The Ordinary Chants of the Roman Mass, with their Tropes

When, in the summer of 1976, Bruno Stäblein invited me to prepare an edition of the melodies for the Sanctus and Agnus Dei of the Roman mass, together with their tropes, for the series Monumenta Monodica Medii Aevi,1 I had to say that I would be happy to do so, but would not be able to work on the project right away because of other commitments. Little did I know that it would be over thirty years, and that the world would go through at least two great recessions, the United States would elect six different presidents and engage in two different wars before I could finish even one part of the project, and that the edition would appear in a completely different series and in a completely different format from that for which it was originally conceived.2 Rather than being an edition of the Sanctus and Agnus Dei and their tropes, the edition will now comprise the Kyrie and Gloria of the Roman mass and their prosulas and tropes as well.3 The fact that I am writing this article (and gave the lecture upon which it is based) is one positive sign that work on the edition is well under way. But much water has flowed under the bridge between the inception of the project and its present state. Granted, there are a number of things that kept me from completing this edition any sooner. One was a series of articles and then a book

on a completely different subject, namely the nexus of tone-system, mode, and notation in early medieval music. When the book of that title appeared in 2009, 4 I could finally return to the editorial project in earnest. But I would have to admit that even greater obstacles were presented by the task of editing a large body of texts with music. Those will be the main focus of this article. But I shall discuss them within the broader framework that marks the history of this edition. 5 As mentioned above, I am preparing an edition that includes the melodies for the Ordinary chants of the Roman mass and their tropes. This edition will indeed comprise both texts and music. But while the texts of the Ordinary chants themselves are relatively stable, the texts of the tropes for these chants are characterized more by their diversity than by their consistency. In her seminal article 'Problems in the Editing of Tropes', which appeared in the first issue of the journal Text in 1984, 6 Gunilla Iversen points out that with regard to the items that make up the Latin mass during the period between the ninth and the twelfth centuries, "the multiplicity, the great variety, is in fact more striking than the uniformity." 7 In that same article, she goes on to discuss editorial problems pertinent to various types of tropes, ranging from the prosulae added to melismas in various types of chant, to tropes for both the Proper and Ordinary of the mass. Since her remarks pertain as much to my edition as to hers, I should like to reiterate some of her points here.
Regarding the character of the Latin texts, she states that the Latinity of the tropes can vary considerably according to type of chant, the provenance of a manuscript, and even across chants within the same manuscript. 8 The language of prosulae, for example, as found in the Sanctus repertoire, can often be "very peculiar" as she describes it. 9 This is because these texts were in many cases added to a pre-existent melody, and the number of syllables of the text had to match the number of pitches in the melody. Beyond this generic difference, one must keep in mind that the Latin language evolved during the course of the Middle Ages, thus is not the same in late as in early texts. 10 This plays a concrete role in the tropes, since newer pieces were constantly being added to the repertoire, but stand side by side with older ones in the manuscripts themselves. 11 In addition, there are regional differences in the quality and character of the Latinity one encounters in tropes that must be dealt with by the modern editor. 12 Regarding the task of the editor of tropes, Iversen points out three fundamental problems, which are pertinent whether one is editing the texts by themselves or with their music. 13 The first is to establish what she calls the "main structure" of a chant and its tropes, presenting the combinations of the elements or verses that belong to a given trope and demonstrating the way the elements and the base liturgical text link together. The second problem is to present the edited text in such a way that variants are presented as readings that are just as valid as those of the edited text itself, a point to which I shall return below. Her final problem is one that is overarching, and plays perhaps an even more important role in preparing an edition of texts with music than one of the texts by themselves: namely, the problem of trying to ascertain who will use a given edition and how much and what kind of information that user will need.
In addressing the special problems of editing trope texts with music, I should like to begin by underscoring Iversen's point that in dealing with liturgical texts we cannot define "authenticity" in the same way that we would for editing literary texts. In the case of the latter we are dealing usually with a text by a known author, and the job of the editor is to produce a text that is as close as possible to the final state of the text by that author. In the case of liturgical texts, however, we rarely know the name of the author, Notker Balbulus and Tuotilo of St. Gall, along with Adémar de Chabannes being noteworthy exceptions. 14 But having the name of an author is not the deciding factor when one is dealing with liturgical texts and their music. As Iversen says: The most important thing is that each different version of a text as long as it belonged to a liturgical practice, is to be regarded as authentic. The most interesting task for us today is not to come back to and re-establish the first, the earliest text, which may be the natural editorial impulse, but to give as accurate a picture as possible of all the versions that were used in their own right in different regions. 15 In accordance with that statement, she and the other editors of Corpus Troporum have adopted the principle of maintaining the text of a trope as it appears in the manuscript, "as long as there is the slightest possibility of making sense out of it." 16 This strikes me as an eminently sensible decision, and is one I plan to adhere to in my own edition of tropes to the chants of the Ordinary. Let me now turn to that edition, relating something of its history, and some of the problems it will have to solve.
I must preface my remarks by returning to the third of the problems Gunilla Iversen outlined under the tasks of an editor: namely, the problem of trying to ascertain who will use a given edition and how much and what kind of information that user will need. Here lies the biggest difference between editing trope texts by themselves and editing those same texts with music. The latter have to be presented in such a way that they can be sung (even if the potential user is tone-deaf!). All sorts of decisions flow from that simple and obvious fact. One cannot simply prepare an edition for silent reading or the comparison of textual variants. One has to make decisions that can be converted into sound.
The need to provide performable music is only appropriate, since the texts were clearly sung in their role as parts of the liturgy, but lacking an oral tradition that extends back to the ninth or tenth centuries, we must present them in a type of musical notation that a modern singer can read, i.e., in ordinary staff notation, using either tenor or bass clefs. This means that we are forced to prepare editions of the melodies using some form of diastematic, that is to say intervallically precise, musical notation. Since the advent of fully diastematic notation takes place only in the eleventh century, in particular as outlined in the Prologus in Antiphonarium of Guido d'Arezzo (ca. 1030), 17 virtually any manuscripts copied before that time are not likely to be useable as primary sources for an edition. We may consult them and perhaps even collate their neumatic readings for the melodies, but without clefs they must be considered non-diastematic, and their melodies transcribed only from later, fully diastematic concordant sources. As an example of this, please see Examples 1 and 2, below.
Example 1, to which I shall return below, presents Gunilla Iversen's edition of the texts of the trope set Omnipotens aeterna Dei, a trope that appears in a number of tenth-and eleventh-century sources from England, France, and northern Italy. Example 2 presents musical settings for the first element of Omnipotens aeterna Dei in four manuscripts dating from the tenth to the twelfth centuries. As one can see in the example, the Aquitanian manuscript settings -A, B, and C-are relatively well heighted, but without clefs; given this lack, one really cannot sing the melody from them. Fortunately, this set of tropes was also copied into several later sources, including the twelfthcentury manuscript Paris, BnF, lat. 10508, from St Evroult, whose setting appears as Example 2 D. 18 This source presents the trope on lines and with clefs, thereby permitting a reliable transcription into modern notation.
Related to the necessity of transcribing from diastematically notated manuscripts is the fact that a number of tropes cannot be found in any diastematic sources at all, hence cannot be part of an edition of the music. By my count there are fourteen of the Agnus Dei tropes and fifteen Sanctus tropes in Gunilla Iversen's editions in Corpus Troporum volumes IV and VII whose melodies are not transcribable. 19 One has to lament this, because some of these tropes have quite attractive texts. The earliest tropes from St. Gall are a special case. They are written for the most part in good Latin, and have clearly legible melodic settings. But because the earliest manuscripts from St Gall are notated in campo aperto -not on a staff, but in open field-the only way we can make those tropes available is by offering them in transcription from a later manuscript, if a concordant source may be found at all. In this regard we are fortunate to have a late manuscript from St Gall, St Gall, Stiftsbibliothek, codex 546, copied in 1507, that offers diastematic readings of a number of early tropes from the monastery that would otherwise be completely lost to us today except for their texts. 20 St Gall 546 is a "collectaneum" compiled by Frater Joachim Cuontz, hence not a manuscript belonging to a specific liturgical tradition. One can question whether a melody copied in St Gall 546 is in fact the same as the one with which a given text was originally sung, but in this case, as long as the neumes concord with each other, having a melodic setting -even a late one-is better than having no setting at all. As examples, I would cite the St Gall Agnus tropes Christe, theos agye and Patris factus hostia, both of which have diastematic concordances only in St Gall 546. 21 The need to resort to transcribing, say a ninth-century trope from a sixteenth-century source is indicative of yet another issue that concerns tropes to the Ordinary in particular. As we know, tropes begin to be composed no later than the early part of the ninth century, as the canon of the Synod of Meaux (845/6) witnesses, and continue to be written throughout the Middle Ages. 22 The great bulk of the repertoire, however, in particular Proper tropes, is made up of texts and music composed between the ninth and twelfth centuries. The decision of the editors of Corpus Troporum to establish 1100 as cut-off date for their editions of Proper tropes was a wise and practical one. 23 As Gunilla Iversen has pointed out in her editions of tropes to the Sanctus and Agnus Dei, however, the composition of tropes for these two chants of the Ordinary of the mass continued through the Middle Ages and on into the Renaissance. Accordingly, she included some "later" texts in her edition of Agnus Dei tropes, 24 and even more in the editions of those for the Sanctus 25 and the Gloria. 26 The problem I face, however, is that my edition will be an edition not just of tropes, but of the Ordinary Of course, nothing about editing is simple, and certainly not editing a repertoire as diverse and problematic as the Sanctus and Agnus Dei, using manuscripts dating from the ninth through the sixteenth centuries. My initial plan was to transcribe each of the melodies whose incipits appeared in Thannabaur's and Schildbach's catalogs, using the same sources from which the two earlier scholars had transcribed incipits. This would have involved transcribing the melodies from a colorful array of manuscripts, but without placing them in any specific liturgical context. It would also have meant transcribing one melody for each trope that appeared in manuscript together with the base melody that had already been catalogued -but it would also have meant that many of the tropes not appearing with those melodies in their manuscript sources would not have been edited at all. When I started collating sources for the tropes by themselves, the question became one of relating the tropes to their base melodies. Since some melodies can appear with a fairly large number of tropes, should one include the base melody along with every trope that appears with it? This is partly a philosophical question, but it is also a practical question of size.
The presentation of a printed version of a piece of music takes considerably more space than does the presentation of its text alone. Even the transcription of an Agnus Dei melody with a set of almost syllabic trope elements takes substantially more space on a printed page than would the transcription of the texts of the base chant and trope texts by themselves. We shall see evidence of this in the following examples.
Another problem that is raised by the musical settings of trope texts is that there can be several different melodies for a given text. Let me use as an example one of the trope sets that Gunilla Iversen used in her earlier article, the trope set "Omnipotens aeterna Dei." It may well be the best known of all tropes to the Agnus Dei, since it appears as the first in the series of such tropes in volume 47 of Analecta Hymnica Medii Aevi. I have provided the set as it appears in Corpus Troporum IV in Example 1 above. As one can see there, the text itself presents a number of questions, starting with the very first line. Is it Omnipotens aeterna Dei or Omnipotens aeterne Dei? Obviously, Iversen decided in favor of aeterna, modifying sapientia, but a large number of manuscripts read aeterne, modifying Christe.
The musical settings of this set of tropes add an additional layeror perhaps several layers-of complexity to the task of preparing an edition. These problems arise in part because this trope complex appears with five different Agnus Dei melodies,numbers 64,78,119,226,and 253 in Martin Schildbach's catalog. Even when a given trope verse or set of verses appears with a single melody, its melodic settings can differ somewhat among themselves. When a different base melody is involved, as is the case here, one can -and very often does-have different trope melodies as well. One can see this in Examples 3 and 4 below. As we shall discover, there are three different melodic traditions for the trope complex Omnipotens aeterna, one for melody 78, which we saw in Example 2 above, another for melodies 64, 119, and 253, and yet another for Melody 226. Let us take a closer look. 32 Example 3  As one sees in Example 3, Melody 64 is a melismatic setting of the text, with a final on D and a range extending from the subfinalis C to the c an octave higher. (I call this "melismatic" because it frequently has five or more notes, or "melismas," over a single syllable of text.) It appears most often untroped, but it can also be prefaced by the introductory tropes Haec festa precelsa, and Pro cunctis deductus. 33 Melody 78, the melody with which Omnipotens aeterna dei most often appears, is a more modest, neumatic setting of the liturgical text -"neumatic" because it typically has between one and five notes over a single syllable, and hence is a bit less ornate than a melismatic chant such as Melody 64. Melody 78 begins on G, and has a range from D, a fourth below, to e a fifth above the opening pitch. Its final is on a, a fact that caused Schildbach to make the remark that "the final a does not correspond to the modal structure of this melody." 34 Schildbach also provides a version of the melody transposed down to start on C, with a final on D, and extending from the A beneath the initial C to the a an octave above it. 35 Although he does not catalog it as such, this transposed version is actually his Melody 253. 36 I have argued elsewhere that this is most probably the original pitch-level of this melody, which was transposed up to G, Schildbach's Melody 78, and also to F with a B-flat, which is Schildbach's Melody 119. 37 All three versions of the melody appear with Omnipotens aeterna dei as an internal set of tropes. As we shall see below, the upward leap of a fifth that occurs on the first syllable of the word peccata in melodies 78, 119, and 253 is a striking gesture that is actually reflected in the melodic ductus of the versions of Omnipotens aeterna that are sung with these melodies.
Finally, the two tropers from Apt set Omnipotens aeterna dei with Schildbach's Melody 226, the most widespread of all Agnus Dei melodies. 38 Like Melody 253, although somewhat less ornate, it is a neumatic setting that has its final on D; its range, however, is narrower than the other melodies, extending only a sixth from the subfinalis C to the a a fifth above the final.
We have already seen several of the earliest settings of the first verse of the set Omnipotens aeterna/Verum subsistens/Optima perpetue in Example 2 above; all of these appear with Schildbach's melody 78 in their respective sources. Given that all the settings but the last one are notated without clefs, it is impossible to know whether those appearing with Melody 78 are to be sung starting on G or on the C below it. This ambiguity of pitch level is only reinforced by the early diastematic settings. For example, Paris, BnF, lat. 10508 (D in Example 2) places the melody on C, which would correspond to Melody 253, the transposed version of Melody 78; other diastematic sources, such as Biblioteca Casanatense 1741 (in Example 3 above), place it a fifth higher on G, corresponding to the pitch level of Schildbach's Melody 78 itself.
Example 4 shows a sample of my collations of all three of the verses of Omnipotens aeterna/Verum subsistens/Optima perpetue as they appear with melodies 78 (on G), 119 (on F), 64 and 253 (on C), and 226 (starting on D) -here Examples 4A-D respectively.
The collations for lines 4A and C, melodies 78 and 253, are only partial; there is a substantially larger number of sources for this version of the trope than for the other two. What one sees at the top of each system of the example is a transcription from a reliably diastematic source -or at least as reliable a source as I could find. Beneath each of the transcriptions in 4A and 4C are my collations. In the cases of manuscripts with good heighting of neumes, but no clefs -here Paris, BnF, n.a. lat. 1871 and BnF, lat. 909-I chose to collate with Schildbach's Melody 78, as one can see in Example 4A. The exception to this decision is the setting in Apt 17. Since it appears with Melody 226, consistently pitched on D, I have transcribed it at that pitch level. The versions pitched on C are clearly indicated by clefs in their respective manuscripts.
In comparing the various versions in Example 4, one will notice that versions A and C are quite similar to each other in that the general melodic shape of the trope melody is fairly consistent, whether placed on C or G, and occurring with melodies 78, 64, or 253 as they appear in Example 4A and C. Example 4B, the version from London, BL, 2 B IV with Agnus Dei melody 119, has an incipit rather unlike the other two, but once it reaches the interval of a fourth above the initial pitch -on the syllables -potens of omnipotens-it continues in a manner very close to 4A and especially 4C. Particularly striking in all three settings 4A-C is the upward leap of a fifth on the first syllable of the word Optima that begins the third verse. This corresponds to the same leap that occurs in the melodies 78, 119, and 253 on the first syllable of the word peccata in the phrase peccata mundi, which I pointed out above in Example 3. Although it does not have a leap of a fifth on peccata, Melody 64 begins by traversing that same interval moving downwards.
The melodic kinship of the base melodies 78, 119, 64, and 253 and those of the trope verses set with them is clearly strong. One can say the same thing about the very different version of the trope in Apt 17, which sets the verses of Omnipotens aeterna/Verum subsistens/Optima perpetue with base melody 226. One can see Apt's setting of Melody 226 in Example 3 and the trope verses in Example 4D. As mentioned above, melody 226 moves only within the range of a sixth, C to a, rather than the more expansive range of an octave or a ninth of the other base melodies. And as one might expect, the trope melody moves only within this same range, hovering around the opening pitch for the first three words of the text before moving upward to a on the word sapientia.
The principal question raised by these versions is which ones to print in an edition, and according to what criteria. Were these to appear in an edition for Monumenta Monodica I would present one representative from each of the four versions presented in Example 4. Those from London, BL, 2 B IV and Apt 17 would be easy to justify, since they are each the only examples of their types -London, BL, 2 B IV because it appears with Schildbach's Melody 119 and Apt 17 because it is a unique melodic setting of the verses that appears with Melody 226. The others would not be quite so straightforward. Despite wanting to include a version from Aquitaine that was sung with Melody 78, I would feel compelled to use Rome, Casanatense 1741 as representative of those settings. This is because 1) the versions from Aquitaine might not actually have been sung with this melody and 2) because with aeterne instead of aeterna in the first verse, they are not as close to the text Gunilla Iversen established for this trope in Corpus Troporum IV. A similar rationale might lead me to choose Paris, BnF, lat. 10508 over Paris, BnF, lat. 1177 for the version in Example 4C, pitched on C. It has perpetuae in the third verse, modifying vitae; the other sources have perpetua, modifying gaudia. One could justify either choice, perhaps, but I would concur with Iversen on perpetuae as the preferred reading. Pa 10508 also has the slightly better Optima as the first word in this verse, as opposed to the Obtima in Paris, BnF, lat. 1177. In the final analysis, though, I would probably choose to present the version in Paris, BnF, lat. 1177 in the edition for the sake of completeness: it is the only setting of Omnipotens aeterna dei that also includes the verse Rex regum, gaudium angelorum, Christe. 39

The musical edition in Corpus monodicum
The reader will have noticed that I have been using the subjunctive mood in my discussion of Omnipotens aeterna dei. This is because for various reasons my edition of Sanctus and Agnus Dei melodies with their tropes will not be appearing in the series Monumenta Monodica Medii Aevi after all. Instead, as mentioned above, it will be appearing in a new series conceived as the successor to Monumenta Monodica, namely the Corpus monodicum, the Project Director and Editor-in-Chief of which is Andreas Haug of the Universität Würzburg. Funded by a major grant from the Union of the German Academies of Science, Corpus monodicum has as its goal the edition of previously unedited repertoires of sacred medieval monophony. 40 Since one of the major repertories of sacred monophony that has not yet received a comprehensive edition of its music is the Ordinary of the mass, it made sense to place the edition of the Sanctus and Agnus Dei with their tropes in this new series and edit them together with both troped and untroped Kyrie and Gloria settings. I am delighted to report that Gunilla Iversen will be working with us on the texts of the tropes, insuring that the editions of the texts will be at the very highest level. 41 The situation we have just seen and discussed with regard to one complex of tropes to the Agnus Dei is characteristic of the situation that obtains in the repertoire of Sanctus and Agnus Dei tropes as a whole. Hence, when Andreas Haug and I started to discuss how best to edit these trope complexes in the edition for Corpus monodicum, this was one of the foremost questions. My first thought was that we could give a representative version of each melodic tradition, in the manner just discussed, and provide information as to both textual and melodic variants in the critical apparatus. In order to test the viability of this approach, I collated all the manuscript settings of the Sanctus trope Admirabilis splendor, which is the very first in Gunilla Iversen's edition in Corpus Troporum VII and no. 4 in the catalog of tropes prepared by Peter Josef Thannabaur. One sees the tip of the iceberg in Example 5 below.
I have provided in Example 5 representative transcriptions of each of the three melodic versions of the trope, together with the Sanctus melodies with which they appear in their respective sources.
As one can see here, the three versions are quite different. The version with melody 111 (Example 5A) has the word Admirabilis beginning with an ascending F-major triad. The one with melody 74 (Example 5B) has the same word hovering around the pitch G until the melisma on the final syllable. The setting with melody 49 (Example 5C) is much more active than the other two, setting each syllable with 3-or 4-note neumes. 42 The textual edition and critical apparatus for both text and music of this trope complex take up 6 single-spaced pages, using a 12-point type font. If one adds the actual transcriptions of the three pieces, one has nine manuscript pages of material for this single trope. Multiply the 9 pages that this one complex requires by the 174 tropes that Gunilla Iversen includes in her edition, or the more than 300 Sanctus tropes that Thannabaur has catalogued -and that I plan to include in my edition-and one realizes that an entire forest of trees would have to be felled in order to provide enough paper! If the apparatus provided the kind of information that a scholar or performer could easily use to reconstruct the versions of the trope melodies as they appear in specific manuscripts, then 2,700 or so pages of manuscript would not be unreasonable. In our view, however, this method of presenting tropes and their variant readings seemed overly cumbersome, to the point of being unwieldy. Andreas Haug and I, along with other members of the team working on Corpus monodicum, have therefore decided to take a different approach.
We felt that it would best serve the needs of both performers and scholars to present editions of texts with their melodies as they appear in specific manuscripts representing specific geographical areas and liturgical traditions -i.e., an edition based on manuscript sources, not on genres of chant. A provisional list of sources grouped according to these criteria, appears as Example 6.
As one can see in Example 6, we are in essence following the same groupings as does Corpus Troporum. Each group of manuscripts will be divided into two subgroups: primary and secondary. Following procedures already established for the other five repertoires being edited in Corpus monodicum, we plan to present not just the settings of the Sanctus and Agnus Dei, but rather the complete corpus of Ordinary chants -Kyrie, Gloria, Sanctus, Agnus Dei-from each of the primary manuscripts in the order in which they appear in that source, complete with rubrics and cross references. Each individual chant will have its own "critical apparatus," but as Iversen suggested in "Problems in the Editing of Tropes," the reading in any given manuscript will be maintained "as long as there is the slightest possibility of making sense out of it." 43 Ordinary chants that do not appear in the primary sources C Paris, BnF, n.a. lat. 3126, fol. 75v: Example 6. Provisional list of core manuscripts for the Corpus monodicum edition of melodies of the Ordinary of the Mass with their tropes: from a given geographical area will be edited in their entirety from the secondary sources of that group; melodies that have already appeared in one or more of the primary sources will be given only via cross reference. Should there be items of the Ordinary sung in a given geographical area that do not appear in either the primary or secondary sources for that area, we shall edit them individually as they appear in other manuscripts from the same region. 44 The trade-off here (as it would have been in the edition for Monumenta Monodica as well) is that this edition cannot pretend to be a "complete critical edition," in the sense that the editions of the texts in Corpus Troporum certainly are. Our hope, however, is that even if it cannot present a picture of all the versions that were sung in a given tradition, it can at least provide a representative sample.
The way individual pieces might look in this edition can be gathered by considering the presentation of two versions of the Sanctus prosula Clangat hodie vox nostra, which appears with two different Sanctus melodies in manuscripts from two different geographical regions: one from the "Southwest" and the other from "Northern France and England" in the groupings given in Examples 8 and 10 present two versions of Clangat hodie. Some readers will recognize it as a piece on which Gunilla Iversen and I have written complementary studies: she in "The Mirror of Music: symbol and reality in the text of Clangat hodie," and I in "Music and Meaning in Clangat hodie," both of which appeared in the Proceedings for the Madrid meeting of the International Musicological Society and its study group Cantus Planus in 1992. 46 Gunilla Iversen has also edited the text of this prosula in Corpus Troporum VII. 47 The reader will see her edition of the text in Example 9, together with my translation.
Clangat hodie appears in some fifteen manuscripts dating from the eleventh through the fourteenth century. A list of these sources appears in Example 7. As one will see in that example, there are two versions of the piece. One, appearing in manuscripts from southwestern France and Spain, sets the text to an extended Osanna melisma for Sanctus 190 in Thannabauer's catalog of Sanctus melodies. A transcription of a representative setting of this version of Clangat hodie, taken from the twelfth-century manuscript Huesca, Biblioteca de la cathedral, codex 4, appears in Example 8. 48 The second version of Clangat hodie appears in Example 10, transcribed from the twelfth-century Norman-Sicilian manuscript Madrid,Biblioteca Nacional 289. 49 This is the version one finds in Example 9. Text and Translation of Clangat hodie: manuscripts from northern France and England, or what Corpus Troporum calls the "Northwest and Zone of Transition." As one can see by comparing examples 8 and 10, the version in example 10 appears with a different Sanctus melody, melody 112 in Thannabauer's catalog. It is noteworthy that in its earliest settings, Melody 112 appears together with the prosula Clangat hodie, suggesting that this is a new composition of both text and melody.
Example 10. Sanctus 112 with Clangat hodie (Version II : Madrid Biblioteca Nacional,MS 289,: I have posited elsewhere that this second version of the prosula with its associated Sanctus melody was newly composed in order better to reflect the meaning of the text, a text that is remarkable for the amount of musical imagery and terminology it contains. 50 That text, in Gunilla Iversen's edition and with my translation, appears in Example 9. With its references to simphonia, organum, harmonia, tonorum modulamina, vocum discrimina, diapason and tetrachordum, along with the more common carmina and hymnus, Clangat hodie stands almost alone in the Sanctus repertoire. Indeed, I know of no other Sanctus prose that makes such extensive and purposive use of technical musical terms. The import of those terms is underscored by the use of the C mode for both Sanctus 112 and the prosula Clangat hodie. The C mode -or more accurately, the C scale-to which the text has now been set had first been discussed theoretically in the ninth-century Scolica enchiriadis and in the De harmonica institutione of Hucbald, in both cases being identified with instruments. 51 One cannot help but wonder whether there is any connection here with the tinnula armonie organa mentioned in line 2a of the prose. But there are yet other gestures that link this melody firmly with harmonic theory as presented on instruments: At the beginning of line 3b, the diapason really does ring out altisona, thanks to the octave leap that introduces it. This remarkable gesture had been prefigured by the octave leap between the end of line 1a and the beginning of 1b -introduced by the word sinfonia. In the Pythagorean mathematics underlying ancient Greek harmonic theory, the diapason, with its ratio of 2:1, was the most perfect sinfonia. 52 In his commentary on Martianus Capella, John Scottus even goes so far as to equate the diapason with harmonia itself. 53 Moving further in line 3a of the prose, one finds that the diapason is reached per vocum discrimina, tetrachordis...conscendens. The first part of this phrase is of course a reference to the "seven discrete pitches" with which Orpheus plays in Book VI of the Aeneid. 54 The second part reminds us that these pitches are produced by a concatenation of tetrachords. 55 As if to render this scalar structure concretely, the melody for lines 3a and 3b of the prose divides into two disjunct tetrachordal strata (G-c, C-F) connected by the four-note figure on the word "tetrachordis." Underscoring this division even further is the fact that the upper tetrachord (G-c) is projected in stepwise motion at the beginning of lines 2a through 3b. Thus, the melody of Clangat hodie does indeed 'ascend in tetrachords through discrete pitches,' as the text says, and in a clear, yet subtle way.
I have discussed this piece in detail because it is one telling exampleamong many that could be cited-of the importance of editing texts with music. Some readers may know that in American musicological circles in recent years there has been a tendency to denigrate the preparation of editions as being old-fashioned, "positivist" musicology. In her plenary lecture for the 1985 meeting of the four major American musical societies, Margaret Bent pointed out that the making of editions is an activity that demands not only a high level of philological skill, but an equally high level of critical ability -precisely the kinds of skills that have characterized the best musicology in the past and will continue to do so in the future. 56 But there is even more to it than that.
In this paper I have focused mostly on the problems involved with editing texts with music. They are indeed myriad -but not insurmountable. And what they make possible, in a piece such as Clangat hodie, is an example of nuptiae bellae philologiae et musicaea beautiful wedding of philology and music-that makes it possible to bring these ancient texts to life in performance. Although performance is not the only desired outcome of a scholarly edition, it is certainly one of things that makes the preparation of editions of texts with music infinitely fascinating and rewarding, and ultimately tremendously exciting. (Kassel: Bärenreiter, 1956-). The series, founded by Bruno Stäblein, began with his own edition, Hymnen: I, Die mittelalterlichen Hymnenmelodien des Abendlandes.

Monumenta Monodica Medii Aevi
2. The edition will now appear as part of Abteilung I: Ordinariumsgesänge, in the series Corpus monodicum, die einstimmige Musik des lateinischen Mittelalters, ed. Andreas Haug, published in Basel by Schwabe Verlag. Rather than being an edition arranged according to genres, as was Monumenta Monodica, Corpus monodicum is a Quellen-Edition ("source edition"), one that presents repertoires as they appear in complete manuscripts. For more on this, and its implications for my own edition, see below.
3. The Kyrie appears in two forms in the earliest manuscripts containing it, both in melismatic versions exhibiting only the transliterated Greek text (Kyrie eleison 3x, Christe eleison 3x, Kyrie eleison 3x), and in versions that contain Latin texts underlaid to the same melismas in a syllabic fashion -one syllable of text for every individual note of the melody. These latter are usually referred to as "texted," "prosulated," (or "Kyrie with prosulae"), or simply "Latin" Kyries. On this see Alejandro Planchart, 'Trope [ii] [a]', in New Grove 32. In the discussion that follows I use letter names to indicate pitches. Gamma (Γ) is equivalent to the G on the first line of the staff in bass clef. A represents the note a second above that; a the note an octave higher; and aa the note yet another octave higher.
33. Haec festa praecelsa and Pro cunctis deductus are trope sets 27 and 49 in Iversen, Tropes de l'Agnus Dei, CT IV. For Melody 64 see Schildbach,Agnus Dei,p. 92. 34. Schildbach,Agnus Dei,p. 95. 35. Ibid.,p. 163. 36. Ibid.,p. 160. Tropes,Agnus Dei, 39. Cf. Example 1 above, where Pa 1177 is the only source with the verse order ABCD. 41. It almost goes without saying that the availability of excellent editions of the texts of tropes to the Gloria, Sanctus, and Agnus Dei, all prepared by Gunilla Iversen for the series Corpus troporum (volumes XII, VII, and IV respectively), will be of great benefit to this new edition of the texts with music. 42. In Example 5, neumes of the original notation that comprise two or more notes have been transcribed under slurs. Single note-heads correspond to puncta in the original, neumatic notation. 43. Iversen, 'Problems', p. 104. Cf. n. 16 above and the text associated with it.

Atkinson, The Earliest Settings of the Agnus Dei and its
44. This will be the case, for example, with the rather large number of both melodies and tropes for the Sanctus and Agnus Dei that are unica, and do not appear in the manuscripts we have selected as primary or secondary sources for each of the regions and orders, as given in Example 6.