¹ Hamm, C., ‘Manuscript Structure in the Dufay Era’, Acta Musicologica, 34 (1962), pp. 166–84CrossRefGoogle Scholar.

² Examples of the former type include Bc q15 (inventory by de Van, G., ‘Inventory of Manuscript Bologna, Liceo Musicale, q15 (olim 37)’, Musica Disciplina, 2 (1948), pp. 231–57)Google Scholar, and Bu 2216(see Besseler, H., ‘The Manuscript Bologna, Biblioteca Universitaria 2216’, Musica Disciplina, 6 (1952), pp. 39–65Google Scholar, and Gallo, F. A., ed., Il codice musicale 2216 della Biblioteca Universitaria di Bologna, Monumenta Lyrica Medii Aevi Italica, ser. iii: Mensurabilia 3/i–ii (Bologna, 1968–1970))Google Scholar. Perhaps the most complex example of the latter type is Ob 213, the subject of a detailed palaeographical study by Schoop, H., Entstehung und Verwendung der Handschrift Oxford, Bodleian Library, Canonici misc. 213 (Berne and Stuttgart, 1971)Google Scholar.

³ The reader is warned that the facsimile reproduction of the Codices, Trent, Codex Tridentinus 87–93 (Rome, 1969–1970)Google Scholar, is unreliable in many details: for example, the omission of material at page edges, such as gathering signatures and marginalia; the distortion by the method of photographic reproduction of void notes so as to make them appear solid and the converse; and the frequent touching-up of damp-damaged initials and staves. Most blank pages are not reproduced.

⁴ Haberl, F. X., ‘Wilhelm Du Fay: monographische Studie über dessen Leben und Werke’, Vierteljahrsschrift für Musikwissenschaft, 1 (1885), pp. 483–94Google Scholar.

⁵ Adler, G. and Koller, O., eds., Seeks Trienter Codices: Geistliche und weltliche Compositionen des XV. Jahrhunderts, Erste Auswahl, Denkmäler der Tonkunst in Österreich Jg. vii, 14–15 (Vienna, 1900), pp. xiii–xivGoogle Scholar.

⁶ On the former see Ward, T., ‘The Structure of the Manuscript Trent 92–i’, Musica Disciplina, 29 (1975), pp. 127–47Google Scholar; for an edition and repertorial study of the latter see White, R., ‘The Battre Fascicle of the Trent Codex 87’ (Ph.D. thesis, Indiana University, 1975)Google Scholar.

⁷ All folio references in TR are to the nineteenth- or early-twentieth-century foliation in the top right-hand corner of each recto. While most of the original foliation of Tr92₁ survives, the sole vestige of an earlier system in TR is the figure ‘144’ onTr92, fol. 144^r. (Presumably all the other numbers were lost through trimming.)

⁸ The various studies that deal with the make-up of TR are as follows: Adler, and Koller, , eds., Sechs Trienter Codices, pp. xiii–xv; Hamm, ‘Manuscript Structure’, pp. 168–75Google Scholar; Hamm, , A Chronology of the Works of Guillaume Dufay (Princeton, 1964), pp. 70–1, 81Google Scholar; Hamm, , ‘A Group of Anonymous English Pieces in Trent 87’, Music & Letters, 41 (1960), pp. 211–15CrossRefGoogle Scholar (see the correspondence on the latter between Hamm, and Trowell, B. in Music & Letters, 42 (1961), pp. 96–7, 295–6)CrossRefGoogle Scholar; and Kanazawa, M., ‘Polyphonic Music for Vespers in the Fifteenth Century’, 2 vols. (Ph.D. thesis, Harvard University, 1966), i, pp. 94–107Google Scholar.

⁹ Gatherings are referred to by small upper-case roman numerals; superscript arabic numerals indicate the number of folios in a gathering. In this context the sigla Tr87₁ and Tr92₂ are contracted to ‘87’ and ‘92’ respectively.

¹⁰ This section of the manuscript is peripheral to the present discussion and is the subject of a separate study which I am currently preparing.

¹¹ All the main types of watermark in Tr87–93 are reproduced in Kanazawa, ‘Polyphonic Music’, ii, pp. 215–22; and Chemelli, A. and Lunelli, C., Filigrane trentine (Provincia Autonoma di Trento, 1980), pp. 223–9Google Scholar. But in both cases the reproductions are only approximations of the original marks.

¹² The marks of paper la (mould side, left folio) and paper 2a (mould side, left folio) are close to Briquet's nos. 11702 and 11882 respectively, and those of papers 3 and 5 similar in type but not in detail to nos. 7890 and 6387 (see Briquet, C. M., Les filigranes (Geneva, 1907; rev. edn, Hilversum, , 1968))Google Scholar. The closest variant I have found to the marksofpaper 4 is no. xi 99 in Piccard, G., Das Ochsenkopf-Wasserzeichen: Findbuch ii (1–3) der Wasserzeichenkartei Piccard im Hauptstaatsarchiv Stuttgart (Stuttgart, 1966)Google Scholar.

¹³ For an explanation of moulds and the marks they produce see Stevenson, A., ‘Watermarks are Twins’, Studies in Bibliography, 4 (1951–1952), pp. 57–91Google Scholar.

¹⁴ The term ‘state’ refers to the variant states of identical marks (that is, marks from the same pair of moulds) as they grow old, ugly and distorted (see Stevenson, A., ‘Paper as Bibliographical Evidence’, The Library, ser. v, 17 (1962), pp. 202–3)Google Scholar. The only safe criterion for establishing states is the pattern of sewing dots (that is, the points at which the mark is attached to the mould), which can never be the same in marks from different moulds. The naked eye is an unreliable guide in such bibliographical minutiae, and it is possible that beta-radiography could lead to a more refined view of the paper in TR. I believe it unlikely, however, that this would disturb my central hypothesis.

¹⁵ All the evidence suggests this was scribe A: most indented staves in his work are extended with a rastrum (for example, Tr92, fol. 163^v), while those in the work of his assistants are extended freehand (for example, Tr92, fols. 160^v, 161^v); moreover, there often appears to be a correlation between the care exercised in ruling the staves and that taken in entering the contents (compare, for example, gatherings 87i and 87ix).

¹⁶ It is possible to compare widely separated folios by tracing and superimposing these patterns. I am thus able to show, for example, that (in paper la) Tr87, fols. 3/10, 15/22 and 73/84 have one pattern of holes and were therefore pricked simultaneously, while Tr87, fols. 63/70, 75/82, 76/81 and Tr92, fols. 149/150 have another, quite different pattern. This seems to indicate that the scribe was himself responsible for arranging these sheets in gatherings.

¹⁷ Sheets of different paper which have the same number of staves to a page and have in common the presence or absence of a staff indentation are generally distinguished from one another by the way in which the staves are distributed on the page. Thus in paper 2c, for example, the staves are much more widely spaced than in paper 2b. The only exceptions to this pattern are papers 1a and 2a which have almost identical lay-outs.

¹⁸ While several writers have recognised this fact, none has attempted to define the extent of scribe A's activities. Flotzinger, R., ‘Buntschucherh-explicit’, Festschrift Walter Senn, ed. Egg, E. and Fässler, E. (Munich, 1975). pp. 89–92Google Scholar, has convincingly shown that the word ‘Puntschucherh’ on Tr87, fol. 161^r is not, as has been traditionally believed, the name of the main scribe, but a form of scribal explicit.

¹⁹ For a discussion of these works see Planchart, A. E., ‘Guillaume Dufay's Masses: a View of the Manuscript Traditions’, in Papers Read at the Dufay Quincentenary Conference, Brooklyn College, December 6–7, 1974, ed. Atlas, A. W. (New York, 1976), pp. 33–7Google Scholar.

²⁰ The Agnus setting on Tr87, fols. 139^v–140^r has internal concordances on Tr87, fol. 23^r–23^v, and Tr92, fols. 207^v–208^r, the second of which bears an attribution to ‘Dunstable’; the Credo on Tr92, fols. 204^r–205^r, is designated ‘Anglicanus’; the two Sanctus settings on Tr87, fols. 103^v–104^v and fol. 105^r–105^v, are elsewhere attributed to ‘Dumpstabl’ (Tr90)/ ‘Bennet’ (AO) and ‘Leonel’ (OH) respectively; and the Gloria Jacet granum (Tr87, fols. 31^v–33^r and 141^v–142^v) is manifestly English. This last piece and a number of works in TR of suspected English origin are listed in Hamm, C., ‘A Catalogue of Anonymous English Music in Fifteenth-century Continental Manuscripts’, Musica Disciplina, 22 (1968), pp. 47–76Google Scholar.

²¹ Hamm, , A Chronology of the Works of … Dufay, pp. 120–1Google Scholar.

²² The few dotted coronae which do occur in scribe A's work (for example, Tr87, fol. 158^r) are clearly in the hand of an assistant, since they are always accompanied by a change of ink and are invariably confined to pieces in which secondary scribal activity can be shown to have taken place.

²³ On the subject of scribal versatility see Van Dijk, S. J. P., ‘An Advertisement Sheet of an Early Fourteenth-century Writing Master at Oxford’, Scriptorium, 10 (1956), pp. 47–64CrossRefGoogle Scholar; and Mynors, R. A. B., ‘A Fifteenth-century Scribe: T. Werken’, Transactions of the Cambridge Bibliographical Society, 1 (1950), pp. 97–104Google Scholar.

²⁴ Such as has been possible with Ob 213, where certain letter- and note-forms undergo a gradual transformation in the course of a succession of gatherings. See Schoop, , Entstehung und Verwendung, pp. 21–2, 33–46Google Scholar.

²⁵ While for present purposes the term ‘layer’ is confined to discussion of the main scribe, it automatically includes the work of his assistants wherever they collaborated with him directly. The manner in which the script may change, even in the course of a single piece (for example, the text-hand of Tr87, fols. 27^v–29^v), is a warning against the dangers of imposing too rigid a system of classification upon scribe A's work. Some sections of the manuscript are less distinctive in character than others, and in such cases the definition of the scope of a layer may involve decisions of a more subjective nature.

²⁶ In the final column of Table 3 I have listed various isolated additions to the layers of the manuscript.

²⁷ Hamm, , A Chronology of the Works of…Dufay, pp. 86–7, 114–15Google Scholar.

²⁸ Hamm, ‘Manuscript Structure’, p. 171.

²⁹ In contrast to the usual solid type of initial. These were clearly supplied by scribe A: compare, for instance, the initials ‘C’ (Tr92, fol. 170^r) and ‘Q’ (Tr87, fol. 18^v) with their majuscule counterparts on Tr87, fol. 16^r and Tr87, fol. 14^r. From the many pen guides, and the presence of a number of redundant and incorrectly supplied letters, it can be seen that the initials were entered some time after the main contents had been copied. Furthermore, a number of offsets have occurred between gatherings that are not adjacent – Tr92, fol. 186^r, for example, contains a reverse image of the initials ‘A’ of fol. 146^v, staves ii–iii, while all three initials of Tr92, fol. 147^r have offset onto Tr92, fol. 177^r — thereby suggesting that the gatherings were unbound when initialled.

³⁰ Hamm, , A Chronology of the Works of … Dufay, pp. 86–7Google Scholar.

³¹ This work is ascribed to Binchois in Ob Mus.c.60. See, however, the arguments for English authorship put forward by Kovarik, E., ‘Mid fifteenth-century Polyphonic Elaborations of the Plainchant Ordinarium Missae’ (Ph.D. thesis, Harvard University, 1973), pp. 451–65Google Scholar.

³² Professor Julian Brown, who kindly discussed my work with me, suggested that this may be the hand of an ageing scribe.

³³ Hamm, , A Chronology of the Works of … Dufay, pp. 120–1Google Scholar. By citing instances such as these, I do not wish to imply that one should accept unquestioningly the dates proposed by Hamm; merely that the palaeographical evidence supports many of his groupings.

³⁴ This gathering also contains the anonymous Gentile alma benigna (fols. 161^v–162^r), the only setting of an Italian text in the entire manuscript.

³⁵ The possibility must be raised that these offsets occurred during restoration of the manuscript. However, a useful piece of evidence may be cited as an argument against this. At the foot of Tr92, fol. 198^r (part of the central opening of gathering 92v) is a reverse image of Tr92, fol. 188^v, staff viii of the previous gathering. Under ultra-violet light both words and music can be seen quite clearly; only the initial ‘T’ is missing. This suggests, therefore, that the image was formed before the initial was entered — in other words, during the copying of the layer in question.

³⁶ The first and last layers have been placed thus for reasons already given; elsewhere, when the evidence permits, I have endeavoured to group layers in a logical order.

³⁷ Gatherings 87i, 87xiv, and the insert to 87xvii are unsigned. 87ii (which is linked by a catchword to 87i) differs from the other signed gatherings of Tr87₁ in that its first six folios are simply signed ‘1’ to ‘6’, each number with a dot below it.

³⁸ The disruption of the intended gathering sequence could have occurred during a rebinding. Gathering 87ii contains two sets of sewing-holes, one of which corresponds to the single set of original holes found throughout the rest of the volume, while the other is an inversion of the first set, thus indicating that this gathering was once bound upside down.

³⁹ To judge by the signatures (a–l), this was: 87v, xi, xiii, vi, vii, xii, x, iii, ix, viii, iv. Bent, M., ‘Some Criteria for Establishing Relationships between Sources of Late-medieval Polyphony’, Music in Medieval and Early Modern Europe: Patronage, Sources and Texts, ed. Fenlon, I. (Cambridge, 1981), p. 303, n. 11Google Scholar, suggests that gatherings 87i–ii were later added as a unit. While concurring with this suggestion, I am unable to agree with Professor Bent's statement that ‘At some stage after the signatures were supplied, some compositions were copied across gathering joins in a manner which imposed some of the present deviations from the order originally intended.’ According to my own findings, 87i–ii are the only gatherings to be bridged by a piece.

⁴⁰ See Monson, C., ‘Stylistic Inconsistencies in a Kyrie Attributed to Dufay’, Journal of the American Musicological Society, 28 (1975), pp. 245–67CrossRefGoogle Scholar.

⁴¹ It is, however, worth noting the absence from the index of the three chansons on Tr92, fols. 180^r, 218^r and 227^v/226^r. Hamm, , A Chronology of the Works of…Dufay, p. 149Google Scholar, places the first of these, Dufay's Franc cuer gentil, after c. 1454 on the basis of its use of fusae, but his argument is invalidated by the fact that these note-values occur only in Esc iv.a.24, the latest of the vocal sources of the piece.

⁴² Adler, and Roller, , eds., Sechs Trienter Codices, p. xivGoogle Scholar, proposed — on the basis of the remark ‘Illud Kyrie pertinet ad Et in terra’ which scribe A appended to the Kyrie by Binchois that he copied into Tr87, fol. 56^v and followed with the first sixteen notes of the superius of the same composer's Gloria on Tr92, fols. 2^v–4^r — that gathering 87v and the first gathering of Tr92₁, must once have belonged together. There is no need to dwell on the untenability of this view, but the assumption that such an improbable link between two unrelated works was made at the scribe's initiative (that is, not inherited from his exemplar) needs to be challenged, especially in view of his apparent lack of interest in Tr92₁.

⁴³ Adler, and Koller, , eds., Sechs Trienter Codices, pp. xiv–xvGoogle Scholar, pointed this out without realising its full implications.

⁴⁴ See, for example, Strohm, R., ‘European Politics and the Distribution of Music in the Early Fifteenth Century’, Early Music History, 1 (1981), p. 320CrossRefGoogle Scholar; and Ward, ‘The Structure of…Trent 92–i’, p. 141. The earliest possible terminus is 1436, the date of Dufay's Nuper rosarum flores. This belongs to the main layer of Tr92₁ and is the latest datable work in the source.

⁴⁵ von Fischer, K., ‘Neue Quellen zur Musik des 13., 14. und 15. Jahrhunderts’, Acta Musicologica, 36 (1964), pp. 94–7CrossRefGoogle Scholar. An inventory of the manuscript is given.

⁴⁶ According to the librarian, Dr Johann Tommaschek, the leaves were discovered loose in a drawer. When I inspected the source in November 1980 they were still loose, and were badly in need of restoration and mounting.

⁴⁷ A number of words were, however, added in a greyer ink as scribe A revised the manuscript. These appear to be as follows: ‘Tenor laudamus te’ and ‘Contratenor laudamus te’ (upper half of fol. 77^r); ‘Amen’, ‘Tenor domine deus rex’ and ‘Tu solus altissimus’ (lower half of fol. 77^v).

⁴⁸ Note the precise correspondence in design between the initials ‘E’ on ZW, fol. 81^r and Tr92, fol. 150^v (the latter is no longer complete in the manuscript but is clear from microfilms made before its restoration in 1975).

⁴⁹ The method of preparing gatherings is essentially the same as in TR, the distinguishing features in ZW being the double marginal rulings in brown ink and the use of a dry-point base line to guide the text.

⁵⁰ TR contains several instances of the main scribe linking movements, sometimes erroneously or suspectly (for example, the index-paired Sanctus and Agnus settings on Tr92, fols. 220^v–226^r); but in contrast to ZW such pieces are always bound by physical proximity.

⁵¹ The ZW variants are listed in Reaney, G., ed., Early Fifteenth-century Music, Corpus Mensurabilis Musicae 11, iii (1966), p. xxivGoogle Scholar.

⁵² The monastery was razed to the ground by the Hussites in 1427 and any cultural life would have been prohibited there during the period of the manuscript's compilation. For details see Özelt, H., ‘Geschichte der Sängerknaben im Stifte Zwettl’, Jahresbericht 1959–60 des Bundes-Gymnasiums und -Realgymnasiums Krems (Krems, 1960), p. 4Google Scholar.

⁵³ Evidently the binder's pile contained a number of sheets of the dismembered choirbook, among which a single leaf and two double sheets happened to be adjacent. These were then halved and the two halves placed at each end of the book. From light offsetting and the pattern of worm-holes one can determine the original order of leaves in the bound volume. A good deal of indecipherable printed matter has offset onto the two half-pages (upper half of fol. 81^v and lower half of fol. 54^r) that would have been adjacent to the covers, though it is uncertain whether these images were transferred from the binding or from a previous book.

⁵⁴ The initial figure could at first sight be a ‘5’ but comparison with other specimens of Zelenka's handwriting confirms it as an idiosyncratic ‘3’.

⁵⁵ Trimming of the upper margin and traces of sewing-holes suggest that ZW was dismantled from a bound state. It could thus have remained intact until the early sixteenth century.

⁵⁶ See Ward, ‘The Structure of…Trent 92–i’, pp. 142–7; and Cobin, M., ‘The Compilation of the Aosta Manuscript: a Working Hypothesis’, Papers Read at the Dufay Quincentenary Conference, Brooklyn College, December 6–7, 1974, ed. Atlas, A. W. (New York, 1976), pp. 76–101Google Scholar.