Texts identified and edited as ordines romani appear in various forms and in various contexts. In the definitions presented by Andrieu and Vogel, an ordo was conceived as a ‘pure’ representation of action and gesture, distinct from other elements of the narrative of liturgical ceremonial.Footnote ¹ For example, in Vogel’s influential handbook, they are described thus: ‘an Ordo is a description of a liturgical action, a directory or guide for the celebrant and his ministers, setting forth in detail the arrangement of the entire ritual procedure and how to carry it out. As descriptive of the actual rite, the ordines are the indispensable complement to the Sacramentary’.Footnote ² This counterpart, the Sacramentary or Mass Book, properly contained only prayers and spoken texts, and these were likewise supposed to be fully separable and extricable in their original conception. Yet it is rare to see any ordo of which one could say such a thing without qualification. In Andrieu’s works, this idea of purity of content is implicitly linked to the purity of Roman origin: that is, his reconstruction of ordines that were created and used in Rome, and represented wholly the Roman liturgy before Frankish copyists began to adulterate and alter them. Originally, such ordines had been copied individually or in small booklets whose aim was to propagate Roman usages; all surviving manuscripts represented the movement away from this. Again, we might quote Vogel: ‘Before they were gathered in Collections, each Ordo – which described a single actio liturgica or some part thereof – existed completely on its own … and were gathered for the first time only in Gaul.’Footnote ³ None of these hypothetical original settings of these texts has survived.

The idea that different books classified in the same genre might have different usages, or indeed that different components of the same book might have been read in different ways, and so help to interpret each other, was not integrated into this understanding of liturgical books, which employed a zero-sum understanding of this category of ‘liturgical’ (as being for ritual use). Such an understanding, however, is not evidenced in Carolingian attitudes to such books. For the Carolingians, texts we describe as ‘liturgical’ were comprehensible in various formats, and the ordines present this phenomenon particularly plainly. Carolingian manuscripts of the ordines romani raise questions concerning modern assumptions about the purpose and nature of the written liturgy, which, when addressed, allow a better understanding of the creative agency of manuscript creators and users, and ultimately the purposes of the ordines as copied and shared by them.

A great deal of background lay behind the assumed framings of Vogel and Andrieu. In the twentieth century, the history of liturgy was discussed in a certain way, with a certain background of assumed truths or ‘laws’ that still remain difficult to fully escape. One might, for example, quote Ellard’s formulation that: ‘the growth of liturgical ceremonial there operates first of all by that law of evolution where the rudimentary tends to perfection’ (i.e. the medieval liturgy is perceived as the evolutionary antecedent to modern practices).Footnote ⁴ Scholars like Andrieu attained a deep acquaintance with the texts and the meaning of liturgy that is difficult to emulate, and were most often liturgical practitioners themselves. But analysing liturgy as a historical source which provides some insight into the mindset of manuscript compilers requires that we sometimes take a different approach from theirs.

In the following treatment, manuscripts of the ‘collections’ of ordines will be examined and, where possible, links to known figures and locations used to help explore the use and reinterpretation of the texts which they contain. Critical and new in Andrieu’s study of the ordines were the discovery of Collections of the ordines romani that circulated in one or more manuscripts. In his editions, these were said to represent deliberate configurations of several ordines to a single purpose. Most widely witnessed were two to which Andrieu gave particular significance: Collection A or the ‘Roman Collection’, and Collection B or the ‘Frankish Collection’.

Within the framing of the medieval liturgy and how it was changed, as delineated above, Andrieu’s analysis assigned a reconstructed purpose and use to various manuscripts. The first of them, Collection A, he suggested, began life as the definitive attempt to establish purely Roman usages in Francia.Footnote ⁵ Only the purest Roman ordines were selected for this purpose. Other scholars wondered consequently if Collection A could have been the product of an official ‘reform’ effort by Pippin III, or even if it was a work of the famous Alcuin of York (735–804).Footnote ⁶ The compilation was dated by Andrieu in around 750, and he inclined to somewhere in modern-day France as the place of compilation. Drawing a specific analogy to the Gregorian Sacramentary that Charlemagne received around 785, he suggested that Frankish churchmen soon found the Collection too specialised in its presentation of specifically Roman liturgical usages. Later copies should rather be interpreted as ‘library copies’ for study outside a liturgical context, entirely different in their conception from the first few. Collection A is given by Andrieu as:

No manuscript evidence for any of these individual components is known to survive from before the closing decades of the eighth century. Surviving manuscripts also suggest a more gradual process of compilation, with no definitive form of the Collection, and considerable freedom to add additional content.

Significant redating of several manuscripts, as well as more precise localisation of them, permits us to give new context to the Collections and individual manuscripts. This is mostly thanks to the verdicts of Bernhard Bischoff. Along with his Katalog, I have consulted his Nachlass in the Bayerische, Staatsbibliothek, ANA 553, in order to better understand and evaluate the reasoning behind these often very important decisions. The earliest available manuscript of Collection A is now identifiable as BAV, MS Pal.lat.487, which actually contains only Ordines 1, 11, 27 and 34, lacking two of the texts, Ordines 13 A and 42.Footnote ¹³ It was written at Lorsch towards the end of the eighth century, and was often used as the ‘best’ manuscript for Andrieu’s editions because it preserves a text of Ordo 1 and the others quite close to what Andrieu had reconstructed as the Roman ‘original’. This suggests the importance of Lorsch as the node for the initial distribution of the Collection. The key to this importance is likely to be Lorsch’s close links with the bishopric of Metz and notably Chrodegang of Metz, bishop of Metz 742/748 to 766, who took a decisive part in the founding of the monastery in 764.Footnote ¹⁴ He provided Lorsch with the relics of Nazarius, a Roman martyr, given to him by Pope Paul I, which became central to the monastery’s liturgical identity.Footnote ¹⁵

Another clue that elements of Collection A perhaps came to Lorsch via Metz and which provides a possible connection with Chrodegang is the manuscript Bern, Burgerbibliothek, MS 289, specifically the parts fols.1–19, which can be dated around the same period as the Lorsch manuscript, at the end of the eighth century.Footnote ¹⁶ This fragmentary manuscript gives us a partial text of Ordo 1 (nn.36–64), the key text of Collection A, which also opens the Lorsch manuscript. The state of the text of this ordo is closest to that of BAV Pal.lat.487. Since the end of the text is lost, and we cannot say what followed, it is quite possible other ordines were originally present – perhaps four ordines from Collection A, just as in Lorsch? Certainly, directly proceeding Ordo 1 in these fragments we find also the Rule of Chrodegang himself for the canonical clergy of his cathedral. Since the manuscript ends before the end of Ordo 1, we cannot know if the manuscript originally contained the rest of Collection A, but perhaps this is not unlikely, since the version of the text otherwise can only be found with the other elements of that Collection. Can we then uncover Chrodegang’s role in the transmission of some early ordines, and link Lorsch’s reception and copying of an early form of Collection A to its connections to the bishopric of Metz?

Of similar age to the Lorsch manuscript are the various fragments of a manuscript probably copied in Murbach monastery and taken to Regensburg, which are found today in Munich, Regensburg and in Sankt-Paul in Lavanttal, of which Andrieu knew only the latter fragment. This book seems to have originally contained only the same four texts as the Lorsch example (pieces survive of characteristic elements of Ordo 34 and Ordo 27).Footnote ¹⁷ An abbot of Murbach from 789 was called Simpert, and a bishop of the same name is found in Regensburg from 768 to 791. It is likely that these were the same man, particularly since Charlemagne himself then took over as abbot of Murbach from 792, presumably following the death of Simpert.Footnote ¹⁸ Clear evidence of the transmission of manuscripts from the Alsatian monastery of Murbach to Regensburg would support an identity of the two, and help to explain our manuscript’s provenance.Footnote ¹⁹ Simpert, through his presence at the court of Charlemagne, was closely associated with the archchancellor and bishop of Metz, Angilram (bishop 777–791), successor to Chrodegang. Furthermore, Simpert of Regensburg went on campaign against the Avars in the company of Charlemagne and Angilram of Metz. Both Simpert and Angilram died during that campaign. Given that the Lorsch manuscript, the Bern fragments and also the Murbach fragments were all copied not in Chrodegang’s lifetime but towards the end of the eighth century, Angilram of Metz emerges as another contender for involvement in the initial transmission of Collection A, at least in its original form with just four constituent texts (Ordines 1, 11, 27 and 34), not the six reconstructed by Andrieu.

In addition to his links with Simpert of Murbach and Regensburg, Angilram also had close relations with Lorsch, and consecrated the church there.Footnote ²⁰ He was also interested in his predecessor’s Rule, a copy of which forms part of the Bern manuscript, which was likely made during his episcopacy, and the text was adapted with Angilram’s own additions.Footnote ²¹ There is also some evidence for a Metz connection, and one to Angilram specifically, in a late copy of Collection A, British Library, Add. MS 15222, produced in Besançon during the eleventh century, which undoubtedly goes back to a Metz archetype.Footnote ²² It contains the unique copy of an institute issued by Angilram recording how much he paid his clergy for various functions on high feasts, including a stational system for Lent in Metz described as ‘iuxta consuetudinem sedis apostolicae’.Footnote ²³ It is difficult to explain the text’s presence in this manuscript without assuming a Metz exemplar probably before Angilram’s own death, from which was copied (either directly or at one or more stages removed) the Besançon manuscript some centuries later. Since a restoration and extension of canonical life was being undertaken in Besançon at this time under Archbishop Hugh of Salins (bishop 1031–1066), and with particular attention to liturgy, a memory of Carolingian Metz’s role in the institution of common life, and of Chrodegang and Angilram as founders, probably helps to explain why the Collection was copied there.Footnote ²⁴ A connection between Angilram of Metz, canonical life in a cathedral city, Roman stational usages, and Collection A of the ordines romani seems to still have been alive in Besançon in the eleventh century.

The complete Collection A, as Andrieu reconstructed it, had two additional texts however, Ordines 13 A and 42, which deal with the Matins biblical readings of the liturgical year and with church dedication respectively. Andrieu’s own ‘best’ example of this complete Collection and the first to carry all six ordines, Montpellier, Bibliothèque de la Faculté de Médecine, MS 412, is from the first quarter of the ninth century, and was written at Tours.Footnote ²⁵ The two additional texts which were added to the Collection in this example, Ordo 13A at the opening of the Collection and Ordo 42 at the closing of it, are notably also witnessed in manuscripts that are independent of the same Collection, indicating differing provenance. Ordo 42 can also be found in Albi, Bibliothèque Municipale, MS 42, which has some connection to Reims.Footnote ²⁶ Though Albi 42 also included Ordo 1 (fols.60r–68r), it has a form of that text that has some independence from Collection A, perhaps representing an earlier version.Footnote ²⁷ It also contains Ordines 24 and 26, the principal sources for Collection A’s Ordo 27 but not that text itself. Despite its late date, Albi 42 therefore gives the impression of being a copy of an older archetype, a possibility that we will subsequently explore. But in this case, it indicates availability of Ordo 42 in France independently of Collection A.

Ordo 13 A appears in a number of other manuscripts besides those containing Collection A, of which ninth- and tenth-century examples include: Douai, BM, MS 14; Vercelli, Archivio Capitolare, MS 183; St Gallen, Stiftsbibliothek, MS 225; Reims, BM, MS 1; Munich, BSB, Clm 6398 and 14470; Rouen, BM, MS 26; and BAV, MS Vat.lat.6018.Footnote ²⁸ In some of these cases, Ordo 13 A was added on spare folios of manuscripts of various kinds.Footnote ²⁹ The St Gallen and Vercelli examples are the oldest, and date to the end of the eighth century, which demonstrates that this ordo already existed before its incorporation in the Collection.Footnote ³⁰ They both include the ordo among patristic extracts, which suggests an understanding of the text as an authoritative one among the ‘Fathers’.Footnote ³¹ However Reims 1, Douai 14 and BAV Vat.lat.6018 are the only three of these manuscripts to include a particular textual addition (Ordo 13A, nn.3–5) that is also found in manuscripts of Collection A which contain that Ordo, such as Montpellier 412. Douai 14 is closest to Collection A in other details.Footnote ³² Two of these manuscripts come from Reims. Both are biblical: Reims 1 is part of the Bible presented by Archbishop Hincmar (845–822) to the Cathedral, and Douai 14 is an Old Testament manuscript, located to Reims by Bischoff.Footnote ³³ Therefore, the circulation of Ordo 13A, like Ordo 42, in Northern France can be securely established, as well as its presence in numerous centres elsewhere. When combined with the evidence of Ordo 42’s presence in Northern France, it was probably in this region that Collection A, having come from Metz, was enhanced with these two additional ordines.

Our earliest known copy of the full Collection is Montpellier, Faculté de Medecine, MS 412, copied at the abbey of St Martin, Tours, in the Loire valley. The only other ninth-century manuscript with the same full collection is Copenhagen, Kongelige Bibliotek, MS Gl. kgl. 3443, dating to the boundary between the ninth and tenth century, locatable to France.Footnote ³⁴ This rather late manuscript is, in fact, the only one which carries Andrieu’s Collection A apparently without the addition of any other ordines (leaves are now missing, the final item ending partway through Ordo 13A). Such evidence is not at all conducive to regarding the accomplishment of Collection A as already completed by the year 750, as Andrieu did.Footnote ³⁵ We should probably therefore trace Collection A’s creation on the basis of the surviving manuscripts, uncovering an initial collection of four distinct ordines towards the end of the eighth century in Metz, and an enhancement of this with two additional ordines around the beginning of the ninth, possibly in Northern France. The second, enhanced version also came to form the source for Collection B, probably redacted in Alemannia or Northern Italy, as we will see.

So, we come to the key question: did the compilers and copyists of the Collection A see themselves as copying a ‘Roman collection’, and were they selecting specifically and purely Roman ordines for the role, as Andrieu assumed? In the original collection, as transmitted in the Lorsch and Murbach manuscripts, which I attributed to Angilram of Metz, none of the four ordines were actually composed by Romans.

Ordo 1 has long been held to be a Roman document, but no knowledge of the text can be demonstrated in Rome. The assertion of Roman origin has tended to go together with the assumption of an early date for the text, around 750. This latter assumption is perilous, since the text cannot be evidenced in any way so early. However, even those who maintain Roman origin for the ‘long recension’ of Ordo 1 concede the ‘short recension’, found in one manuscript, to be Frankish.Footnote ³⁶ But if we accept Andrieu’s argument that the ‘long recension’ is based on the ‘short’, as I do, and that the ‘short recension’ was known in Francia and used before the ‘long recension’ was known there, the Frankish origin of the latter can likely also be assumed.Footnote ³⁷ The precise origin of this text, however, is less important for this study, which is focused on the purpose or purposes for which manuscript copies were made. However, in Chapter 4, I will present additional evidence against Andrieu’s belief that Ordo 1 constitutes an accurate representation of Roman liturgical practice, namely a blatant error in the discussion of the immixtio in the text, for which the ‘short recension’ presents a more accurate description of Roman practice than the ‘long’. The more it becomes clear that the Romans did not record texts such as Ordo 1, and that the ordines romani are Frankish documents, the more the evidence is weighed to the probability that Ordo 1 was a Frankish document too. To this effect, I will also argue below that Ordo 11, the baptismal text and another key element of Collection A, was not a Roman text at all, but was almost entirely written by Franks to begin to fill in very sparse rubrics in the Sacramentary.Footnote ³⁸

Of the four texts in the Lorsch manuscript, Ordo 34 has the strongest claim to be a Roman document, but this too becomes more problematic upon closer examination. This text describes the ordinations of the clerical orders from acolyte to subdeacon. The terminology and procedures are accurate for practice at Rome: all are undertaken by the domnus apostolicus (the Pope) and the episcopal ordination takes place in the ‘aulam beati Petri apostoli’ (Ordo 34, n.18). Nevertheless, in one key intervention, the text includes the note (Ordo 34 n.16) that the women called ancilla dei sacrata in Rome are called nuns by the Franks (‘quae a Francis nonnata dicitur’).Footnote ³⁹ This indicates that a Frankish audience was also in mind. This part of the ordo also cites, it seems, the true Roman wording of the interrogation quattuor capitulis secundum canones (four things the bishop must not have done), but also interprets the complex vocabulary and Greek terms which the Romans seem to have used for them, again it would seem for a non-Roman (and presumably Frankish) audience.Footnote ⁴⁰

Andrieu assumed that the text had been redacted by a cleric in the Lateran Basilica in Rome, but the intervention directly addressed to Franks makes it more likely that a foreign cleric present in Rome, who observed ordinations, carefully recorded the text for a Frankish audience, with much the same care as we see in Ordo 1. This suggestion should not be dismissed simply because, as Andrieu has put it, the ordination rites were not directly applicable in the Frankish realms, for it is evident that the Franks were interested in the rites they themselves could not do (as we shall see in the Frankish manuscript witnesses to Ordines 1 and 26 below).Footnote ⁴¹ Andrieu also adduced as evidence in support of his argument that Pope Hadrian I had cited ‘toute au long notre document’ (‘quoted our document throughout’), in a letter to Charlemagne of 790–791.Footnote ⁴² However, although Hadrian’s description accords in general with Ordo 34 and secures its general accuracy in describing Roman practice, he did not directly cite the text at all, nor indicate its availability in Rome. He spoke only in very general terms of his interrogation of the newly elected bishop, without citing any of the details of the ritual surrounding it which are described in Ordo 34.Footnote ⁴³ This corresponds generally with the interrogation by the Pope on the quattuor capitula described by Ordo 34, n.16.Footnote ⁴⁴ In fact, the only one of the questions to the bishop that Hadrian quoted directly (perhaps from memory) – ‘Vide ne aliquam promissionem cuiquem aut dationem fecisses, quia simonicum et contra canones est?’ – is not found in Ordo 34’s narrative of the interrogation.Footnote ⁴⁵ Hadrian’s citation supports in a broad way the general accuracy of Ordo 34, but not the availability of the text itself to him, and thus its assumed redaction in the Lateran. Ordo 34 could thus also most likely be a work of a Frankish pilgrim who had observed papal ceremony closely.Footnote ⁴⁶

The next ordo in the collection, Ordo 27, is the product of the merging and adaptation of the pre-existing Ordo 24 and Ordo 26, themselves found in three Frankish manuscripts important for the pre-history of the ordines prior to Collection A: Albi 42 already mentioned, also Brussels, Bibliothèque royale, MS 10127–10144 and St Gallen, Stiftsbibliothek, MS 614. Ordo 24 deals with the Day Office of Holy Week, and Ordo 26 with the Night Office. The Brussels manuscript is the oldest of these, and secures the existence of these two sources for Ordo 27 probably before the end of the eighth century.Footnote ⁴⁷ But, as we will see, the other two manuscripts were probably copied from exemplars that were just as old. Ordo 24 describes the Day Office of Holy Week, from the Wednesday before Maundy Thursday through to Easter, while Ordo 26 describes the Night Office. But, unlike other ordines (and just like Ordo 11), the main narrative of Ordo 24 evidently does not take place in Rome. The celebrant is a pontifex, and he is clearly distinct from the Pope: at Ordo 24, n.28, the pontiff prays for the ‘apostolicus’ (the Pope) on Good Friday, whereas the priests name their bishop.Footnote ⁴⁸ The ceremonies unfold in various churches, but no Roman names were given to them: for example in Ordo 24, n.22: ‘in ecclesia statuta infra urbem, non tamen in maiore ecclesia’.Footnote ⁴⁹ None of Rome’s exotic personnel appear, only notarii and the secundus and magister scolae (there is no equivalent to this latter position in Rome in Ordo 1). Finally, the fact that everyone communicates on Good Friday (Ordo 24, n.38: ‘Et communicant omnes cum silentio’) directly contradicts what Amalarius of Metz learned from the Roman clergy concerning the papal celebration in Rome, where no one communicated on that day at all.Footnote ⁵⁰

In contrast, Ordo 26 does describe a specifically Roman, papal usage, but presents it as differentiated from the normal usages and rather exotic. Rome was evidently viewed as foreign to the presumed reader: for example, in the first part, the Sunday before Holy Week is that ‘which the apostolic see calls mediana’.Footnote ⁵¹ During the Easter Vigil, the text details that other cities beyond Rome (‘forensibus civitatibus’) undertook the blessing of the Easter candle as usual (the blessing is not described, but a pointer to the Sacramentary is given), but that ‘the catholic church within the city of Rome’ did not do this.Footnote ⁵² Instead, the archdeacon of the papal church made the agnus dei from blessed wax to be distributed to the people on the Octave of Easter.Footnote ⁵³ This description conveys the flavour of one of the ‘travel records’; it was, according to the very wording of the text, not a practice to be imitated, but one to be read and wondered about.

Three mentions of the use of a Sacramentary in the two ordines (a Sacramentary which was specifically not the Roman Gregorian but rather the Gelasian) – Ordo 24, n.2: ‘in Sacramentorum continetur’; Ordo 24, n.23: ‘ordine quod in Sacramentorum continetur’; Ordo 26, n.5: ‘ordine quod in Sacramentorum continetur’; and the specific note that a prayer ‘pro rege Francorum’ (Ordo 24, n.3) would be said on the Wednesday of Holy Week – would incline one to believe the two texts were describing a particular Frankish ceremony of Holy Week, with the interlude of the Roman practice of the Agnus Dei on Holy Saturday in Ordo 26 as a clearly indicated digression. Was the setting, in fact, Metz? Andrieu admitted the possibility, at least, that Ordines 24 and 26 had been redacted by Frankish observers of the Roman liturgy, but still preferred to argue for a Roman origin by proposing that these were created for the churches ‘suffragen’ to Rome, still partaking of the Roman liturgy, and that they described truly papal ceremonies.Footnote ⁵⁴ However, this statement relies on Chavasse’s theory that the Old Gelasian Sacramentary was originally compiled for use in the liturgies celebrated by a priest in one of Rome’s titular churches, thus ingeniously explaining divergences from what we know of papal practice by hypothesising a distinctive ‘presbyterial’ Rite in Rome.Footnote ⁵⁵ These theories have been strongly criticised, and it does not seem that the Old Gelasian can really be located to the titular churches of Rome at all, or seen as a presbyterial counterpart to the Gregorian’s papal liturgy.Footnote ⁵⁶ Further discussion on the origin of Ordo 24 and Ordo 26 as part of a set of ordines that can be specifically linked to Chrodegang of Metz, and which includes the ‘short recension’ of Ordo 1 (as witnessed, for example, by St Gallen 614), will be detailed below. For now, we should note that the Holy Week ordo in Collection A, Ordo 27, combines two ordines that were not clearly of Roman manufacture, and made little claim to be describing Roman usages at all. Their combination was not done with much finesse.Footnote ⁵⁷ Very little was added, except a formula for chrismation, which was specifically by an episcopus.Footnote ⁵⁸ This combination of the ordines, at least, clearly took place within the Frankish kingdoms, as even Andrieu acknowledged.Footnote ⁵⁹

There was a third source for Ordo 27. Ordo 27, nn.67–94 was a different kind of text, a description of paschal Vespers for the week after Easter (entitled ‘Ad vesperas die pasche sanctum’) which originated in neither Ordo 24 nor 26. This was of immediately different character to the preceding texts. Roman personnel like the parafonistis infantibus appear at Ordo 27, n.70 and the notario vicedomni at Ordo 27, n.28. Quite particular Roman placenames are also a feature: at Ordo 27, n.76 ‘Ad sanctum Iohannem ad Vestem’, or Ordo 27, n.77 ‘ad sanctum Andream ad Crucem’, both oratories in the Lateran Baptistery, as well as the Lateran itself (Ordo 27, n.80, n.83).Footnote ⁶⁰ A number of the chants are even delivered in Greek (singing in Greek took place within the liturgy in Rome, and was imitated in Frankish cities, as at Metz).Footnote ⁶¹ Andrieu was clear, this part, at least, was a ‘true ordo romanus’, that is, one that was redacted in Rome and able to tell us how the paschal Vespers really unfolded in Rome.Footnote ⁶² However, Van Dijk demonstrated plainly that the text in Ordo 27 is not, as Andrieu said, the accurate record of Rome’s own ceremonies on these days.Footnote ⁶³ In fact these texts are a clumsy Frankish adjustment of a more accurate Roman original text that survived elsewhere, given in the important ordo romanus manuscript, Wolfenbüttel, Herzog August Bibliothek, MS Weissenburgenses 91 (examined below) but perhaps originally transmitted in the Antiphoner, since pieces resembling it do certainly survive in the Antiphoner of Compiègne, a Carolingian copy of a Roman chant book.Footnote ⁶⁴ The fact that this piece could be taken by Franks from a Roman chant book adds to the range of sources of ordines and undermines Andrieu’s conception of the texts as wholesale independent inventions in Rome. Rather, the forms the ordines take in Collection A are in many cases Frankish reinterpretations, constituting a new format invented principally for Frankish consumption. Before it was copied into Collection A, this text of the Easter Vespers had been clumsily adjusted in order to make space for a stational service on the Thursday of the week after Easter (nn.78–91). Having a stational service on the Thursday was a more recent innovation in Rome which the Roman Sacramentaries available to the Franks displayed but which the original text, representing an earlier stage of the Roman liturgy, did not.Footnote ⁶⁵ As those like him were accustomed to do, the Frankish author ‘borrowed’ texts in order to fill out a stational ceremony which they knew occurred in Rome.

If we suggest a role for Angilram of Metz in the transmission of this initial Collection A, it is not at all clear that Angilram had intended to gather a purely ‘Roman Collection’. Ordines 11 and 27 actually described, for the most part, Frankish usages and could be replicated in a Frankish cathedral like Metz without any difficulty. Ordo 1 and 34 described Roman usages, but both were observations by Frankish observers, and not truly Roman texts, in the sense Andrieu had conceived. A dynamic contrast between Roman and Frankish liturgies was thus a part of the ordo romanus tradition from the earliest gathering of the texts. The reader was clearly supposed to read this collection with a mind to what he might find useful and what he could or should not.

The final addition of Ordo 13A and 34 to a second enhancement of Angilram’s original collection did strengthen the ties of the Collection to Rome. Ordo 13 A is perhaps the only one of the ordines that can uncomplicatedly be attributed to Rome itself, and it very likely accurately describes the practice of the clergy of St Peter’s, probably only within or after the reign of Pope Zacharias (741–752).Footnote ⁶⁶ The Roman text is notably not concerned at all with the things Franks were interested in, in the ordines we know they wrote, which is to say topography, movement, gestures and arrangement of clergy. It is simply a list of readings. Ordo 42 is also entitled specifically as a Roman usage: ‘ln nomine Dei summi ordo quomodo in sancta romana ecclesiae reliquiae conduntur.’Footnote ⁶⁷ Here the church was dedicated simply by the deposition of relics. In this case, there is nothing that would argue against Roman origin: the prayers and antiphons listed as part of the ceremony all belong to the Roman Gregorian Sacramentary and the Roman Antiphoner of Compiègne.Footnote ⁶⁸

The impression gained of a complex relation to Roman practice is strengthened when we examine anew the manuscript copies of Collection A. None of the manuscripts show plainly that the copyists of the collection prized only the Roman ordines. Instead, the three earliest manuscripts each carry an additional ordo from outside what Andrieu defined as his Collection A, all of Frankish manufacture. In the case of BAV Pal.lat.487 and the Murbach/Regensburg fragments, a second, alternative Holy Week ordo was added to the manuscript not long after it was copied.

In Lorsch, the manuscript of what was probably Angilram’s collection (BAV Pal.lat.487) had an extra folio at the end, and this was used, a few decades after the manuscript had originally been written, to begin a new text, which was completed with the addition of a new quire (fols.25r–30v).Footnote ⁶⁹ This text is a reworking of Ordo 27 as a new text, edited by Andrieu as Ordo 29. In substance, this is a repetition of many of the same rituals but it now takes place in a Frankish monastery rather than an urban setting.Footnote ⁷⁰ A simple ‘presbiter’ undertakes the prayers on Wednesday and Friday.Footnote ⁷¹ We can tell that the Lorsch reworker of the text wished to emphasise two particular aspects. One was the Eucharistic customs on Maundy Thursday (Ordo 29, nn.24–26):

The other was the gradual extinguishing of lights on each of the days, or the Tenebrae ceremony (Ordo 29, n.12).Footnote ⁷³ Both these aspects were highly symbolic and theatrical, and demonstrate what are clearly non-Roman embellishments. The Vespers portion of Ordo 27 (with the Roman place names and characters) is also entirely absent. On the one hand, then, we have a clear reworking of Ordo 27 to make it less Roman and more reflective of Frankish customs and peculiarities. This was, in general, how we might imagine Holy Week really unfolded in the monastery of Lorsch itself. At the end of the text, however, is an extract from what appears to be a longer letter, in which a person reports that they consulted Pope Hadrian himself about some specifics of the Holy Week Ceremony, specifically aspects on which Ordo 27 was unclear or silent:

The two answers of Pope Hadrian about Holy Week were actually incorporated into the preceding Ordo 29.Footnote ⁷⁵ Bischoff’s dating of the manuscript means that the Pope consulted must have been Pope Hadrian I (d.795), and not Hadrian II (867–872) or III (884–885), as Andrieu had assumed.Footnote ⁷⁶ Ordo 29 shows the same complex adaptations of the ordo to diverse Frankish sensibilities. On the one hand, they pruned back Roman details and added new elaborations to the ceremony, which probably had a long history in Francia. On the other hand, the Pope was still an ultimate arbiter and judge in matters of liturgical confusion. The question of who wrote this letter remains unanswered. Presumably it was a Frank who travelled to Rome and who was a cleric of sufficient standing to be admitted to Pope Hadrian’s presence, but on the other hand, they were clearly tasked to ask about these liturgical matters by a superior (their abbot or Bishop Angilram of Metz?). The addition of Ordo 29 was accompanied in the Lorsch manuscript by a theological extract from Alcuin of York’s De fide Trinitate, the summary ‘profession of faith’ entitled Credimus sanctae trinitatis which would later be recited in a liturgical setting.Footnote ⁷⁷ Was Alcuin also a possible recipient of the letter about Pope Hadrian?

The Lorsch manuscript, today BAV Pal.lat.487, was available to a copyist before the end of the ninth century, and he or she makes it plain that Ordo 29 and the note about Pope Hadrian were of highest interest, and, not necessarily, the descriptions of ‘pure’ Roman liturgical practice, which was reconstructed by Andrieu as Collection A’s principal interest. The copy is now in St Petersburg’s Russian National Library, and has been dismembered as three physical units: Q. V. I, n° 34, Q. V. I, n° 56 and Q. V. II, n° 5.Footnote ⁷⁸ The original manuscript may have been from Cambrai, although soon after its production it was in Corbie.Footnote ⁷⁹ Even though they had access to BAV Pal.lat.487, the copyists did not simply copy Collection A from the Lorsch exemplar. The manuscript they produced began with Ordo 11, followed by Theodulf of Orleans’ letter to Bishop John (dating from c.812), a commentary on a ritual of baptism that describes a somewhat different rite of baptism.Footnote ⁸⁰ Three diverse Frankish commentaries on the Mass follow, including the popular Dominus Uobiscum on the Canon of the Mass.Footnote ⁸¹ But there is no sign here of several key texts of Collection A: Ordo 1, 27 and 34. A component of Collection A in other manuscripts but not in BAV Pal.lat.487, Ordo 13 A (the annual cycle of biblical books read in the Night Office), is found in the Corbie manuscript. But the Holy Week Ordo unique to the Lorsch manuscript, Ordo 29, was inserted into the body of this text at the relevant place in the year, and this copy includes the note about Pope Hadrian.Footnote ⁸² Here it appears as the only narrative of Holy Week available. In this case, it seems the Lorsch manuscript was consulted, and the most interesting element was our Ordo 29, a narrative of Holy Week as it unfolded in Lorsch. The ‘completing’ and filling in of ordines with spoken elements was elsewhere used by Andrieu as a sign that a book was intended for liturgical use, and this manuscript is also notable in this regard, having the full texts of the prayers and homilies for Ordo 11 taken directly from the Gelasian Sacramentary and inserted into the body of the ordo.Footnote ⁸³ But this addition also exists here with commentaries and expositiones that contrast with and complement the ordines, offering different versions of the rites the ordines describe.

Our other closely related manuscript, the Regensburg fragments, also acquired another account from Holy Week, specifically about Maundy Thursday, in another hand, celebrated by a mere presbyter and lector, and perhaps revealing a hint of the customs of St Emmeram, where the manuscript had probably come by then (note the Old High German term clocca also used in Ordo 29, n.8, though there the bells did not sound at all).Footnote ⁸⁴

While agreeing with the preceding Ordo 27 in some respects (Ordo 27, n.2 reports that the chant Deus in adiutorium is not said, nor the Gloria, but only the antiphon), in others it actually contradicts it (Ordo 27, n.5: ‘Lumen autem ecclesiae ab initio cantus nocturnae incoatur extingui’). The monks of St Emmeram may not have agreed with all the customs presented in the Murbach manuscript that came their way, and indicated here their differences from them. In the same way as the Lorsch text, this Ordo therefore provided an alternative to the Collection A ceremonies, and indicated that ‘Frankish custom’ was not itself uniform between communities like Lorsch and St Emmeram. Readers were apparently not surprised by this, and expected to read these ordines with a critical eye, comparing the narratives closely.

The same is clearly true for our first manuscript with the complete Collection A, Montpellier 412, from Tours, in which the Collection is presented only after a first non-Roman ordo, Ordo 15. This was the original disposition of the manuscript, the ‘additional’ ordo not being an addition, in this case. The new text, Ordo 15, is a set of ordines and regulations strung through along the liturgical year, and itself contains elements both Roman, including Roman church names, and Frankish, such as baptism at Epiphany, forbidden in Rome.Footnote ⁸⁶ The interest of this text is that it still declares itself to be ‘according to how the holy apostolic Roman church celebrates’, but it is clearly of Frankish design and manufacture.Footnote ⁸⁷ Andrieu had the text written as a group with the rest of the ordines in another early manuscript, the Collection found in St Gallen 349, but that sequence of texts seems rather to have different audiences, and certain details suggest that Ordo 15 in fact originated at Tours, where Montpellier 412 was written.Footnote ⁸⁸ Another early manuscript to which a partial copy of Ordo 15 was added was also in Tours, The Hague, Museum Meermanno-Westreeanium, MS 10.B.4.Footnote ⁸⁹ The ordo makes provision for celebration in a community called a monasterium in an urban setting, in which a bishop would be also present. Such a setting was the ambiguous position of the Basilica of St Martin at the time Alcuin of York was abbot there (796–804), and when the first copies of Ordo 15 begin to appear.Footnote ⁹⁰ This is not to argue for Alcuin’s role in the creation of the ordo (the level of Latinity is much below him), but the text does probably present something of the liturgy at Tours at the time he arrived there.

Like Ordo 29, Ordo 15 covered similar ground to the Collection of ordines that follow it in the Montpellier manuscript, including an extensive narrative of a pontifical Mass which is clearly a simplified version of Ordo 1, and a narrative of baptism which accounts for similar rituals as Ordo 11.Footnote ⁹¹ Alternative versions of the same rituals are also displayed side by side for consultation and comparison.

But Montpellier 412 also has its ordines and Collection A following after the text of Augustine of Hippo’s Enchiridion, and nothing would suggest this was not the original disposition of the manuscript.Footnote ⁹² It was therefore also possible to place this collection in a book along with a resolutely non-liturgical patristic work, as well as the Collection of St Amand, copied on spare leaves of a manuscript of another work of Augustine, the De gratia et libero arbitrio, Paris, BnF, MS lat.974.Footnote ⁹³ We know of another book that placed a Roman order of the Mass with the same text, one that was in St Gallen in 1446, but which is not today identifiable.Footnote ⁹⁴

Some later manuscripts hint at other iterations of Collection A. Paris, BnF, MS lat.2399 is also an eleventh-century manuscript, but certain features of the palaeography suggested to Andrieu and Wilmart that it was copied from a ninth-century exemplar.Footnote ⁹⁵ If so, that exemplar offered only the basic parts of the Collection, Ordo 1, 11 and 27, and these texts were here sandwiched between two contrasting Carolingian commentaries which dealt with liturgical matters, Amalarius’ Liber Officiis and Hrabanus Maurus’ De institutione clericorum.Footnote ⁹⁶

Andrieu assumed more consistency in both the function and form of the Collection than these manuscripts would suggest. Clearly, Collection A was accomplished in steps and stages, and a significant proportion of the manuscripts do not display the full set. This was a fluid set of texts that could easily be adapted, and indeed encouraged the contrasting of the components with alternatives, often adapted or entirely non-Roman ones. The ordines were not therefore seen as definitive accounts of a ceremony but admitted the possibility of the reader’s contrast and comparison. Andrieu’s understanding of ‘Collection A’ as having a single ideological function does not bear the scrutiny of manuscripts. Furthermore, the majority of the ordines in the Collection can be identified as significantly interpolated or written by the Franks, not the pure Roman texts they were supposed to be in Andrieu’s reconstruction.

The idea that pure Roman texts needed to be preserved and made sense only among others of the utmost purity does not seem to be one that animated Frankish recipients of that tradition. The creation of Collection A already associated Roman and Frankish texts which were allowed to interpret each other. Individual manuscript copyists continued to enrich and deepen this relation by adding new texts.