circle of Willibrord is the correspondence of the
56
Borst (2006), 5467. In Munich, Bayerische Staatsbibliothek, Clm 14456, the
relevant passage occurs on f 66v. For this MS see especially Warntjes (2010), CCXI
CCXXI; the Munich Computus is there ed. and trans. on pp. 1317.
57
Paris, Bibliothque Nationale, Lat. 10837, 41r. Willibrords Easter table and its
continuations survive in four separate parts in this MS: 1) AD 684702 on f 44r; 2) AD
703721 on f 40v; 3) AD 722759 on f 41rv (which is the one containing the passage
in question); 4) AD 760797 on f 43rv. Only the first three parts (with part 3 concerning us here) are generally dated by Lowe (CLA 5, 26 (no. 606a)) to ante A.D. 728, but he
also states more specifically: In the Paschal table, on fol. 40v, in the margin, is a cross opposite the year 717, perhaps marking the year in which the MS. was written; part 1 may
already have been written in Ireland, the following parts, then, on the Continent, probably in Echternach. For Willibrords Easter table see especially CLA 5, 26 (nos. 606a and
606b); Wilson (1918), ix-x; Kenney (1929), 2334; Crinn (1984), 1556; Obrist
(2000), 745; Warntjes (2010), XC-XCI.
58
Computus Cottonianus (London, British Library, Cotton Caligula A XV, 75v;
Gomz Pallars (1994), 106 transcription is misleading, since he did not realize that the
list is to be read in columns rather than lines): Item Kalendae mensium, qua die intrent,
hoc calculo cognoscitur uoluentibus annis: Octember II habet litteram, Nouember V, December VII, Ianuaris II, Februarius V, Martius V, Aprilis I (add. later), Maius III, Iunius
VI, Iulius I, Augustus IIII, September VII. His semper adiect aepactas solis anni cuius uolueris, et deinceps partire per septiam partem, septies asse aut septies bini, sicque calculationis
seriem sine errore reperis. Willibrords Easter table (Paris, Bibliothque Nationale, Lat.
10837, 41r): Argumentum ebdomadarum: In Octembri I, in Nouembri IIII, in Decembri
VI, in Ianuario II, in Februario V, in Martio V, in Aprili I, in Maio III, in Iunio VI, in Iulio
I, in Agusto IIII, in Septembri VII; et adiece postea concurrentes. Note the difference in the
regulars for October to December. In Willibrords Easter table, the values are correctly
calculated for 1 October, 1 November, 1 December preceding the concurrentes of 22
March; those in the Computus Cottonianus are calculated for 1 October, 1 November, 1
December following the concurrentes of 22 March and are subsequently shifted to the beginning of the list. It appears, therefore, that the original custom was to list the regulars
from January to December, that the author of the Computus Cottonianus then decided to
change that order to an October beginning by artificially shifting the final three months
to the beginning. This order was then kept in Willibrords Easter table, the regulars for
October to December, however, corrected to agree with this order.

immo warntjes

date of composition of this text with the date of Willibrords mission.


The Computus Cottonianus includes numerous explicit dating clauses,
which refer to AD 688 at the beginning, but later in the text turn to the
following year, AD 689.59 This suggests that the text was written over a
period of two years, or, given the rather small size of the text, at the turn
from AD 688 to 689. Willibrords mission, on the other hand, is securely
datable to AD 690: First, Bedes account suggests that Willibrord set out
for Frisia in AD 690.60 This is fully confirmed by the only written passage that can confidently be ascribed to Willibrords own hand. On the
November page of Willibrords famous calendar the following statement
can be found:61
In nomine domini Clemens VVillibrordus anno sexcentesimo nonagesimo
ab incarnatione Christi uen\i/ebat ultra mare in Francia.
In the name of the Lord, Clemens Willibrordus came across the sea
to Francia in the 690th year from the incarnation of Christ.

The reason for this entry occurring on the November page is not that the
trip to Frisia was undertaken in that month (at a most unsuitable time of
year), but rather because Willibrords consecration as archbishop, which
is mentioned immediately after the above given quote, took place on 21
November in AD 695.62 Hence, the entry only reveals that Willibrord
59
The AD 688 datings on f 73rv, the AD 689 datings on f 75v of London, British
Library, Cotton Caligula A XV.
60
Bede explicitly refers to AD 690 in Historia ecclesiastica V 8; he then introduces
the following chapter (V 9), which relates Ecgberhts preparations for his trip to Frisia
and the divine command that he should turn to Iona instead, with the phrase eo tempore,
before turning to Willibrord setting out for Frisia as a direct consequence of Gods command to Ecgberht in cap. V 10 (Plummer (1896), i 294301).
61
Paris, Bibliothque Nationale, Lat. 10837, 39v; a facsimile of the passage in
question can be found in CLA 5, facing p 26, in Wilson (1918), under fol. 39b with a
transcription on p 13, in Flaskamp (1929), between pp. 80 and 81 with a transcription p
81, and in Weiler (1989), 132; cf. also Arndt (1877), 293; Hauck (1887), 4167; Levison
in MGH SS rer. Merov. 7, 912; idem (1940), 329; Wampach (192930), i 78; and
the most recent detailed discussions in Howlett (2008a), 15461 (with translation) and
Pelteret (forthcoming; I would like to thank David Pelteret for providing me with the
manuscript of his paper before publication).
62
Besides the note in the left-hand margin of the November page, there is a brief
comment under XI Kal. Dec., i.e. 21 November, in Willibrords calendar: ordinatio
domni nostri clementis. This information contradicts Bedes statement in his Historia ecclesiastica V 12 (Plummer (1896), i 3023), arguing that Willibrord was consecrated
a year and a day later, on Caecilias day, i.e. 22 November, in AD 696. The evidence of

The Computus Cottonianus of AD 689

travelled to Frisia (or rather to the Frankish court) in AD 690. Spring


or summer would generally have been the most likely seasons for such a
long sea journey. On the other hand, if Alcuins chronological data in his
metrical and prose lives of Willibrord are taken at face value, then the
expedition would, in fact, have been undertaken in the final two months
of AD 690.63
The one-year difference between the latest dating clauses in the Computus Cottonianus and the beginning of Willibrords mission should not
irritate us here. It may well be that the book production for such an expedition started well in advance of the expedition proper, since not only a
computistical manual, but also far larger books had to be produced, like a
Gospel book, all kinds of liturgical texts, a grammar, etc.;64 the computistical formulary may simply have been compiled first. On the other hand,
the expedition to Frisia may equally likely have been scheduled for AD
689 and was then postponed by a year due to unknown circumstances.
In fact, Bede explicitly says that the initial expedition under Ecgberht
was abandoned, with the leadership then being given to Willibrord; the
space of time between Ecgberhts abandonment and Willibrords setting
sail to the Continent may only have been a few days, but it may also have
been a few months.65 Moreover, Bede relates the story of an unsuccessful
attempt to convert the Frisians to the true faith by one Wihtberht, and
his return after two years of preaching in vain may have led to a period
of reconsideration of such an undertaking.66 A third possible reason for
a postponement of Willibrords Frisian mission is also implicit in Bedes
account. He argues that Willibrord and his 11 companions went to Pippin II, the Frankish major domus, who had just recently conquered Frisia
Willibrords own calendar must obviously be regarded as superior to Bedes account, the
more so since 21 November was a Sunday in AD 695 (cf. Wilson (1918), 43; Wampach (192930), i 37; Harrison (1973), 6970; however, Flaskamp (1929), 157 argues
strongly for Caecilias day on the basis that Willibrords consecration took place, according to Bede, in the church of St Caecilia in Trastevere: therefore, the date was dependent
on the place of consecration).
63
Wampach (192930), i 710, 60.
64
For the texts that an Anglo-Saxon missionary would have brought with him to
the Continent see Schlings excellent study of Bonifatius Handbibliothek (19613);
for the need of extensive preparations for Willibrords mission, not only in terms of book
production, see also Weiler (1989), 93.
65
Bede, Historia ecclesiastica V 910 (Plummer (1896), i 298301).
66
Bede, Historia ecclesiastica V 9 (Plummer (1896), i 298). Cf. Flaskamp (1929),
9, who argues that Wihtberhts failure resulted in an attitude among the Anglo-Saxon
missionaries to wait for more favourable circumstances.

immo warntjes

citerior.67 The fighting between Franks and Frisians and Pippins victory
are confirmed by Frankish sources, though their dating relies heavily
on Bedes statement and the note in Willibrords calendar: At the battle of Tertry in AD 687 (the date being inferred from internal evidence
of the Liber Historiae Francorum), the Austrasian major domus Pippin
defeated his Neustrian counterpart, the latter being subsequently killed
the following year.68 This made Pippin factually the ruler of the united
Frankish kingdoms, and he immediately, in AD 689 (if this event is connected to Bedes narrative), waged war against the Frisian dux Radbod,
whom he defeated at Duursted that year. With this victory he occupied
the south-western part of Frisia.69 Thus, in AD 689, Frisia was in political
turmoil, and if a missionary expedition was planned for that year, the circumstances certainly did not allow for its execution;70 a postponement
of the undertaking to the following year, when the political situation
was stable and even more favourable with a Christian ruler now officially
ruling a part of this heathen region, appears like a logical consequence.
Therefore, keeping the dearth of information about the beginning of
Willibrords mission in mind, the date of the Computus Cottonianus
Bede, Historia ecclesiastica V 10 (Plummer (1896), i 299).
Liber Historiae Francorum 48 (MGH SS rer. Merov. 2, 3223; for the date, 322 n
5); Fredegarii chronicarum continuationes 5 (MGH SS rer. Merov. 2, 171); Annales Mettenses priores s.a. 687 (MGH SS rer. Germ. 10, 712; the date given here is AD 690).
69
Liber Historiae Francorum 49 (MGH SS rer. Merov. 2, 3234); Fredegarii
chronicarum continuationes 6 (MGH SS rer. Merov. 2, 172); note that the year is not
explicitly mentioned in these accounts; Bedes statement that Pippin had just recently
(nuper; cf. n 67) defeated the Frisians when Willibrord set out for Frisia in connection
to Willibrords calendar entry relating that he sailed to the Continent in AD 690 makes
Rettbergs dating (1848), 503 of Pippins victory over the Frisians to AD 689 very plausible (though AD 689/690 would be more accurate). The Annales Mettenses priores s.a. 689
(MGH SS rer. Germ. 10, 13) record a victory of Pippin over Radbod under the year AD
692, but because of their late date (early ninth century) and generally unreliable chronology (for which see especially Levison (1933) and Haselbach (1970), 314), Bedes
account is given more credence here; still, it is worth noting that the same annals date
the battle of Tertry to AD 690, an event that is securely datable to AD 687 (cf. previous
note); if, therefore, the Metz annalist is consistent in his relative chronology, i.e. in the
three-year difference between the actual event and his own dating, then AD 692 of the
Annales Mettenses would, in fact, lead to AD 689 for Pippins victory against the Frisians;
on the basis of the Annales Mettenses, which record not only this one battle between the
Frisians and Pippin, but also a second one, fought at Duursted five years later (MGH
SS rer. Germ. 10, 17), Fritze (1971), 109, 1415 (cf. also Haselbach (1970), 36) dates
the battle of Duurstedt to AD 694/695, believing that Bede refers to the earlier clash
between the two armies recorded by the Metz annalist, while the other Frankish sources
appear to have telescoped these two events. See also n 77 below.
70
Levison (1946), 56 suggests that news of the political developments in Francia
of these years would have reached Ireland and Britain very quickly.
67
68

The Computus Cottonianus of AD 689

agrees reasonably well with that of the beginning of the Willibrordian


mission.
Of course, the date can also be interpreted differently. The end of the
AD 680s saw one of the peaks of the Insular Easter controversy, which
led to an extensive production of computistical texts. On a diplomatic
visit to Northumbria in AD 687, the Iona abbot Adomnn, the head
of the community that most ardently supported the latercus (i.e. the 84
(14)-year Easter reckoning), was converted to the Dionysiac reckoning.
Even though his own brethren did not follow his example, Adomnns
conversion certainly led to renewed discussions about the correct method of calculating Easter, especially in his native Ireland, which he visited
directly after his conversion.71 The southern Irish church followed the
Victorian reckoning since AD 632, while the northern Irish, under Ionas lead, still adhered to the latercus. Some of the Irish churches, as well
as the Anglo-Saxon monasteries in Ireland, may already have converted
to the third reckoning, i.e. the Dionysiac system, which was to oust the
two others in the decades to come; but only Adomnns conversion to
Dionysius in AD 687 appears to have set a trend with far-reaching consequences. It does not surprise, therefore, that one of the few datable Irish
Bede argues in his Historia ecclesiastica V 15 and V 21 (Ceolfriths letter to the
Pictish king Nechtan) (Plummer (1896), i 3156, 3456) that Adomnn was converted
to the Dionysiac reckoning when visiting Northumbria on a diplomatic mission; this
diplomatic mission probably was Adomnns negotiation of the release of 60 Irish prisoners of war noted in the Annals of Ulster s.a. 686.7 (Mac Airt and Mac Niocaill (1983),
151; for the dating of this passage see Mc Carthys reconstruction of the chronology
of the Irish annals at http://www.scss.tcd.ie/misc/kronos/chronology/synchronisms/
Edition_4/K_trad/Synch_tables/s0679-0694.htm). The fact that Adomnn did not
manage to convert his own community but was far more successful in Ireland is also
related by Bede in the same chapters. Picard (1984), 6070 is certainly right in arguing
that this account of Bede has to be read with caution, as it is unlikely that Adomnn
spent most of the rest of his life in exile in Ireland, expelled by his own community,
as Bede has it. Still, Adomnns conversion to the Dionysiac reckoning can hardly be
doubted, as Ceolfrith could not have twisted the truth or simply invented this event
when writing the letter to Ionas neighbour, the Pictish king Nechtan, only some six years
after Adomnns death. Kirby (2002), 52, more vehemently than Picard, concludes that
supporting evidence that he [Adomnn] campaigned in Ireland for the acceptance of
the Roman Easter is non-existent. Adomnn may not have expressed missionary zeal
for promoting Roman customs, but his example alone must have led to serious discussions; the influx of computistical texts in AD 689, as witnessed by the now lost Victorian
computus of AD 689 cited by the Munich computist and the tract De comparatione epactarum Dionysii et Victorii of the same year (to be discussed presently), provide just the
source material that Kirby was missing, as they demonstrate that considerable debates
took place in this year between southern Irish adherents of Victorius and presumably a
northern Irish clergy convinced by Adomnns example that the Dionysiac method was
the one to be followed. For the context cf. Warntjes (2010), CLVIICLVIII.
71

immo warntjes

computistical texts of the seventh century, a Victorian computus which


has not survived, but which is quoted in the later Munich Computus of
AD 719, was evidently compiled in AD 689.72 This text may have been
written by some southern Irish churchmen as a direct response to Adomnns conversion in order to make a stand for the Victorian reckoning
followed by them. In fact, direct evidence for adherents of Victorius
challenging the Dionysiac system in AD 689 on the basis that the latter
did not provide data for Christs crucifixion and resurrection in accordance with the Gospels comes from a text just recently introduced into
scholarship and located in this very context.73
If, then, the Computus Cottonianus had originated in this Irish context, one would expect some influence of this text in Ireland itself. Yet,
all Irish computistical textbooks written in the late seventh, early eighth
centuries avoid the Dionysiac Argumenta and their seventh-century recensions, apparently preferring older and simpler methods of calendrical
calculations.74 In general, the Computus Cottonianus has left almost no
traces in the rich Irish computistical literature of the late seventh, early
eighth centuries. The analysis of the sources and contents of the Computus Cottonianus outlined above rather revealed that this text is to be
connected to a different context, Willibrords Frisian mission, and the
date of composition of this text fits very well into that argument, which
will be further strengthened by the discussions of the provenance and
reception of this formulary.
The Transmission
The Computus Cottonianus survives in only one manuscript, London,
British Library, Cotton Caligula A XV. As outlined above, the provenance of this codex is highly disputed, since palaeographers have failed
to connect it to a specific scriptorium. Still, the rather vague localisation
of the origin of this manuscript in north-eastern France appears to have
found general acceptance.75 This provenance of the only manuscript
transmitting the Computus Cottonianus certainly agrees well with Willibrords area of activity on the Continent. Shortly before Willibrords
72
For the chapters of the Munich Computus that were copied from this now lost
Victorian computus of AD 689 cf. Warntjes (2010), CXXIVCXXVI.
73
This text, termed De comparatione epactarum Dionysii et Victorii according to
its content, is edited for the first time in Warntjes (2010), 3226 and discussed ibidem,
CLIICLVIII.
74
Warntjes (2010a), 93; idem (2010), LXXIILXXIII, CLXCLXI.
75
Cf. n 13 above.

The Computus Cottonianus of AD 689

mission a certain Wihtberht, also a monk from the circle of Ecgberht in


Ireland, had spent two years among the Frisians unsuccessfully preaching the Gospel with the permission of their king, Radbod.76 This experience apparently led to a change in tactics in Ecgberhts circle concerning
the Frisian mission. Willibrord, rather than directly approaching the
Frisian king, went to the Frankish major domus Pippin instead to seek
support for his activity.77 This support was graciously granted, with Pippin obviously hoping that the spread of Christianity in the newly conquered south-western part of Frisia would consolidate Frankish rule. In
fact, Pippins victory over Radbod in AD 689 may have been decisive
for the timing of Willibrords expedition; the mission itself, however,
as outlined above, appears to have been planned further in advance.78
Five years later, Willibrord went to Rome to receive the pallium with
the consent, if not under directions, of Pippin, who appears to have realized the potential of archbishoprics for the consolidation and structure
of secular power. The city chosen to be the see of the Frisian archbishop
was modern-day Utrecht.79 Here Willibrord spent most of his missionary life, retreating to his monastery Echternach (of which more in due
Bede, Historia ecclesiastica V 9 (Plummer (1896), i 298).
Bede, Historia ecclesiastica V 10 (Plummer (1896), i 299); Alcuin, Vita Willibrordi 5 (MGH SS rer. Merov. 7, 11921; Veyrard-Cosme (2003), 44, 46). Alcuins statement that Willibrord first went to Utrecht is unreliable, as Angenendt (1973), 108 n 279
and others have demonstrated; Fritze (1971), 1145, on the other hand, is inclined to
believe Alcuins account (as did Hauck (1887), 406 earlier), but Schferdiek (1994), 182
has further shown that Alcuin here employs literary models rather than outlining historical truth, bracketing Willibrords continental activity by references to Utrecht at the
beginning and end of his narrative of the saints mission (chapter 5 and 13 respectively);
for Alcuins further application of this technique and more generally for the structure of
his Vita Willibrordi see Reischmann (1989), 2642. For the connection between Willibrords mission and Pippins victory over the Frisians see especially ibidem, 1856, an
argument that can already be found in Rettberg (1848), 518 (cf. also n 69 above); the
fact that Willibrord approaching the Frankish major domus rather than the Frisian king
constitutes a major change in Anglo-Saxon missionary tactics is stressed by Angenendt
(1990), 178. Flaskamp (1929), 7 suggests that Wilfrith coordinated the mission and
therefore the agreement with Pippin from York. If Willibrord was of royal blood, as
Crinn (2007), 202 suggests, then this may have helped to secure Pippins support.
78
Cf. pp. 1934 above.
79
For the date of Willibrords consecration see the two notes on the November
page of his calendar (cf. n 62 above); the narrative in Bede, Historia ecclesiastica V 11
(Plummer (1896), i 3013); Alcuin, Vita Willibrordi 68, 13 (MGH SS rer. Merov. 7,
1213, 127; Veyrard-Cosme (2003), 46, 48, 54); Epistola Bonifatii 109 (MGH Epp. sel.
1, 235). For Pippins initiative and interest in Willibrords consecration see Wampach
(192930), i 389; Levison (1946), 59; Angenendt (1984), 1978 (who later, in (1989),
41, presents doubts about the Carolingian interest in archbishoprics); Fritze (1971),
1249 places more emphasis on the papal missionary zeal of the time, but still concludes
76
77

immo warntjes

course) at the end of his life, as well as during the periods when the political situation made it impossible to retain the archdiocese. His contact
with the Carolingian dynasty also remained very positive throughout
his lifetime, receiving dominions for his monastery Echternach as gifts
from Pippin II and his wife Plectrudis; the monastery was then in return
to become an Eigenkloster of the Carolingians.80 That the favourable
relations to this dynasty survived the death of Pippin and the ensuing
struggle for succession between Plectrudis and Charles Martell is illustrated by the facts that Charles Martell donated part of his inheritance
to Echternach, that Willibrord baptized Charles Martells son Pippin
III, that he called Charles dominus and senior in his so-called testament,
and that his calendar records many of the important events of Charless
life.81 These observations alone suggest that Willibrord was in constant
contact with the Carolingian majores domus, who had their main area of
activity and centre of power in the region between Paris and the Rhine,
and that he regularly lived in and travelled through the north-eastern
parts of Francia whenever he left Utrecht and southern Frisia. It does
not surprise, therefore, to find a text from the circle of Willibrord in
an eighth-century manuscript from north-eastern France; in fact, such a
provenance is exactly what would have been expected for the transmission of a text brought to the Continent by Willibrord.
The Reception
Due to the very nature of computistical formularies, it is extremely difficult to establish the reception of any such work. The general algorithms
that a Frankish reconquest of the Frisian area around Utrecht initiated Pippins sending
of Willibrord to Rome; similarly Schroeder (1985), 910.
80
For the close ties between Echternach and the Carolingians see especially Echternach Charters 1315, 24 (Wampach (192930), ii 3743, 5760); cf. also n 87 below. The term Carolingians here obviously also refers retrospectively to the ancestors of
Charles Martell, as Charles himself did not found a new dynasty but largely built on his
fathers successes.
81
For Charles Martells donation to Echternach see Echternach Charter 27 (Wampach (192930), ii 658); cf. also Echternach Charter 41 (Wampach (192930), ii 98
102). The baptism of Pippin at the hands of Willibrord is recorded in Alcuin, Vita Willibrordi 23 (MGH SS rer. Merov. 7, 133; Veyrard-Cosme (2003), 64); for this passage see
now Palmer (2009), 823. Willibrords testament is Echternach Charter 39 (Wampach
(192930), ii 8397; the terms in question on pp. 95, 97); a discussion of the titles given
to Charles Martell in this instance can be found in Angenendt (1973), 769. The events
of Charles Martells life noted in Willibrords calendar are thoroughly discussed in Levison (1938), 3436. For the succession crisis and Willibrords role in it see especially
Gerberding (1994), 2106 and now Fouracre (2000), 5778, here 624 on Willibrord.

The Computus Cottonianus of AD 689

outlined in the formulae usually remain unaltered, while the phrasing


may vary considerably. As long as the wording is not exactly the same, direct dependency between two formularies is difficult to prove and such
an analysis may, in fact, lead to very controversial results.82 Besides identical (not similar!) wording (including especially identical mistakes),
the only other criterion that proves dependency of one formulary from
another is corresponding dating clauses; e.g., the fact that the Computus
Digbaeanus of AD 675 preserves Dionysius original dating of AD 525
for all argumenta copied from his formulary proves that the author of
82
E.g., Cordoliani (1943), 60 and Springsfeld (2002), 73, 76 have referred to MSS
Vatican, Biblioteca Apostolica, Reg. Lat. 1260, starting 118r and Basel, Universittsbibliothek, F III 15k, starting 37r respectively as containing the original Dionysiac Argumenta; they have been deceived by the incipit, which is the same as the one in the printed
edition of Dionysius Argumenta (Incipiunt argumenta de titulis pascalis Aegyptiorum
investigata solertia; Krusch (1938), 75); had they analyzed the argumenta proper, however, they would have realized that these contain dating clauses for AD 788 and AD
789 respectively, which alone are proof enough that the formulae are later recensions of
Dionysius orginal composition; cf. Warntjes (2010a), 778. More problematic is Borsts
approach to the text Lect. comp. of AD 760, edited by him for the first time in (2006),
527659; he lists some 54 manuscripts for this text, divided into four recensions: ) the
original complete text, composed in the Rhine region from AD 760 onwards; ) a rearranged complete West-Frankish recension from c.AD 825 onwards; ) fragments from
AD 800 onwards in calendar manuscripts; ) fragments from AD 810 onwards in noncalendar manuscripts. Yet, even of the four manuscripts listed for the -recension, only
one (Cologne, Dombibliothek, 832; Ko in Borst) contains the texts as it is published; in
the other three, the text is not only arranged differently, but it is also not cohesive, as it
appears scattered throughout the MSS; additionally, some of the formulae of these MSS
listed as belonging to the original text show dating clauses which are markedly different
from the AD 760 date. In fact, at least in one instance a part of an -manuscript declared
by Borst to be a copy of a section of Lect. comp. constitutes a separate text in its own right,
namely f 90rv of the -MS Berlin, Staatsbibliothek, Phillipps 1831 (Bg in Borst); this
text is clearly a strictly defined, separate formulary datable to AD 757, a Vorlufer to the
AD 760-text (cf. Borst (2006), 5278, 532), and therefore deserves to be treated and
edited separately. The situation is even more complex concerning the - and -MSS. One
example may suffice: The excerpts listed under the -MS Cologne, Dombibliothek, 103
(fols. 184v185v and 189v190r; Borst (2006), 539) are, in fact, part of an independent text extending fols. 184v190v, the Computus Rhenanus of AD 775 to be discussed
below; a copy of this text can also be found in Borsts -MS Wolfenbttel, Herzog-August-Bibliothek, Weissenburg 91, 169r173v (note that Borst records the relevant formulae only for the Cologne MS, not the Wolfenbttel one, as I have mistakenly stated
in (2010a), 76: the argumenta of the Wolfenbttel version of the Computus Rhenanus
were used by Borst for the edition of Lib. ann. rather than Lect. comp.; for his edition of
Lect. comp. he used a different part of the Wolfenbttel MS); the formula providing this
dating clause of AD 775, a variation of Argumentum II, is edited as Lect. comp. IIII 2 in
Borst (2006), 5934, with AD 775 not being the only variation of date: others are AD
757 (from the Berlin text mentioned above), 809, 814, 818 (recte 813), 817, 826, 844,
885, 937, and all of these should be treated separately, as part of the collections or texts
in which they are contained.

immo warntjes

the Digby computus worked directly from Dionysius original, or rather


the extended version of the Dionysiac Argumenta that has survived to
the present day (the Group B corpus).83
As for the Computus Cottonianus, there is only one text, to my present
knowledge, which also incorporates dating clauses for AD 689, namely a
computistical formulary found in Cologne, Dombibliothek, 103, 184v
190v (= C) and Wolfenbttel, Herzog-August-Bibliothek, Weissenburg
91, 169r173v (= W). For ease of reference, this formulary, which has
escaped the attention of modern scholars, may be termed Computus
Rhenanus of AD 775.84 As the title suggests, the latest dating clause given in this text refers to AD 775, but it nevertheless also mentions the
earlier AD 689 three times.85 Now, all formulae which include the AD
689 dating in the Computus Rhenanus can also be found in the Computus Cottonianus, though the date given there is AD 688 rather than AD
689. This difference of one year should not lead us to the conclusion
that the author of the Computus Rhenanus worked from an entirely different text: The Computus Cottonianus also has the AD 689 dating later
in the text, and it is not unlikely that the examples in the formulae were
standardized to AD 689 in a subsequent copy of that formulary, and that
the author of the Computus Rhenanus then worked from this standardized version now lost. Indeed, a detailed analysis of the relevant passages
reveals that the Computus Rhenanus is dependent on the Cotton text. It
may suffice here to illustrate this dependency between the two texts by
only one example. When the different recensions of the Dionysiac Argumentum I as outlined in Appendix II are compared, it becomes immediately obvious that the author of the Computus Rhenanus worked directly
from an updated copy of the Computus Cottonianus: Not only do both
texts omit the same sentences when compared to earlier versions, they
also incorporate the same mistakes, e.g. the regular is wrongly given as
XV rather than XII, while indictiones is misspelled as dictiones in both.
As mentioned above, the Computus Rhenanus of AD 775 is the
only text, to my present knowledge, which evidently used the Computus
83
The Computus Digbaeanus is edited as Dionysius Argumenta by Jan (1718),
7994 and Krusch (1938), 7581. The Group B corpus of Dionysiac argumenta has not
been edited yet; in fact, a new edition of the original Dionysiac Argumenta will have to
be based on the Group B manuscripts, for which see Warntjes (2010a), 647. Cf. notes
17 and 18.
84
This formulary is first discussed in Warntjes (2010a), 756.
85
The dating clause for AD 775 can be found C 184v185r; W 169r (Argumentum II). For the AD 689 date cf. the passages listed in Appendix III below.

The Computus Cottonianus of AD 689

Cottonianus as one of its sources. Accordingly, the discussion of the reception of the Cotton text must solely be based on the Computus Rhenanus.
This latter formulary of AD 775 is transmitted in two manuscripts of different provenance, one from Cologne, the other from Worms (or, possibly, Weissenburg).86 The fact, then, that the reception of the Computus
Cottonianus can geographically be placed in the middle to lower Rhine
region strongly supports the argument that the Cotton text is to be ascribed to the circle of Willibrord. In AD 697 or 698, the abbess Irmina of
Oeren handed over part of the villa Echternach to Willibrord by counsel of the bishops of Trier; this estate was then further increased by dominions presented to the Anglo-Saxon missionary by Pippin II and his
wife Plectrudis, Irminas daughter, in the early eighth century.87 On this
estate, Willibrord founded the famous monastery of Echternach, which
he regularly used as a place of retreat, and to which he eventually retired
in the final years of his life.88 As far as can be reconstructed from the limited source material, Willibrords area of activity from this base appears
to have been mainly to the East and obviously especially to the North
towards the Frisian border, as well as to Echternachs immediate vicinities
(especially Trier): The connection to Oeren and Trier existed ever since
Irminas donation of the villa Echternach, and Alcuin relates in his Vita
Willibrordi that the saint visited her monastery at least on one occasion;
likewise, Alcuin mentions a trip to the Maastricht region, namely to the
86
Cf. the catalogue descriptions of these manuscripts listed in Warntjes (2010a),
75 n 110.
87
Echternach Charters 34 (Irminas donations), 1415 (Wampach (192930), ii
1723, 3843); for further donations by Irmina to Willibrord see Echternach Charters
6, 910, 12 (Wampach (192930), ii 2426, 3134, 367); for the implications of these
charters see Angenendt (1973), 6876; idem (1974), 2701; Semmler (1974), 3136,
3712; Anton (1989), 11521. For the archaeological evidence of the site before the
foundation of the monastery see Krier (1996), 46671. On Irmina, her family, and her
relation to Willibrord and Echternach see especially Wampach (192930), i 11335
and Werner (1982), 1175, 32631 (her relation and donations to Willibrord and Echternach 448, 6083; relation between Irmina and the Carolingians as well as Carolingian relation to Willibrord and Echternach 8490; Irminas relation to the bishops of
Trier 904; her family relations 12170; genealogical chart 331) as well as the thorough
critique of Werners study, especially of his doubt about the relation between Irmina and
Plectrudis, in Hlawitschka (1985).
88
For the hagiographical source evidence concerning the occasions and the time
spent by Willibrord in Echternach see Schroeder (1990); for his tomb in Echternach see
Krier (1996), 4768. A good summary of the evidence for Willibrords retreat to Echternach in the years AD 714719, the time of Frisian reconquest, as well as at the end of
his life is outlined in Rettberg (1848), 5212; an excellent overview over the geographical traces left by Willibrord is provided, in chronological order, in Wampach (192930),
i 5964.

immo warntjes

cella Susteren, which Pippin and Plectrudis had presented to Willibrord


in AD 714.89 Most of the territorial donations given to Echternach during
Willibrords lifetime recorded in the surviving charters of that monastery
refer to places to the north, principally the region between Schelde and
Rhine.90 Willibrords calendar, for its part, is another strong reminder of
the connection between Echternach and Trier, as it shows a great interest
in Triers saints and bishops, while some of the scribes who can tentatively
be identified among the hands that were at work in this calendar may
actually have come from the Trier region.91
Accordingly, Willibrords area of influence was principally between
Mosel, Schelde, and Rhine; the most important and profitable link to
any Frankish aristocratic family that Willibrord was able to draw on was
the one to Plectrudis and her mother Irmina, with Cologne being Plectrudis stronghold, while most of the estates of this family being located
further south in the Trier region.92 Thus, Worms and Cologne, the provenances of the two surviving manuscripts containing the only known
89
Alcuin, Vita Willibrordi 15, 21 (MGH SS rer. Merov. 7, 1289, 132; VeyrardCosme (2003), 56, 62, 64). For the donation of Susteren see Echternach Charter 24
(Wampach (192930), ii 5760) and Angenendt (1973), 701; Felten (1993), 189;
Dierkens (1996), 463 for its motivation; for the donation of a vineyard outside of Triers
city walls by Irmina to Willibrord and his monastery of Echternacht in AD 704 see
Echternach Charter 10 (Wampach (192930), ii 334).
90
Cf. the Echternach Charters 141 (Wampach (192930), ii 1102), their discussion in Wampach (192930), i 345404, as well as Karten II and III attached to
Wampachs first volume; a good account of the donations to Willibrords Echternach
and an analysis of the donors can be found in Theuws (1991), 31837 (particularly suggestive is Fig. 5 on p 332). For evidence of churches in the Low Countries at Willibrords
lifetime and the Echternach daughter foundations of this region see Rombaut (1990);
Pffgen and Ristow (1996b), especially Abb. 321 on p 409; Parsons (1999), 1368; Bijsterveld, Noomen and Thiessen (1999); good detailed accounts of Willibrords continental activity can be found in van Berkum (1989) and Weiler (1989), 11347; for the
historical context see now Palmer (2009), 1078.
91
The Trier bishops noted in the calendar are (Wilson (1918), 3, 5, 7, 9, 10, 14):
Valerius (29 Jan), Basinus (7 Mar), Maximius (29 resp. 31 May), Hildulfus (11 Jul), Paulinus (31 Aug), Eucharius (8 Dec); for the Trier scribes that may be identifiable in the
calendar see McKitterick (1990), 3845. For the relation of Willibrord and Echternach
to the Trier bishops and the noble families of the region (including Irminas and the
Carolingians) see Ewig (1954a), 13343.
92
For Cologne being Plectrudis stronghold in the succession crisis after Pippins
death see Liber Historiae Francorum 52 (MGH SS rer. Merov. 2, 326); Fredegarii chronicarum continuationes 10 (MGH SS rer. Merov. 2, 174); for Cologne becoming the centre of Carolingian power during Pippin IIs reign see Ewig (1953), 2245; idem (1965),
2978; Pffgen and Ristow (1996a), 159; both aspects are discussed in Semmler (1977),
5, 89, 33. For Irminas estates in the Trier region see especially the discussions in Werner
(1982), 7583.

The Computus Cottonianus of AD 689

text citing the Computus Cottonianus of AD 689 directly, fall exactly in


the area of Willibrords continental activity and of his greatest support,
lay as well as ecclesiastical. In fact, it may be suggested that the Computus Rhenanus itself may have originated in a computistical centre like
Trier before it spread further North and East to Cologne and Worms.
On the other hand, the date AD 775 may suggest that this text was compiled for the same purpose as its main source, the Computus Cottonianus,
namely to serve as a standard text for missionary activity; in AD 775,
however, the mission was to the Saxons rather than the Frisians, and if
this background to the Computus Rhenanus is accepted, then Cologne
rather than Trier or Worms would be the most likely candidate for its
composition, as newly conquered Westphalia became part of the Cologne archdioceses in that very year. Whatever the origin of the Computus Rhenanus may be, the history of its transmission certainly agrees very
well with the area one would expect for the reception of texts brought to
the Continent by Willibrords Frisian mission.

Summary
Summing up the results of this article: The analysis of the sources and
content of the Computus Cottonianus has led to the conclusion that the
origin of this work is best placed in the circle of Ecgberht in Ireland, to
which Willibrord belonged for twelve years, from the age of 20 to the
age of 32; one formula even shows direct links to Willibrords Easter
table. The date of composition, AD 688/9, agrees reasonably well with
Willibrords departure for the Continent, which, it must be presumed,
was envisaged for AD 689, but postponed to the following year due to
the political circumstances in Frisia. The transmission and reception of
the Cotton text fall exclusively into the area of Willibrords continental
activity and influence. These facts strongly suggest that the Computus
Cottonianus was compiled in Ecgberhts circle in Ireland in AD 688/9
for Willibrords Frisian mission.

immo warntjes

appendices
Appendix I: Facsimiles

Plate 1 London, British Library, Cotton Caligula A XV, 72v73r.

The Computus Cottonianus of AD 689

Plate 2 London, British Library, Cotton Caligula A XV, 77r.

immo warntjes

Plate 3 London, British Library, Cotton Caligula A XV, 80r.

Si nosse uis, quotus sit annus ab


incarnacione domini nostri Ihesu
Christi, conputa quindecies
XXXIIII, fiunt DX. Is semper
adde XII regulares, fiunt DXXII.
Adde et indiccionem anni cuius
uolueris, ut puta terciam
Consulato Probi Iunioribus, fiunt
simul anni
DXXV. Isti sunt anni ab incarnacione domini.

Si nosse uis, quotus annus ab


incarnatione domini nostri Ihesu
Christi, conputa quindecies
XXXIIII, fiunt DX. His semper
adde regulares XII, fiunt DXXII.
Adde et indictionem anni cuius
uolueris, ut puta tertiam
Consulato Probi Iunioris, anni
simul
fiunt DXXV. Isti sunt anni ab
incarnatione

Si uis scire, quotus annus est ab


incarnatione domini nostri, XLV
per quindecim conputa, fiunt
DCLXXV. Is semper adde
regulares XII, fiunt
DCLXXXVI<I>. Adde ad
indictionem annis cuius uolueris,
ut puta primam,

Si nosse uis, quotus annus est ab


incarnatione domini nostri Ihesu,
computa XVes XLV, fiunt
DCLXXV. His semper adde XII
regulares, fiunt DCLXXXVII. Adde
et indictionem anni cuius uolueris,
ut puta II,

Computus Rhenanus
Computus Cottonianus
(London, British Library, Cotton (Cologne, Dombibliothek,
103, 187v)
Caligula A XV, 73r)

fiunt DCLXXXVIII. Isti sunt


anni ab incarnatione domini
nostri Ihesu
Christi.

fiunt DCLXXXVIIII. Isti sunt anni


ab incarnatione domini
nostri Ihesu
Christi,
Christi.
hoc est DCLXXX et VIIII.
Hoc tantum memor esto sollicitus Hoc tanto memor esto sollicitus ut Hoc tantum memor esto sollicitus Hoc tantum memor esto sollicitus
ut succedentibus annis semper,
ut succidentibus annis semper,
succedentibus annis semper,
ut succedentibus annis semper,
quando ad quindecim indictionem quando ad XV indictionem
quando ad XV indiccionem
quando ad XV indictionem
perueneris,
ueneris,
ueneris,
ueneris, qua argumentum
integrum possit custodiri XV non
adsumas, sed ad sumam primam
unum semper adicias, ut puta

Computus Digbaeanus
(Oxford, Bodleian Library,
Digby 63, 72v73r)

Group B text of the Dionysiac


Argumenta (Vatican, Pal.
Lat. 1448, 13r)

Appendix II: Comparison of Argumentum I

The Computus Cottonianus of AD 689

si XXXIIII per XV conputabas,


XXXV per XV conputa
et annos domini sine ullo errore
poteris reperire.
Item cum ad aliam XV indictionem perueneris, XXXVI per
XV conputa.
Quibus adiectis semper XII
suprascriptis regularibus inplebis
numerum annorum domini et nihil
remanet, quia post XV addere
posis. Sicut in aliis indictionibus
remanent, ut in prima indictione
unus, in secunda II, in tertia III, uel
usque ad XIIII tot remanent de
conputo suprascripto quota fuerit
indictio. In XV uero indictione, si
ad summam superiorem, id est a
XXXV uel XXXXVI, non XV sed
unum addas, ut puta XXXVII per
XV conputes, et suprascriptos XII
addas, hoc argumento annos
domini inuenies.

Quibus additis semper XII


suprascriptis regularibus inplebis
numerum annorum domini et
nihil remanet, quod post XII
addere possis. Sicut in aliis
indictionibus remanent, ut in
prima indiccione I, secunda II, in
tercia III, uel usque ad XIIII tot
remanent de conpoto suprascripto
quota fuerit indictio. In XV uero
indiccione, si ad summam
superiorem, id est ad XXXV uel
ad XXXVI, non XV set unum
addas, ut puta XXXVII per XV
conputes, et suprascriptos XII
addas, hoc argumento annos
domini inuenies.

Item ad aliam XV indiccionem


XXXVI per XV conputa.

si XXXIIII per XV conputabas,


XXXV per XV conputa.

Quibus additis semper XII


suprascriptis regularibus inplebis
numerum annorum domini et
nihil remanet, quod post XII
addere possis. Sicut in aliis
<in>dictionibus remanet, et in
prima indictione I, in secunda II,
in tertia III, uel usque ad XIIII tot
remanent de compotu suprascripto quota fuerit indictio. In XVma
uero indictione, si ad summam
superiorem, id est ad XLVI uel ad
XLVII, non XV sed unum addas,
ut puta XLVII per XV conputes,
et suprascriptos XII (XV MS)
addas, hoc argumento annos
domini inuenies.

si XLV per XV conputabas, XLVI


per XV conputa.

Quibus additis semper XII


suprascriptis regularibus implebis
numerum annorum domini et nihil
remanet, quod post XII addere
possis. Sicut in aliis <in>dictionibus
remanet, ut in prima indictione I, in
secunda II, in tertia III, uel usque ad
XIIII tot remanent de computo
suprascripto quota fuerit indictio. In
XV uero indictione, si ad summam
superiorem, id est ad XLVI uel ad
XLVII, non XV se<d> unum addas,
ut puta XLVII per XV computes, et
suprascriptos XII (XV MS) addas,
hoc argumento annos domini
inuenies.

si XLV per XV (XII MS) computabas, XLVI per XV computa.

immo warntjes

74r

Argumentum II
Argumentum III 1
Argumentum IV 1
Argumentum V

2
3
73v 4
5

Computus Rhenanus (with AD 689; C


186v, W 170v)

Argumentum VIII

9 Calculation of the Julian calendar date and weekday of the


Argumentum XIV
Easter full moon from epacts of 22 March and concurrentes of
24 March (concurrentes of the examples: 4, 5, 6)

Computus Rhenanus (C 187v, W 171v)

Argumentum VII

Argumentum VI

Computus Rhenanus (with AD 689; C


187v, W 171rv)

Argumentum I

73r 1 Calculation of AD from indiction (AD 688)

Calculation of indiction from AD (AD 688)


Calculation of epacts of 22 March from AD (AD 688)
Calculation of concurrentes of 24 March from AD (AD 688)
Calculation of the cycle number in the cyclus decemnovenalis
from AD (AD 688)
6 Calculation of the cycle number in the cyclus lunaris from AD
(AD 688)
7 List of cycle numbers of the cyclus decemnovenalis in which
the Easter full moon falls in March
8 Calculation of bissextile year from AD (AD 688)

direct reception

no description

ultimate source

Appendix III: Inventary, Sources, and Reception


of the Computus Cottonianus of AD 689

The Computus Cottonianus of AD 689

76r

75v

74v
75r

Argumentum IX

Argumentum XII

Argumentum XI

15 Calculation of weekday of any given Julian calendar date from Argumentum X


concurrentes of 24 March
16 Calculation of lunar age of 1 January from cycle number of
Argumentum XIII
cyclus lunaris

14 Calculation of lunar age of Easter Sunday from epacts of 22


March

11 Calculation of epacts of 22 March from AD (AD 689; cf.


no3)
12 Calculation of weekday of 1 January from AD (AD 689)
13 Calculation of weekday on the Calends of each month from
concurrentes of 24 March and given regulars

10 Julian calendar limits for luna 1 and luna 14 of the Easter


lunation, as well as for Easter Sunday

Cf. Willibrords Easter table in Paris,


Bibliothque Nationale, Lat. 10837, 41r as
well as Munich, Bayerische Staatsbibliothek, Clm 14456, 66v.

Computus Rhenanus (C 186v187r; W


170v171r). Cf. BC 1112; Lib. ann.
3335, 37, cited in Lib comp. IIII 18, Lib.
calc. 55; only the beginning in RM 81.

immo warntjes

78r
78v

77v

77r

76v

22 Discussion of the eras used by Dionysius


23 Quattuor differentiae lunae

21 De flexibus digitorum

18 Calculation of lunar age of any given Julian calendar date,


apparently referring to the lunar year from luna 15 of Easter
lunation of AD 672 to Easter full moon of 673, as in the
previous argumentum: Easter Sunday 25 April, epact of 22
March (adiectiones lunae) 28; the algorithm remains obscure
to me.
19 Gloss to the following text: Explanation of the Greek term
enneacedecaeterida
20 Suggestio Bonifatii

17 Calculation of weekday of any given Julian calendar date,


apparently referring to the lunar year from luna 15 of Easter
lunation of AD 672 to Easter full moon of 673: indiction 1,
feria 6 on 1 January, concurrentes 5 (here termed soladictio),
Easter Sunday 25 April (24 April MS, but the 25 April of the
following argumentum appears to be correct here); the
algorithm remains obscure to me.

Computus Rhenanus (C 187r; W 171r).


Cf. BC 15; Lib. ann. 5859.

Computus Rhenanus (C 188r189r; W


172rv)

The Computus Cottonianus of AD 689

Divisions of a month
Divisions of a year
Discussion of lunar years
Discussion of the lengths of the seasons

30 Discussion of the etymologies of the seasons


31 (later hand): List of Victorian epacts of 1 January in numerical
order as well as those that do not occur in the Victorian
reckoning

26
27
28
29

24 Calculation of an annual bissextile increment of 3 hours by


division of the total number of hours of a year by 7
25 Calculation of the epacts of 22 March from the cycle number
of the cyclus decemnovenalis
Argumentum III 2

Argumentum XVI 2

Editions: Items 118 are ed. in Gomz Pallars (1994), 2231; item 10 ed. by Jones in CCSL 123C, 6801; item 19 ed. by Krusch (1926), 56; item 20 ed. by Krusch (1926), 558; item 21 ed. by Jones (1939), 1068 and in CCSL 123C, 6702.
Abbreviations are the ones introduced by Borst, Schriften, and the following: BC = Bobbio Computus; RM = Rabanus Maurus,
De computo.

80r

79v

79r

immo warntjes

You might also like