The manuscript: codicology and palaeography

The two surviving volumes are substantial and expertly presented, measuring 300mm x 240mm (written space 245mm x 200mm) with two columns usually of twenty-six lines, except for the handful of biblical readings which are written in long lines. The preparation and structure of the book, with ruling on the hair side on the outside of quires, is standard for Italian books from the early Middle Ages, though in this case the ruling appears to have been done two leaves at a time before folding. Prickings to guide the ruling are in the margin in the early eighth-century portions, but slits were used as guides for the late eighth-century folios. The gatherings of eight are signed with Roman numerals preceded by either a capital Q or half uncial q in the lower margin of the last page of the quire, underneath the right–hand column. Red as well as black uncial or capitals are used for many titles announcing the (supposed) author and the liturgical feast for which each sermon or extract from patristic exegesis is intended. Text incipits and explicits are also differentiated in red uncials or capitals. The pen-drawn initials in the early eighth-century section are finely drawn, sometimes using a compass to draw exact circles. The initials in the added late eighth-century sections are more elaborate and coloured. There is every indication in the use of abbreviations, citation marks, omission signs and punctuation, as well as in the character of the script, in both the early and late eighth-century portions, that this is the work of well-trained scribes, thoroughly accustomed to the general scribal conventions in use in early medieval Italy. The particularly monumental style of uncial used by the scribes of both the earlier and later eighth-century portions, with the distinctive character of the letter forms, have been identified by Armando Petrucci as late survivals of Roman uncial.Footnote ¹¹ The two surviving volumes of Agimund's Homiliary originally comprised 330 and 276 folios containing 116 and 100 texts respectively, and the first volume can be assumed to have been of similar length. The entire project therefore assembled approximately 300 texts on just short of 900 leaves (438 bifolia) divided into three more or less equal-sized volumes.

I mention these codicological details in order to emphasise that these two volumes are an expert example of book production. The original three-volume set represents a considerable investment in physical materials as well as in time and expertise. The books appear, moreover, to be a carefully-designed fair copy of the exemplar or sets of exemplars. Quite apart from the Roman uncial in which it is written, that it is a compilation produced for use in Rome is confirmed of course by Agimund's colophon.

Agimund and the ‘Roman Homiliary’

Any assessment of Agimund's achievement has to acknowledge that these two surviving manuscripts are a fair copy of something. But is it possible, contrary to current assumptions, that this ‘something’ was in fact Agimund's own draft, and thus his own compilation made from resources available in Rome in the early eighth century, as distinct from a reproduction of an earlier compilation? If the latter, how old might it have been? To ask what kind of resources Agimund deployed and how much Agimund himself contributed when making his selection has the potential to enhance our understanding of early medieval Roman liturgical and cultural life. My use of the shorthand ‘Agimund’ accepts the identification of Agimund the scribe as compiler for the moment.

To address the questions concerning the resources available in early medieval Rome raised by the contents of Agimund's Homiliary, they need first of all to be considered in relation to a small number of homiliaries understood to be Roman in origin and organised in relation to the Roman lectionary. These compilations form a related group:

1. 1. The Homiliary of Alan of Farfa (c. 769), extant in both a Frankish and a Bavarian recension, in manuscripts dating from the end of the eighth century and from the ninth century (for example, Bayerische Staatsbibliothek, Munich, Clm 4547 s.VIII/IX; Clm 4564 s.VIII/IX; Clm 18092 s.VIIIex; Clm 14368 s.IX; Clm 17194 s.IX) from Bavarian centres such as Benediktbeuren, St Emmeram in Regensburg and Tegernsee and others that remain unidentified.

2. 2. The Homiliary of Egino of Verona (bishop 780–99; †802 at Reichenau), extant in the late eighth-century codex also famous for its illustrations and known as the ‘Egino Codex’; originally from Verona, now in the Deutsche Staatsbibliothek, Berlin, Phillipps 1676. The homiliary section occupies fos 20v–23r, 29r–309v.Footnote ¹²

3. 3. The Homiliary in the BAV, Archivio di San Pietro C.105, dating from s.IX/X and adapted to take account of liturgical developments in the course of the ninth century. It contains a sequence of texts for the period from Advent to Maundy Thursday.Footnote ¹³

Rather than seeing these three compilations in a copying sequence, a forceful argument emerged in the course of the twentieth century that the considerable overlaps in text selection between these three – Alan of Farfa, Egino and the Homiliary of the Archivio di San Pietro – could be accounted for if they each independently had used an earlier Roman liturgical homiliary. Based on the evidence of the authors whose works are represented in these compilations, the most recent of whom is Isidore of Seville, this now lost ‘Roman archetype’ was surmised to have been produced between the middle and end of the seventh century and to have been originally composed for the basilica of St Peter in Rome. Jean-Paul Bouhot has argued, moreover, that the reconstruction of ‘L'Homéliare Romain’ by Réginald Grégoire was mistakenly based on a mixture of the two different recensions of Alan of Farfa.Footnote ¹⁴ Instead, Bouhot has proposed that the closest idea we can gain of what this lost hypothetical seventh-century homiliary contained is the Homiliary of Egino. Others maintain, however, that Egino simply adapted Alan of Farfa's Homiliary.Footnote ¹⁵

Consensus, either about the existence of a lost Roman archetype from the seventh century or how the similarities of content of these three homiliaries might be explained, has not been achieved. The complex transmission histories of the texts of the various authors of the sermons included in non-Roman (that is, Frankish or north Italian) manuscripts dating from the late eighth century onwards, the different origins of the copies of the compilations, and the varied scenarios that these interesting ideas suggest, all prevent this. Nevertheless, whether or not these three compilations are indeed related in the way that has been surmised, the extraordinary richness and diversity of the contents of these homiliaries for the reception of patristic texts in the early Middle Ages, as well as how much work still remains to be done, are clear. They need furthermore to be set within the context of the early development of homiliaries charted by François Dolbeau, from the single author collections of late antiquity to the early medieval compilations of texts by many authors.Footnote ¹⁶

The Roman homiliaries: the liturgical framework

As a further connection between these three homiliaries compiled by Alan of Farfa, Egino and the anonymous compiler of the Homiliary of the Archivio di San Pietro, Antoine Chavasse stressed that they are organised according to the liturgical sequence and readings as part of the Office in religious communities recommended in Ordo XIV for St Peter's basilica. The liturgical framework for Agimund's Homiliary, however, follows the (incomplete) sequence set out in the probably somewhat later Ordo XIIIA which is designated for the Roman church (and without reference to St Peter's), insofar as this Ordo XIIIA can be reconstructed from the extant Frankish manuscripts dating from the turn of the eighth century onwards.Footnote ¹⁷ As Peter Jeffrey has argued, Ordo XIIIA was a reform possibly designed to replace the arrangement outlined in Ordo XIV and marked a ‘shift of liturgical leadership’ away from the Vatican basilica towards the pope in his seat at the Lateran.Footnote ¹⁸ Jeffrey was inclined to date the more generally designated arrangement in Ordo XIIIA to the pontificate of Pope Zacharias (741–52). The earlier eighth-century date of the manuscripts of Agimund's Homiliary (BAV, Vat. lat. 3835, 3836), however, may indicate that the pontificate of Pope Gregory ii (715–31) is the more likely context for the introduction of Ordo XIIIA.Footnote ¹⁹ Gregory is particularly noted in the Liber pontificalis (in both redactions of his Life) for his work in relation to monastic communities and their observance of the liturgy, as well as for his encouragement of Boniface's missionary work in Germany.Footnote ²⁰ In this context, the use of this Ordo XIIIA framework is also a point in favour of the homiliary itself first being compiled in the early eighth century.

The stational liturgies also enhanced the growing emphasis on the pope's liturgical role.Footnote ²¹ The famous comes in Universitätsbibliothek, Würzburg, M.p.th.f.62 is usually invoked as a further early witness to both the scheme of biblical lections used in Rome and the organisation of the stational liturgy. The biblical lections as a map of the annual liturgical feasts of the temporal (the commemoration of the events of Christ's life and associated seasons such as Advent, Lent, Easter week and the period after Pentecost) and sanctoral (the feasts commemorating particular saints in calendar order during the liturgical year) themselves of course have a far longer history than that of the stational liturgy.Footnote ²² The Würzburg codex, written in a confident insular minuscule datable on palaeographical grounds to the later decades of the eighth century, records the principal Roman feasts and many of the stational churches for the readings throughout the liturgical year (fos 1r–2v), as well as the Epistle pericopes (fos 2v–10v) and an incomplete set of Gospel pericopes (fos 11v–16v). Each set includes incipits and explicits. The pericopes are organised in relation to the annual feasts of the temporal and sanctoral as well as ferial celebrations, for independent events such as ordinations, and readings from the Pauline Epistles probably intended for ordinary Sundays. The reference to the church of Santa Maria ad Martyres (the Pantheon consecrated to Christian ecclesiastical use in about 613), and the absence of particular feasts, such as the exaltation of the Cross and the Marian feasts for which litanies were promoted, according to his Vita in the Liber pontificalis, by Pope Sergius i (687–701), may indicate the consolidation of this set of lections between 613 and 687, that is, between the consecration of Santa Maria ad Martyres and the beginning of Pope Sergius’ pontificate.

Nevertheless, confidence in this Würzburg set of lections as a copy of a mid seventh-century compilation may be misplaced.Footnote ²³ As I have noted elsewhere, any estimate of the time lag between the initial compilation and earliest extant witness to a text remains guesswork. Thus, it is not clear whether the information or lists used by the scribe of the Würzburg comes may themselves have been up to date or not, what his source(s) may have been, nor how or in what context the copies may have been made.Footnote ²⁴ Possible scenarios may be the list being compiled by an insular-trained scribe in the entourage of one of the later eighth-century bishops of Würzburg known to have visited Rome (Burchard, Megingoz or Berowulf) while in Rome. Alternatively, it may be a copy made, at Würzburg, by an insular-trained scribe, of a stational and pericope list acquired in the course of the many exchanges between Rome and the Frankish kingdoms in the course of the eighth century,Footnote ²⁵ not least in relation to the Carolingian rulers’ well-attested interest in the promotion of Roman liturgy.Footnote ²⁶

The Roman list of Gospel pericopes and the stational churches in which they would be read, albeit probably from an independent source, also appears in Godes(s)calc's Lectionary prepared within the circle of scribes working for Charlemagne as a gift for Pope Hadrian in 781 (BNF n.a. lat. 1203).Footnote ²⁷ The intended recipient of this book may perhaps act as its validation as a record of the pericopes and stations in Rome at the end of the eighth century. Yet it is a well-known aspect of the manuscript transmission of every category of liturgical text in the early Middle Ages that ostensibly obsolete or superseded texts continued to be copied and circulated in the Frankish kingdoms and Italy alongside newer compilations well into the ninth, and even the tenth century.

It is against the backdrop of these more recent discussions of the Lectionary, the Ordines and the transmission of the sermons of the patristic authors included in the homiliaries that Agimund's contribution now needs to be assessed. All three of the homiliary compilations based on the St Peter's basilica lectionary framework (that is, the Homiliaries of Alan of Farfa, Egino and the Archivio di San Pietro C.105) postdate Agimund's work as far as both the earliest extant manuscripts are concerned, and the probable dates of their compilation. Agimund appears to have made independent use of texts supposed to have been part of the lost mid seventh-century Roman Homiliary, albeit to a far more limited extent than the other three later homiliary compilers. A closer look however, suggests other possible interpretations for the small number of texts in common.

The overlap between Agimund's texts and the ‘Roman Homiliary’

In BAV, Vat. lat. 3835, volume ii, there are twelve sermon extracts that were also included in the Homiliaries of Alan of Farfa and Egino of Verona. These comprise the sermons on Passiontide by Pope Leo the Great (De passione domini viii and x = Tractatus 60 and 61) and John Chrysostom (De proditione Iudae, Hom. 1, 1–6), a sermon credited to Augustine of Hippo but actually no. 12 of Eusebius gallicanus, Leo the Great's sermon x on the Ascension, sermons viii and x on Pentecost and sermon vi on the feast of the Apostles = Tractatus 74, 76, 77, 82, followed by three sermons attributed to Augustine for the feasts of the Apostles Peter and Paul = Maximus sermon 1, Pseudo-Augustine 205 and 189.

The overlap in BAV, Vat. lat. 3836, volume iii, at first sight seems more significant. It comprises forty-three sermons, extracted texts and examples of what Reginald Grégoire labelled as a ‘centon’, or assembly of shorter extracts. Many of these are credited to Augustine, Leo the Great and Pope Gregory the Great (the homilies on the Gospels), but others are left anonymous. Some of these sermons of Augustine have subsequently been identified as the work of Caesarius of Arles, Maximus of Turin and others. A few texts comprise extracts from larger works such as Isidore of Seville's Sententiae and Gregory the Great's Moralia in Iob. Although it has seemed significant that the texts Agimund has in common with the homiliaries of Alan of Farfa and Egino of Verona occur in the same order, this is also an order following the liturgical cycle of feasts during the year for which the original texts were designated, and is thus not so surprising. Separate author collections organised according to the liturgical year might also have served as resources.

Indeed, the existence from late antiquity onwards of compilations of sermons or exegetical homilies by single authors on liturgical lectiones seems clear. In BAV, Vat. lat. 3835, fo. 29r, for example, at the end of a group of ten of Leo i's sermons on the Passion, four of which were also used by Alan of Farfa and three by Egino, there is a note ‘Expliciunt sermones Sancti Leonis Papae de passione domini nostri Ihesu Christi numerum decem.’ These ten sermons of Leo the Great follow the liturgical order in which one of the two major collections of Leo's sermons were preserved. One of the earliest extant copies of this collection, Landesbibliothek, Karlsruhe, Aug. 227, is from southern Germany, was dated by Bischoff to the last third of the ninth century, and was the base text used by Antoine Chavasse for his edition of Leo's sermons half a century ago.Footnote ²⁸ Chavasse observed that this codex contains indications that it was copied from an exemplar written in uncial script.Footnote ²⁹ This raises the possibility that much earlier collections of Leo's sermons had been compiled, possibly even in Rome itself. Certainly, many other sermons of Leo are attested from ninth-century manuscripts in addition to their appearance in Agimund's Homiliary, and are cited by earlier authors, not least Pope Hadrian i (772–95).Footnote ³⁰

The works of Augustine, or homilies attributed at the time to Augustine, John Chrysostom, Gregory the Great and Isidore could also have been independently consulted. In other words, the overlap between Agimund's Homiliary and the trio of ‘Roman’ homiliaries is not an obstacle to regarding Agimund's compilation as the fruit of an independent assembly of material already familiar in liturgical contexts in Rome and available in Rome's libraries or churches. This surmise is strengthened if one considers the bulk of Agimund's Homiliary, whose contents are not replicated in any other compilation.

Agimund's resources

That Agimond drew on other collections of sermons or patristic works available in Rome is also suggested, for example by a number of notes included in the collection. In BAV, Vat. lat. 3835, as mentioned above, fo. 29r notes that this concludes a set of ten sermons by Pope Leo on the Passion of Christ. At the end of a small set of Augustine's sermons on fo. 276v in BAV, Vat. lat. 3856 is a reference to a larger collection from which they appear to have been extracted: ‘Expliciunt ser(mones) s(an)c(t)i aug(ustini) de solemnitatib(us) diversorum sanctorum numero centum.’ The use of already circulating small or larger collections of sermons by single authors (or credited to single authors), such as Augustine, Leo the Great and John Chrysostom in Latin versions, becomes clearer if we look briefly at a schematic summary of the contents of his two remaining volumes and their principal sources, set out so that you can see what is in the two volumes. (Authors and texts added in the late eighth-century sections are indicated in bold.)

Homiliary of Agimund: contents (as tabulated by Grégoire)

The disadvantage of setting out Agimund's sources like this is that it takes the works out of their liturgical context and does not reflect sufficiently the overall arrangement, the creative choice of topics and particular homilies; nevertheless it provides some sense of the resources on which the homiliary drew. As already noted, the list is based on Grégoire's identifications. There is particularly copious use of Augustine's sermons (or those credited to Augustine) and the precise source is difficult to determine. The transmission of Augustine's sermons after their first delivery in Carthage or Hippo appears to have been in the form of both formally supervised or ‘authorised’ redactions in the order in which they were delivered during a liturgical year, and informal, individual and ‘unauthorised’ records and collections, subsequently reassembled in a number of different contexts in Africa, Italy, Gaul and Spain in late antiquity and the early Middle Ages.Footnote ³¹ Extensive use was also made of Leo the Great's sermons. These too were likely to have been available in Rome, though, as noted above, the earliest extant manuscript was produced in south-west Germany in the ninth century. In the late eighth century, Leo's famous letter 28 was inserted, in which he sets out in full his understanding of the two natures of Christ in response to Eutyches. This letter was included in many early medieval canon law collections from the sixth century onwards and other later homiliaries as well as the Homiliary of Agimond.Footnote ³² Further, the compiler made a particular selection from Latin versions of groups of John Chrysostom's sermons, both those in the so-called collection of thirty-eight Latin homilies (not all of which are actually to be credited to John Chrysostom) attested in a number of ninth-century manuscripts, as well as others.Footnote ³³ On Vat. lat. 3835 fos 304v–305v, for example, the seventh reading in natale beati Pauli apostoli is a sermon by John Chrysostom which appears to be its earliest manifestation, and for which Reginald Grégoire provides the full text.Footnote ³⁴ Some of these Chrysostom sermons were also known in Rome from at least the fifth century; the sermon on the De ascensione domini, for example, was quoted by Pope Leo the Great.Footnote ³⁵ There is more limited use of Ambrose, Jerome, Maximus of TurinFootnote ³⁶ and Eusebius Gallicanus, though the transmission of their sermons is no less complex than those of Augustine and John Chrysostom.Footnote ³⁷ The biblical citations in Jerome's Vulgate translation from the Acts of the Apostles already mentioned are included for the Feasts of the two Apostles Peter and Paul. In Vat. lat. 3836 only there are extracts from Caesarius of Arles, the Rule of Benedict, Gregory the Great and Isidore of Seville, as well as from Peter Chrysologus in the eighth-century addition. Certainly the first three of these at least were known in Rome by the seventh century.

Agimund's own contributions?

The extensive compilation for the Feasts of the Apostles Philip and James, Peter and Paul, and the Apostles, may be one of Agimund's many unique contributions to his homiliary (fos 301–13) and indicate specifically Roman associations. Another may be the occasional instances of a medley of short texts. In volume ii, fos 309v–313r, although headed as by Hilary of Poitiers, is a ‘centon’ taken from various other commentaries. Marginal notes by the scribe of the main text (possibly trying to remedy an omission of the title in the main text) indicate that one extract is from Jerome's Commentary on Matthew and that another is by nili monachi (fos 311v–312r). Grégoire conjectured that this might be a Latin version of an otherwise unknown Greek text.Footnote ³⁸ All these texts are set out interspersed with twelve readings from the Acts of the Apostles. A further text described by Grégoire as a ‘centon’ has been established by Clemens Weidmann as the second recension of a sermon by Augustine on Pentecost (sermon 271) that appears in a number of liturgical collections, the earliest of which is Agimund's Homiliary.Footnote ³⁹ The source for this could have been a collection of Augustine's sermons arranged in liturgical sequence.Footnote ⁴⁰ Another instance discussed by Weidmann is what he describes as a ‘patchwork sermon’ de ligno crucis et de latrone for Holy Saturday in Agimund volume ii, Vat. lat. 3835, fos 89r–92r. The topics addressed are the wood of the cross, the virtue of wood (illustrated by Old Testament examples such as Noah's ark) and the good thief. Weidmann has established how Agimund appears to have selected the various extracts from homilies by ‘pseudo-Augustine’ and ‘pseudo-Chrysostom’, from the historical narrative attributed to Hegesippus, and the De errore profanorum religionum of Firmicus Maternus, but to have linked them together with explanatory words and phrases of his own in order to construct a coherent text for his readers.Footnote ⁴¹

I have mentioned explicits and headings which Agimund could have taken over from existing compilations of sermons, but there are also indications of the provision of his own navigational aids for the collection in addition to the headings or incipits and explicits for each selected text. These add to the impression of a compiler at work. The insertion of the full texts of the Lections from the Acts of the Apostles for the Apostles Peter and Paul is one obvious contribution in Vat. lat. 3835, fos 240r–243r, 255v–260v, 273r–276v and 295r–297v. Each set of three readings from Acts is followed by a set of homilies numbered consecutively so as to provide a complete sequence. After the first set of lections i, ii and iii for the vigil of St Peter, for example, there are, on the following fos 243r–255v, two sermons by or attributed to Augustine on the Apostles Peter and Paul numbered iii and v, sermons by Leo i (vi), Augustine (vii and viii), and an extract from Jerome's Commentary on Matthew (ix) which concludes ‘Explicit [sic] sermones in natale Sancti Petri Apostoli de prima vigilia.’ After the sequence of three lections (i–iii) for the vigil of St Paul, four Augustinian sermons follow (iv–vii) with a note after them: ‘alius sermo sancto Sancti Augustini scriptum est in secunda vigilia beati petri apostoli lectione iiii’. On fo. 313r, the reader is informed again: ‘alius sermo sancti augustini scriptum est in prima vigilia beati apostoli petri lectione vii’.

Other indications of the assembly of appropriate material are Vat. lat. 3835, fo. 51r, where the note ‘expliciunt sermones de v. feria passionis numerum v per lectiones viiii’ concludes the selection of five sermons credited to Augustine, John Chrysostom and Ambrose. At the end of the selection of ten sermons numbered i–x on the Ascension by Ambrose, Augustine and John Chrysostom, are a pair of Leo i's sermons, with the note: ‘expliciunt sermones domini leonis papae de ascensione domini nostri Ihesu Christi numero duo. deo gratias’. For Pentecost there is a similar assembly of ten sermons, seven attributed to Augustine, and three to Leo with the note on fo. 203r: ‘Expliciunt sermones Sancti Leonis almi pontificis urbis Romae de Pentecosten. numero tres.’

Agimund: scribe and compiler

I have only offered here an indicative sample of particular homilies, assemblies of extracts and the explanatory rubrics. A comprehensive analysis of all the texts in relation to both their specific readings and their transmission history would be desirable. It has also not been possible in the compass of this paper to make a full study of the interesting and varied choices of text, topic and feasts to which particular texts are attached in Vat. lat. 3836 in relation to their intended audience. These will have to wait for another occasion. The explanatory and linking phrases Agimund integrates into his assemblies of extracts or ‘centons’, such as those Weidmann has identified, also need to be investigated further. Certainly Jean-Paul Bouhot's detailed reconstruction of the transmission history of some of Agimund's possible sources has added substantially to Antoine Chavasse's earlier conclusions on the texts in Agimund's Homiliary, as well as offering important further indications of Roman liturgical creativity in the seventh and eighth centuries. The portion of Vat. lat. 3835, fos 240–316 containing the readings for the Feasts of Peter and Paul, for example, is also attested in two Frankish manuscripts of the ninth and tenth centuries

(Bibliothèque municipale, Orléans, 196 (173), fos 1–35, and BNF lat. 18297, fos 15–66v respectively) which appear to preserve these texts in a form that predates those found in Agimund. On the basis of the evidence of the texts relating to the Feast of the Maccabees, Bouhot has also posited the possible creation of two other homiliaries at San Pietro in Vincoli: an earlier seventh-century compilation that he thinks has left its traces in Vat. lat. 3828, and an early eighth-century compilation that was drawn on by Agimund.Footnote ⁴²

It cannot be proven that Agimund the scribe was also the compiler of this remarkable homiliary, even taking into account how rare colophons naming a scribe are from the early Middle Ages.Footnote ⁴³ Nor can it be certain that it was he who constructed the ‘centons’ and patchwork sermons and engaged so fully with texts as he made his selection with unfailing attention to the potential needs of his readers. Yet the characteristics of the compilation discussed above all suggest that he should indeed be given the credit for it, and credit too for devising such a collection, the earliest extant multi-author homiliary that assembled material probably already used in liturgical contexts in Rome. It is essentially the provision of a portable library, and perhaps was devised for a community lacking its own copies of the great variety of smaller collections of Augustine, Leo, John Chrysostom or the other patristic works on which Agimund's Homiliary draws.Footnote ⁴⁴ In some instances these coincide with those cited in papal sources from material presumably available at the Lateran, though it is unlikely that Agimund worked in only one library. Even the small sample of the choices considered in this article reflects a compiler intelligently and systematically gathering together local resources, many of which had probably been available in Rome for many decades, constructing his own sets of extracts, and even ‘patchwork homilies’ for particular occasions or to make particular exegetical points. The liturgical framework used is consistent with the palaeographical evidence that this is an early eighth-century compilation. Alternatively, for those who remain unwilling to accept Agimund as the compiler, it was Agimund who made a fair copy of a compilation by a close contemporary who remains anonymous. All discussions of the transmission of the various patristic authors and texts represented in Agimund's Homiliary are hampered by the fact that for so many of them the transmission history, in terms of extant manuscripts, starts with ninth-century manuscripts produced in Frankish scriptoria. In many instances (the precise number remains to be ascertained), Agimund's text is actually the earliest extant witness. Whether or not the homiliary is to be credited to Agimund himself, therefore, it appears that it can be understood as reflecting what was available in Rome by the early eighth century.

Vat. lat. 3856: the late eighth-century additions

I now turn to the late eighth-century additions to the third volume of Agimund's Homiliary, Vat. lat. 3836. They take the form, firstly, of two quires near the beginning of the volume (fos 55r–70v), written by two different scribes in a late eighth-century and rather less expert Roman uncial than Agimund's fine monumental uncial script. The scribes laid out the text per cola et commata (that is, in grammatical sense units which also accord with the rhythm of speech when reading out loud) with enlarged letters at the beginning of each new sentence, and noted citations with symbols in the margin. There are mostly ink-drawn initials at the beginning of each new homily in the quire containing fos 55r–62v; those in the following gathering are more elaborate, coloured and the bowl of the letter P on fo. 63r contains a portrait of a woman, presumably Mary.Footnote ⁴⁵ The parchment is inferior in quality to that used by Agimund. These leaves were inserted after quire vii which had ended pseudo-Augustine's homily on Susanna and the elders with a simple ‘Explicit deo gratias’ on fo. 54v.

The inserted texts relate to the Feast of the Assumption of Mary. The homilies comprise one attributed to Proclus of Constantinople on Marian theology, a homily attributed to Augustine on Luke's brief account of Mary and Martha (Luke x.34), and a Latin version of a sermon on Mary the Virgin by Antipater of Bostra, all assigned to the Feast of the Assumption (15 August). When the original work of Agimund resumes on fo. 71 with sermon 91 of Pope Leo the Great, usually assigned in mense septimo, only nine words from the beginning are missing. That quire is numbered quire viii at the end on fo. 78v. Thus, only one quire of the original Agimund volume is missing; the two late eighth-century quires therefore could either simply have been late eighth-century replacements of what was originally there, perhaps even made at Santi Apostoli, or were substitutes of texts from a different collection.

Vat. lat. 3836, fos 55r–70v: texts for the Assumption of Mary

The inserted folios between quires v and viii all relate to the Feast of the Assumption of Mary and comprise texts attributed to John Chrysostom, but in fact by Proclus of Constantinople, and Antipater of Bostra. They may have been part of this late eighth-century homiliary, or could have been a way in which the Homiliary of Agimund was augmented liturgically.Footnote ⁴⁶ The homilies of Proclus of Constantinople (434–46) at the beginning of the set of texts on the Assumption are rather rarer than the texts in the ‘Appendix’ (fos 277r–314v, discussed below), at least in a Latin version. Proclus played a prominent role in the Nestorian controversy. This example appears to be the earliest instance of a version in Latin of his famous first homily (wrongly identified as Homily v by Grégoire) on Mary as mother of God, preached a year before the Council of Ephesus and widely distributed in its Greek original. The earliest Greek text extant is in a manuscript dated to the eleventh century, though a short extract survives in a Greek florilegium dated to the late eighth century, now BAV, Vat. gr. 2200.Footnote ⁴⁷ The texts by Antipater of Bostra, a fifth-century critic of the theology of Origen, also suggest familiarity with, or access to, Greek texts or early Latin translations thereof.

Vat. lat. 3836: the added fos 277r–314v

The layout of the text and style of initials in the leaves we can term the appendix, now at the end of the volume, fos 277r–314v, is so different as to make it unlikely that fos 55r–70v were originally part of the same volume as fos 277r–314v. In other words, the set of texts occupying six further gatherings or quires in this appendix are the work of a different late eighth-century scribe from those of fos 55r–70v and should be regarded as part of a different homiliary entirely. Closer investigation of these leaves and the quiring structure and surviving quire marks indicates that they once formed the beginning of a different volume. The first, second, third, fifth and sixth quires survive. The quire marks i (fo. 284v), ii (fo. 292v) and v (fo. 306v) are clearly visible. Quire iv is missing. All save quire iii are gatherings of eight (four bifolia). Quire iii comprises six leaves (three bifolia) and the text finishes in the first column. The final leaves in the last quire, quire vi of this set, are mutilated. The contents of these folios comprise, first of all, extracts from five sermons by Petrus Chrysologus, archbishop of Ravenna, on the story of the prodigal son from Luke xv, and following these, with the transition in the middle of quire ii (fo. 290v) and the start of third week of Lent, are extracts from Augustine's in Iohannem Evangelium tractatus, cc. 15, 44 and 49, on John iv.1–42 and John xi.1–16. The subjects are Jesus’ conversation with the Samaritan woman which culminated in the Samaritans’ recognition of him as the Saviour of the world, and the account of the death of Lazarus.

Not only are these texts labelled in the manuscript as homilies for the second, third and fourth weeks of Lent but the papal stations are also provided: the second week is indicated as for the church of SS Marcellinus and Petrus (also the site of the Mausoleum of Helena), the third week for San Lorenzo in Lucina, and the fourth in the basilica of San Paolo, that is, San Paolo fuori le mura, and for Sant'Eusebio. These correspond to the stations LV (sabbato), LX (feria vi), LXV (feria iv), LXVII (feria vi) in the later eighth-century Comes of Würzburg (Universitätsbibliothek, M.p. th. fo. 62), and the Godes(s)calc Lectionary (BNF n.a lat. 1203), both of which are usually accepted as the earliest witnesses to the Roman lections in association with the stations.Footnote ⁴⁸

Script and paratexts together, therefore, confirm that this is a Roman compilation, but when exactly it was joined to the Agimund Homiliary is anyone's guess. I regard it as a lucky accident, for this section of BAV, Vat. lat. 3856 codex obviously raises many interesting questions, far more than can be pursued here. In relation to the availability of patristic texts in Rome, however, the texts need further comment.

The extracts from Augustine's Tractatus are among the earliest witnesses to a work of Augustine that has a particular diversity of manuscripts, partly due to what we are told about its process of piecemeal composition by Possidius.Footnote ⁴⁹ The extracts in both Agimund's compilation and the appendix are among the earliest witnesses to the Tractatus in general and to these particular selections, and it is clear that this was one of Augustine's texts known in Rome in the later seventh and the eighth century.Footnote ⁵⁰ The substantial corpus of homilies of Peter Chrysologus, archbishop of Ravenna (433–50), is best known in the so-called ‘Felician’ compilation made by Felix of Ravenna, one of Peter's successors, in the early eighth century. Two earlier collections were made, however, one of which is extant in a compilation made perhaps in Verona in the later sixth century (Biblioteca Ambrosiana, Milan, C. 77sup).Footnote ⁵¹ The other is thought to be represented by the small group of the first five sermons of Petrus in BAV, Vat. lat. 3836. They are unattributed to any author, contain many unique readings and omissions which distinguish them from the Ambrosiana ‘Severianus’ selection, but are also considered to be pre-Felician.Footnote ⁵² They would appear therefore to attest to the availability of a pre- or at least non-Felician selection of Petrus Chrysologus' sermons in Rome at the end of the eighth century.

Latin and Greek in Rome

The additions of a collection of texts on the Assumption suggest that the compiler was responding to the theological and doctrinal preoccupations of Rome in the later seventh and the eighth century. Further, the homiliary as a whole belongs in the context of the mixed Latin and Greek community we associate with Rome in the seventh and eighth centuries, the intense Christological discussions and the clear indication in the proceedings of the Lateran Council of 649 of how many texts were accessible in Rome.Footnote ⁵³

Indeed, the bilingual nature of the proceedings of the Lateran Council of 649, with the constant recourse to Latin and Greek and the Latin translations of Greek texts read out during the council sessions, is a striking confirmation of the dominance of both Latin and Greek as the principal languages of formal communication in the notably cosmopolitan and multilingual city of Rome in the early Middle Ages. Comment is often made on the way the Liber pontificalis draws attention to the ‘Greek’ or Syrian family origins of the popes, but it also makes clear how many of them had been trained in Rome, if not in the Lateran household itself.Footnote ⁵⁴ We are in a far stronger position to understand the presence and activity of Greek-speaking officials, immigrants and political and religious refugees (many from Palestine and Syria), often into the second and third generation, as a result of the classic study by Jean-Marie Sansterre on Greek monasteries in Rome,Footnote ⁵⁵ and the more recent work of Clemens Gantner, Vera von Falkenhausen, Maya Maskarinec, Philipp Winterhager, Stéphane Gioanni, Camille Gerzaguet, Filippo Ronconi and many others.Footnote ⁵⁶ All of these scholars have drawn attention to the cultivation of particular eastern saints’ cults, the introduction of elements of Eastern liturgical observance and the creation of new texts, especially saints’ Lives and Passiones, the availability of Latin translations of Greek patristic texts, as well as Greek translations of existing Latin texts in Rome, for which there was a Greek-reading audience.Footnote ⁵⁷

Zacharias's translation of the Dialogues of Pope Gregory i (BAV, Vat. gr. 1666) is a case in point.Footnote ⁵⁸ The painted inscriptions in Greek in the church of Santa Maria Antiqua need to be understood in this wider context.Footnote ⁵⁹ So do bilingual texts, still extant, from the sixth-century codex, Bodleian Library, Oxford, Laud gr. 35,Footnote ⁶⁰ and Latin and bilingual Latin and Greek glossaries from the eighth and the ninth centuries, though these still need further work.Footnote ⁶¹ A number of bilingual individuals as well as communities of Greek speakers in Rome have been identified, notably within the Lateran and among those holding the papal office itself. Philipp Winterhager, for example, has made a cogent case for crediting the production of the Lateran synod's texts to Lateran officials who were competent in both Latin and Greek, rather than to a notional group of Greek or Greek-speaking monks elsewhere in Rome.Footnote ⁶²

A short paper cannot do justice to the richness of the material in Agimund's two surviving volumes and the appended folios from another, hitherto unrecorded, Roman homiliary. The lack of the Christmas readings and the first five weeks of Lent in Agimund's compilation also deprives us of potentially useful comparative material. Agimund, the late eighth-century insertion of readings for the Feast of Assumption, and the late eighth-century Roman Homiliary fragment acting now as an ‘Appendix’ to Vat. lat. 3856, are also crucial witnesses to the possibility of collections of patristic sermons, exegesis and other theological works, and possibly florilegia of excerpts, in circulation in Rome. They demonstrate the additional value of homiliaries generally as a source of very early witnesses to and, therefore, knowledge and use of patristic texts which we otherwise only know from later manuscripts.Footnote ⁶³ Agimund's homiliary and the eighth-century additions, therefore, have also given us a glimpse of the rich resources in Rome in the later seventh and the eighth century in terms of texts, knowledgeable readers and liturgical creativity, and of the degree to which patristic theology and exegesis were embedded in Roman culture. There is a tantalising hint, with Latin versions of texts originally in Greek, of the interchange between the Latin- and Greek-speaking communities in Rome and the Lateran in the early Middle Ages and, above all, further evidence of the intellectual productivity and cultural versatility of early medieval Rome.