Studies in Bibliography 56 (2003-2004) 195-242

We cannot verify your location (Log In)
Bibliographical Aspects of Italian Printed Music of the Sixteenth and Seventeenth Centuries
Stanley Boorman

The earliest printed music was presented in two different layouts on the page, both derived directly from those used in manuscripts. The first publisher of music (Ottaviano Petrucci, working in Venice in 1501) used both layouts: for many of his titles, he presented compositions in separate partbooks, so that each voice (or, later, instrument) had a separate book, much in the manner that modern orchestral musicians play from separate parts. For his successors, most books were in quarto or octavo, the choice depending on the repertoire presented. This was by far the commonest method of printing music, and will be the focus of the present study. At the same time, some prestigious volumes (of mass settings, in particular) were printed in what is called "choir-book" layout, with the music which each musician was to sing printed in a different part of the two pages of an opening. Later, and not before the seventeenth century, some music was printed in score, with all the parts synchronised and laid out one below another.

Books in choir-book layout or in score presented some problems for the compositor, mostly concerning casting-off and adjusting spacing to allow for different densities of notation for simultaneous music. These problems did not arise when working with partbooks, although the compositor would still have to adjust the spacing of the music to that of the text, or vice-versa. This was difficult only in multiple-impression printing (most frequently found in liturgical books), when music and text were printed at separate passes through the press.

However, partbooks for a single title do raise problems for the modern bibliographer, in deciding whether or not they are to be treated as separate units. 1 The easy solution would be to see them as parts of a larger work, akin to the volumes of a multi-volume prose work, even though, textually, they do not work in that manner. The contents of the different partbooks are to be "read", performed, simultaneously; one book is useless alone, for it contains only one voice-part for a multi-voice composition. Further, there is no evidence that single partbooks were ever intended to be sold alone, without the other books of a title. For many titles, the dedicatory letter or a publisher's letter to the reader appears in only one partbook of the set, and even an index may not be in all books—although that is often a result of a shortage [End Page 195] of space. Finally, many sets of parts are signed consecutively through the set, from one partbook to the next, and some are even foliated similarly. All these features suggest that the set of parts was seen by the publisher and user as a single unit, only complete when all parts were present.

On the other side, apparently suggesting that each partbook was thought of as a separate item, is the fact that, after the earliest years of the sixteenth century, all partbooks had full title-pages, with publication details. In practice, this was necessary for most users, for a set of parts was often not kept together. Instead, sets of Tenor books from a number of editions would be bound together—for they would be sung by the same person—and the corresponding Cantus or Bassus books similarly kept together. But this was a matter of convenience, not a reflection of the relationship between part and whole. It is perhaps analogous to the way in which collections of maps were made up, with a selection that met the needs of the owner.

If, as I suggest, musical partbooks are seen as strictly part of a single unit, the collation of a set of parts can also be seen as a unit, presented on one line. I believe that printers and publishers saw the books in that way, although they did not always arrange the parts in the same order—Cantus before Tenor before Altus before Bassus. The indication of which part is in a reader's hand does not appear on the direction line, but only on the title-page and headline: because of these features (discussed below), I suggest the addition of editorial indications of the part-names on the collation line— initial letters in brackets with a colon, so that they may not be confused with editorial signatures, as in the following examples. 2 This provides the evidence for printers following differing patterns of work, and even for stylistic change in the music itself.

It also avoids the extensive use of preceding superscripts to indicate repeated signatures: the part-name (abbreviated) is sufficient to distinguish signatures re-appearing in different partbooks, as in Example 43, below. Superscripts are reserved for repetitions within a single partbook.

The part-name indication is not always necessary in the continuation of a description. When each partbook in a set is signed differently, the signature itself identifies a folio absolutely: only when partbooks repeat signatures is it necessary to distinguish them, and I propose the simple solution given after Example 47.

The first three decades of the sixteenth century the period of Ottaviano Petrucci, Andrea Antico, and the Doricos, saw the gradual establishment of standardised procedures for music printing. The following period, roughly that of the second third of the sixteenth century, the period of Antonio Gardano and Girolamo Scotto, saw the appearance in Italy of single-impression type-faces, various techniques fairly stabilised, and methods of production well organised: some of these techniques, such as "vertical [End Page 196] composition", 3 seem to have been new, and a response to the special opportunities offered by repertoires such as madrigals and canzonets. But, by the last decade of the century, we enter a period in which many printers appear to have had difficulty in planning their books efficiently. To some extent, this is probably a result of the changes in style that emerge at the end of the sixteenth century. The appearance of monodic and continuo-supported solo song and motet, sometimes printed in score; the need to present different compositions within the same book in differing layouts on the page: all these helped to reduce the viability of the standard typesetting procedures that had prevailed in the 1550s to 1580s, and the problems were certainly compounded with the emergence of music printers who seem not to have had a systematic mind-set, nor the practical non-musical experience of earlier craftsmen.

Many printers therefore abandoned "vertical setting" entirely, and lost the benefits of consistency and clear organisation that went with it. Others adopted patterns of format and signing which are remarkably diverse, and (on occasion) seem to be almost haphazard. Not infrequently, we must find it difficult to conceive how the printer could have had a full volume of copy before him and yet have worked as he did. Indeed, the evidence often indicates some change of plan, such as a belated expansion in the content of a book. At the same time, however, it will usually tell us less about the music itself.

This article comprises a series of illustrations of printers' habits and bibliographical curiosities, drawn from Italian music printing of the sixteenth and seventeenth centuries. They are accompanied by suggestions as to how to describe the printed books bibliographically, and (where feasible) by commentary on their implications for the musical content and its organisation. As far as possible, the descriptive suggestions are built on the methodology outlined by Fredson Bowers, although some minor adaptations are proposed. 4 [End Page 197]

Standard abbreviations are used in what follows: A = Alto/Altus; B = Basso/Bassus: BC = Basso Continuo (or any of the related titles, for example Organo or Partimento); C = Canto/Cantus; T = Tenor/Tenore; 5 = Quinto/Quintus; 6 = Sesto/Sextus; etc. Books are given conventional titles, rather than formal transcriptions of their title-pages. Editions are identified by their sigla in the Répertoire international des sources musicales (hereafter RISM): entries with an initial letter followed by a number are found in Einzeldrucke vor 1800, ed. Karl-Heinz Schlager, RISM, Ser. A (Kassel: Bärenreiter, 1971- ); entries with an initial year followed by a superscript number appear in Recueils imprimés, XVIe-XVIIe siècles: liste chronologique, ed. François Lesure, RISM, Ser. B/I, i (Münich-Duisburg; Henle, 1960).

I. Signatures

Normal Patterns

The most common format for printed music produced during the mid-sixteenth century is well known. Works were printed in partbooks, in quarto (either portrait or landscape). Each partbook was distinguished from the others by two indications. The first was, of course, the part name—Tenor, Cantus Chorus Secundus, etc.—which normally appeared on the title-page, and also on at least one headline of each opening. The other comprised two elements, the system of signatures, and an additional phrase, usually an abbreviated title of the book, printed on first folios of most gatherings. 5 At other times, as in the case of some books printed by Petrucci, Pierre Attaingnant and Tylman Susato (among others), 6 this phrase was replaced by [End Page 198] a typographical emblem or sign, which would differ from those used in other partbooks or other titles. Taken together, these acted for the binder as a confirmation that he was collecting together only the sheets for that one title, and arranging them in the correct order. Normally, therefore, each partbook had a different sequence of signatures, often selected so that the whole series formed a continuous alphabetical sequence. The following examples start from the standard practice of the sixteenth century and diverge into a number of different situations and solutions, all within the basic concept of a sequence of signing letters uniting all the books of a set.

This was the most widespread pattern for much of the sixteenth century: the Cantus partbook was almost always signed first in sequence (whenever it was actually set in type), and the Tenor normally second. The Bassus usually appears fourth, after the Altus, and before any subsequent parts— Quintus, etc. This sequence reflected a traditional hierarchy of voices in composition and style, although that hierarchy was largely obsolete well before the middle of the sixteenth century.

The pattern would necessarily be modified whenever the music did not fit exactly into a sequence of standard quarto gatherings.

The principal parts for each of these books all contain 14 folios, though imposed differently. The Amadino edition supposes that Amadino had decided to have a six-folio last gathering before reaching the middle of gathering C and its equivalents. This was not necessary for Scotto. While Scotto [End Page 199] often followed the pattern of Example 3, other printers tended to adopt the other. But all are still following the normal sequence of letters, and signing the parts in the conventional order.

This conventional sequence of signatures seems to have been taken seriously during much of the sixteenth century: care is taken to ensure that all the letters are used, and in the correct order within books. In one case, the Gardano firm prepared a book in which the last partbook, the Sextus, apparently started two signature letters too late: a correction was made in the preceding part, by signing gathering T with "T.V.X." and "T.V.X.ij":

For the next example, the only surviving partbook has a similar signing: the last of five gatherings was signed with the double letters "EF". Presumably, the next partbook (probably the Tenor), was signed from the letter G.

This order of parts was not always maintained. Girolamo Scotto was one printer who varied the order of partbooks, even from early in his career. Particularly interesting for an analysis of his craftsmen's activities is the way in which different orders of signing seem to occur only for short periods. The pattern shown here, for example, can be found in a number of books published in 1549, with the Bassus signed first, followed by Cantus, Altus, Tenor and fifth or further books: 7

Scotto was certainly not the only printer to adopt different orders of books, though his examples are perhaps the most consistent within themselves.8 [End Page 200] 8 Other printers adopted various patterns at odd times, with no apparent reason:

Such patterns must say something about the order in which the partbooks were set for the press, but, in the absence of any other evidence, we can not speculate as to possible reasons for the unusual sequence—in Example 8 because the two choirs are not set and signed in the same order.

A different pattern was increasingly adopted during the seventeenth century, in which the Altus was signed before the Tenor:

It seems reasonable to suggest that this followed, with the delay traditional to craft skills, a recognition of the declining pre-eminence of the Tenor in the construction of sacred music.

It would be interesting to trace the stages at which different publishers adopted this arrangement as normal practice. The rather random evidence that I have collected so far suggests that Filippo Lomazzo in Milan was among the first, for the pattern appears in editions he put out with the Tini heirs in 1607 (B. Regio: I Missarum ac sacrarum cantionum à 5, 8, RISM R725) and 1608 (Baglioni: I Sacrarum Cantionum, Op. 2, RISM B644) and [End Page 201] it continues throughout his career: he was followed by Rolla (although with a few exceptions, such as Levi's I Salmi di Terza of 1647). Roman printers soon followed suit, for the Altus is signed second in books by Bartolomeo Zannetti in 1614 (Selectae Cantiones, RISM 16143), by Luc' Antonio Soldi from 1620 (Tarditi, II Psalmi [etc.], RISM T225), by Giovanni Battista Robletti from 1622 (Nenna: Sacrae hebdomodae responsoria à 5, RISM N381), though not consistently, and by Paolo Masotti from his earliest editions in 1626 (Talone, Armonicus parnasus à 3-4, Op. 1, RISM T62). Lodovico Grignani was using this arrangement by 1647. On the other hand, the two principal publishing families in Venice, the Magni and Vincenti, both preserved the traditional arrangement for much longer: for Vincenti, the change occurred during the years 1649 and 1650. In 1649 his edition of Capuana's Motets, Op. 3 (RISM R952) uses the new sequence of parts: although this rapidly became the norm, at least one edition of 1650 retained the old order. For Magni, the change seems to have come even later, perhaps as late as 1660. Bolognese publishers had certainly adopted the new sequence by then: indeed Cazzati takes the practice somewhat further, in that solo instrumental parts are signed first, even before the principal vocal parts: for his Messa e salmi à 5, Op. 36 (Bologna: M. Silvani, 1665), the violin parts are signed with A and B, and the Alto Viola with C; solo vocal parts use D-H and the Ripieno I-L, with the two continuo parts signed M and N. The same pattern can be found in his later books, though not consistently: Le quattro Antifone annuale, Op. 42 (Bologna: s.n., 1667) has the single vocal part signed with C, with the two violin parts using A and B and the Organ part D. However, his Salmi brevi à 4, Op. 58 signs both choirs of singers before any instruments.

If the set of books used more than 23 gatherings, the series could be extended with a few conventional signs—an asterisk, an obelus, even an ampersand:

If several gatherings were involved, a second alphabet of signatures would become essential. Ideally (and normally during the sixteenth century), this second alphabet would be distinguished from the first by some simple typographical device (perhaps an asterisk following the letter), or, for example, with the use of doubled letters:[End Page 202]

This pattern of four gatherings per partbook, each with four folios in portrait quarto, is surprisingly common, given the wide variety of styles that were included in printed books. It is as if the printers and publishers felt that 16 folios contained as much music (with title, table of contents, and perhaps dedication) as the average purchaser was willing to pay for, or perform. It is even to be found fairly frequently in music for two choirs: apart from Examples 9 and 12, see also Lambardi's Psalmodia Vespertina II of 1605 (printed by the Cenobio Santo Spirito in Venice) or Sorte's Vespertina (Venice: Angelo Gardano, 1593), among many others.

The use of small gatherings, usually in quarto, would continue for many years. But already before 1600, Vincenti was printing in longer gatherings, and signing them as in the next examples:

Such large gatherings could raise a problem for the print-shop: the house foreman had to be able to cast off copy fairly accurately, so that a compositor would be sure of the size of a gathering and sheets could be set and printed without any fear of miscalculation. Sometimes, things did not quite work out: the compositor might decide too soon that he had reached the midpoint of the gathering, and start to complete the central formes. The result could then be a need for a short additional gathering at the end of a part-book:

The first gathering of the Cantus is signed normally, from A2, with arabic numerals. The second gathering is signed on the first sheet with the phrase "A in fine" and on the second with "A2". After 32 folios, the compositor was apparently confident that he was about at the middle of his material, but on reaching the end, he found that there was a little more, [End Page 203] necessitating a supplementary gathering. A similar miscalculation marks all the partbooks, and is reflected in the organ book, where the second gathering has to be slightly larger than the first. 9

In this book, it seems that the compositor for the Tenor book had learned from the experience with the Cantus or Altus book.

Some editions appear to show a compromise solution, in which two signature letters were allocated to each partbook: the second could act to collect together whatever was left after the first had been set up and printed. This probably explains the irregular structure of gatherings in the following example:

For some repertories, given the manner in which they were presented, this may have been almost the only possible solution. The many "scores" for solo madrigals and motets, and so on, printed in score layout in a single book, provide an excellent example. The very diverse rhythmic structure of these works, not to mention the presence of recitative sections, produced different amounts of music to the page: as a result, it would be much more difficult to cast off the music in advance. Only a very experienced musicprinting shop could have divided such a book between equal gatherings, or between two compositors. A second result, for long gatherings, is that we can normally assume the presence of many more sorts in the case for a music fount, whenever the musical text could not be cast off in advance.

Added Material

The following sequence is clearly different, for the conventional signs do not follow on from the end of a complete sequence of 23 letters, but are found in gatherings added at the end of each partbook:

The placing of the conventional Maltese crosses to sign an additional bifolium at the end of each part is significant. The book was apparently planned to have five gatherings in each part, and signatures were assigned accordingly. Either Scotto miscalculated, and had to add pages at the end of each part, or else the editor produced some additional music after printing was well under way. But the book was set by "vertical setting", so that the Tenor gathering F, the Alto L and the Bassus Q were prepared with the Cantus A, and before any other gatherings. Thus is would not have been possible, if adding another gathering at the end of each partbook, to sign it with the next available letter; for example, the sixth Cantus gathering could not have been signed with the letter F without causing potential confusion. Even so, it is significant that the additional pages were not merely added to the last lettered gathering of each book, to make six folios. 10 This suggests that the change came late in the printing process, after much of the work on the last lettered gatherings (E, K and P) had been completed, after the midpoint of those gatherings. Given that the last pieces were additional to the contents found in earlier editions, the late decision to include this new music is the likely explanation. 11

The same effect can be seen in Rinaldo da Montagnana's first book of five-voiced motets (Gardano 1563), with a Maltese cross for the last gathering of each part, although in other ways that is a normal gathering, with four folios. Once again, it looks as though further pieces were added at a late point in preparing the edition. The last four works, occupying six pages, are for six voices, and the first three of them are by a different composer. It seems reasonable to argue that someone other than the publisher was financially responsible for the book, either paying for it directly, or undertaking to take a large number of copies. This is the easiest explanation of the [End Page 205] willingness of the publisher to add a whole gathering to each partbook, with the extra costs of labour and (especially) paper, late in the process.

A similar phenomenon occurs in Scotto's edition of the combined first and second books of madrigals by Verdelot, published in 1540:

Here again, we might assume that the short final gathering merely included the few madrigals that were left over and to be included after the completion of gathering G—especially since the edition represents a reprinting of earlier books. But the evidence suggests otherwise: at the foot of G4v in all partbooks is the word FINIS. Gathering H contains three additional madrigals, attributed to Willaert, Verdelot and Barre, none of which had appeared in earlier editions of either book. 12 Given that Scotto had printed a five-voiced anthology in the same year, in which Willaert and his "discipulo" Leonardo Barre were highly featured on the title-page and were first in the contents, 13 it seems likely that one of these two composers (probably the younger and less well-known Barre) requested the addition to the present volume.

A more complex example of the same situation is the following edition, which started out with partbooks planned to be three gatherings long. Apparently a whole gathering of music was added at some stage. The printer could well have chosen to sign the new gathering according to Example 18, above. The solution actually adopted must have raised questions in the minds of booksellers, binders or others faced with unbound sheets.

In this example, two different methods of coping with the problem of pre-assigned signatures were used. Cantus and Altus merely employ the next letter (even though that means a duplication of signatures D and K): Tenor and Bassus use a duplication of the final letter (even though that was not necessary for the Bassus). The implication is probably that two different compositors were involved: there is evidence that, even as late as this volume, compositors had some freedom in presentation, not merely of the verbal text (as commonly elsewhere), but also of the musical notation. Therefore, volumes with this sort of pattern should be examined for evidence of the [End Page 206] compositors reaching slightly different editorial decisions both in layout and in details of musical notation.

An example such as the following might also suggest that two craftsmen were involved; but further examination indicates clearly that there was a change of plan before the book went on sale:

I have arranged the parts in this tabular fashion, in order to highlight several significant features. The most obvious is the use of the two sets of signatures for each of the first eight partbooks, representing two four-voiced choirs. The ordinary run of signatures for the opening gatherings of these books reached into a second alphabet by the end of the second Bassus, with the entirely conventional use of Aa. These gatherings carry a standard presentation and signing for a book of eight-voiced motets, apparently intended to be complete and self-contained. Folio Aalr has the word "finis" after the signature: this reflects a common practice in Gardano's shop, indicating the last gathering of a set of partbooks. In addition, an index of the preceding pieces is found on folio 6r of each third gathering of these eight partbooks. After the book was completed in this form, it was expanded to include additional pieces scored for more than eight voices. The additional signatures for these works in the first eight partbooks could not use the pattern "Aa", which had been started in the second Bassus: that was more logically used for the additional partbooks for voices 9-12. For the added gatherings of the original partbooks, a new style had to be adopted, and one with double capital letters, "AA", etc., was employed. 14

Miscalculations

In many volumes (principally those not containing madrigals or other secular vocal music), the various partbooks had to be of different sizes, to contain music of differing complexity in the various parts. This would not normally cause a problem in planning the signatures:

[End Page 207]

In this case, a compositor or his foreman realised that all partbooks would need at least part of a fourth gathering, and planned accordingly. It does not matter which voice-part he used to make the calculation, for all would have yielded the same conclusion. Occasionally, however, the lengths of the various parts might differ significantly. A competent compositor (or house editor) could plan ahead, and assign signatures accordingly, before work started:

In some cases, the printer seems not to have been able to foresee this need, and was faced, late in the process, with inserting an extra gathering at the end of one or more of the partbooks:

Here, the original plan had evidently been to have three gatherings for each partbook, though in fact three of the parts required more space, only the Alto conforming to pattern. For the Basso, which was last in sequence, there was no problem in assigning a signature: the additional fourth gathering could be signed with the next letter in the alphabet. For the Canto and Tenore, however, this would have resulted in duplicated letters, and conventional signs were used instead. 15

When an additional gathering had to be inserted in this manner, an alternative method (and one perhaps of more use to the binder) was to use a different form, or a double statement, of the immediately preceding signing letter, as for the Altus of the following example:[End Page 208]

This example does indeed look as though it represents a miscalculation in the printing shop: but in many cases, we can not tell whether the signatures indicate a change of plan (for example, involving the addition of more compositions) or a miscalculation:

In both these cases, I suspect, there was a slip on the part of the printer, who had apparently allocated four gatherings (and signature letters) to each voice part. In the first case, he had discovered the problem before the Quinta Pars was set, so that the Bassus could have a normal fifth signature. In the second, it looks as though Gardano planned the book on the basis of the Tenor part. All the other parts have an extra gathering, and all except the Sesto employ a repeat signature. 17

On occasion a section of music is omitted. This could occur because the compositor made a simple mistake, perhaps a haplograph, or because the imposition of the pages in the forme omitted one or two complete pages. The latter seems to be the likely cause of the following:

The folio after C4 is a single leaf containing omitted material. The leaf has a line reading:] Questa Carta Va posta drieto il numero 22 in Cantus. 18

Unsigned Gatherings

Many music books, even as early as from the press of Ottaviano Petrucci, used the same title-page for each part, with the minimal adjustment of changing the part-name. This was eminently practical, saving not simply the labour of setting up the type each time, but also the more time-consuming work of centering each line, and arranging the whole into a standardised, possibly pleasing ensemble. By early in the seventeenth century, this practice was sometimes extended from a simple title-page to a complete bifolio, with title on the first recto, and perhaps a dedication, a Letter to the Reader, or a table of contents on the other three pages. Very frequently, these two folios, which needed no change except for the part-name, could be printed as a separate gathering. Then, this initial gathering need not be signed, so that even the signature would not need to be changed.

In the next example, Muti's compositor allowed for the unsigned gatherings of lower partbooks in his signing pattern:

In this case, it is certain that two craftsmen were involved, and that there was little contact between them. One man set the Cantus, Tenor and Organ books, using the common pattern in which each partbook was signed sequentially starting with A: it is interesting, too, that he seems to have intended to use a different form of the letter for each book, in the manner shown in Examples 34-36. As was customary, the Organ was signed as if independent of the rest of the parts. The second man set the Altus and Bassus according to the pattern I have just been describing, and assumed (without checking) that Cantus and Tenor would be signed A and B respectively, therefore beginning the Altus with the letter C.

The following book shows a common extension of this seventeenth-century practice of an introductory gathering, one that implies a certain lavishness in presentation and the use of white space, a lavishness often at odds with the poor quality of type, paper, and workmanship. 19

This was obviously laid out so that it could be used in every partbook, with only the part-name changed. Despite the extravagant use of paper, a large number of seventeenth-century musical books seem to have had half-titles, and quite a few also had a blank folio, preceding that or the title. In many cases, of course, these blank folios have subsequently become lost, or (more probably) used for some other purpose. Yet it is an aspect of describing these books that we determine whether what looks like a fly-leaf was not in fact a blank first folio of the book itself.

Diverse Styles in the Same Edition

Most of the above examples have started from what was by far the most common pattern, that of a sequential alphabet, followed through all the gatherings of each partbook. Among the possible alternatives, two were frequently used: one involved using some form of conventional sign for each book (as is to be found in the work of Susato, or of Sengenwald in Jena); 20 the other involved different forms of the alphabetical sequence, with each book starting with some form of "A":

A number of other editions printed by Scotto in his early years follow a similar practice. An interesting example can be found in his 1540 edition of masses by Morales, Gombert and Jachet: 21

This pattern of roman and italic letters was employed in Scotto's house with fair consistency—and a number of exceptions—for two years, after which a different pattern, in which the Cantus used lower-case letters in italic or gothic characters (as in RISM G269=15434), was adopted alongside it.

Other possibilities involved using a mix of roman and italic initials, capitals and lower case letters, or capitals of different sizes in the same signature:

Scotto was by no means the only printer to do this on a regular basis: others did similar things, often simply increasing the number of initials from one part to the next.

Here is one place where the musical bibliographer needs to modify the practices proposed by Bowers: his general recommendation is that, whenever letters are doubled or tripled (AA or AAA), the bibliographer should use the formulation 2A or 3A to indicate the pattern. Yet clearly such a habit cannot distinguish between the signatures of the Tenor and Altus of Example 39, or indicate the style of letters used in the volume of Villotte (Example 36). To write "2a" in Example 37 may not raise the question of whether one is referring to the Bassus or the Quintus, but to write "2A" certainly does. It seems to me that there is no choice but to spell out all the sequences as they appear in the book, and as I have done above. At the same time, and for the same reason, the collation should normally distinguish between signature letters in roman or italic or boldface, just as it does between capitals and lower-case letters.

Similar patterns of change can occur in a single partbook, particularly when each book was assigned only one letter:

Sometimes, the mix of letters and letter styles seems to be no more than a product of caprice on the part of the typesetter:

There is little else that can be done here, but list the signatures as they actually appear. Not surprisingly, there are also errors in the signing of this edition: for example, the second folio of gathering B of the Tenor has the signature "Cc*2", presumably left in the forme after the Quintus was completed. [End Page 214]

Deliberate Repetition

The need to distinguish the different forms of repeated letters is not the only place where following Bowers' (now traditional) procedures produces problems when handling partbooks. One of the simpler is where each partbook has exactly the same form of signatures:

This pattern was also used when some of the parts were of different lengths, sometimes (as in Example 46) reflecting the arrangement of increasing scoring in volumes:

Clearly, in such cases, no reference can be adequately made to the content or bibliography of an individual partbook without prefacing the citation with the name or initial of the partbook involved. It is perhaps a nuisance, though nothing more, to have to do this: but a form such as "C1:C2v" is not particularly long-winded, is essentially clear and is still easy to read. 23

I have already mentioned the (largely seventeenth-century) practice of printing with large gatherings, one per partbook, and with one signature assigned to each book. As with the earlier examples given here, this is often very straightforward. But there are many instances where the principle of [End Page 215] using one letter per partbook clashed with a desire to print in separate quarto gatherings. An amusing example of this clash between two operating principles is the following:

Here, the original intention was apparently to assign one letter to each partbook, in sequence from "A" to "O". However, any part that is over ten folios in length was printed in separate quarto gatherings: these, the principal vocal parts and the two continuo parts (Organo and Violone), were then each signed in a traditional sequential pattern. As a result, there are, for example, five gatherings signed with the letter "F". It may be that two compositors were involved, one assigned to these parts, and the other to the remainder. However, I suspect that further research on editions signed by Dozza will reveal a reluctance to have long single gatherings.

More difficult to describe in a modern bibliography is the practice of retaining the same initial for all the gaterings in a partbook, but numbering the signed folios consecutively. 24 Given the practice of signing to the mid-point of each gathering (in which the second half of a gathering was not included in the count) the following signature pattern would be found in a quarto book of 16 folios:

folio 1 2 3 4
gathering 1 A2
gathering 2 A3 A4
gathering 3 A5 A6
gathering 4 A7 A8

When the gatherings are not all of the same size, as often happens, the patterns can seem even more complicated—although, in practice, they do reveal the gathering structure very quickly:

It is evident that the leaf signed "A3" in this example is not the third folio of the first gathering, following A2, as one would normally expect: similarly, "A4" would normally refer to the last folio of the same gathering, whereas the printer has used it to designate the sixth folio of the book. In effect, though, the printer's method is entirely logical, for "A4" represents the fourth signed leaf, but it makes a mess of all modern practice of description.25 25

We have to distinguish these gatherings in the collation, and also to indicate the manner of signing: one can not merely write A4 A4 A8, for there is no way of distinguishing between the first two gatherings, and the presumption would be that each carried a signature of "A2" on the second folio: nor can one write A14 A34 A58, for this implies that the third gathering is signed "A5" throughout, 26 and the other gatherings similarly.

I have already used the convention proposed by Bowers, following on from McKerrow, in which preceding superscripts in the collational formula indicate the repetition of a signing letter or sequence of letters. 27 For him, a book could be described as A-Z4, 2A-Q4 (note the essential use of the comma), indicating a series of 23 gatherings signed from A to Z, followed by 16 more signed in an identical manner. For such a book, he proposed the shorthand version A-2Q4. This notation presents certain drawbacks, more apparent in music printing than elsewhere, for the implication is that all the signature letters between A and the second Q are present, a total of 39 gatherings. However, this is rare in musical editions. Let us suppose a music partbook of four gatherings in quarto, all signed with A. To write A-4A4 would imply three complete alphabets of letters, followed by a fourth appearance of the signature A, a total of 70 gatherings. Similarly, to write A4 2A4 3A4 4A4 or 1-4A4 implies gatherings signed A, AA, AAA and AAAA. The musical bibliographer would therefore have to describe the book as A4, 2A4, 3A4, 4A4; this [End Page 217] could be abbreviated to 1-4A4, even thought this does not indicate the actual signing as clearly as traditional formulae do.

In cases of this sort, where this may be the most convenient formulation, the signature pattern could then be shown in a normal manner. Cazzati's book (Example 49) would be described as follows:

This may seem clumsy, but it has an advantage over all other possibilities, in being precise, and in allowing the bibliographer to understand how the book was constructed (as well as signed).

In practice, musical bibliographers could perhaps adopt a short-hand, using the phrase "numbered sequentially through the part" to indicate this practice of numbering each gathering from where the previous one ended. This would save us having to indicate in detail the numbering pattern in many music books of the seventeenth century, an area of research which is fortunately not yet cluttered with other solutions.

This phrase "numbered sequentially through each partbook" would then imply the pattern of signatures shown here: 29

Cantus: folio 1 2 3 4 5 6
gathering 1
gathering 2 A1 A2
gathering 3 A3 A4
gathering 4 A5 A6
gathering 5 A7 A8 A9
Tenor:
gathering 1
gathering 2 A1 A2 [End Page 218]
gathering 3 A3 A4
gathering 4 A5 A6
gathering 5 A7 A8
gathering 6 A9 A10 A11

[and similarly for the other books].

The practice of repeating a signature letter throughout a book becomes a widespread manner of printing partbooks during the seventeenth century, and I have found examples in the work of Camagno, Rolla, Sala and Tradate, among others. It seems to have been particularly employed by printers who specialised in music for many parts. However, there were obviously likely to be other occasions on which it was of use.

An interesting example concerns the Armonici Entusiasmi, Op. 9 of G. B. Bassani, printed by Sala in 1690 (RISM B1186). Each part is signed in this manner, but the choice of signatures suggests a lack of communication between two craftsmen. The four parts of the solo choir are signed with A-D, as are the four of the Ripieno choir: the two violin parts are signed with I and K, and the Organ with M, evidently following a presumption that the Ripieno would be signed with E-H. The Violone (which presumably should have been signed with L) has G as its signature, perhaps assuming that the two violin parts would be signed with E and F. Perhaps one craftsman set solo A-D and I, K and M, while the other was responsible for the Ripieno A-D and G.

There are rare instances in the sixteenth century, where a signature letter is repeated as a result of a change of plan or a miscalculation, 30 and the second gathering is signed according to the pattern of Examples 49-51. The next two examples seem to me to be cases where material was added to the volume during the process of printing. In the first, I do not believe that Gardano, with over twenty years of experience, would have failed to recognise a four-gathering collection, as opposed to a three-gathering one:

The signature patterns described by this are illustrated here with the Tenor partbook:[End Page 219]

gathering folios: 1 2 3 4
D Dij
E E Eij
F F Fij
2F F3 F4
* * *2

I find it impossible to argue that the need to add two full gatherings of music at the end of each partbook, after only three signed normally, was the result of miscalculation. Instead, I assume that the music that begins at the end of the third gathering was a later addition. This comprises works for nine or more voices, while all the earlier pieces were composed for seven or eight voices. The volume is entitled Libro Primo, although no volume 2 seems to have appeared. I suspect that volume 1 was originally intended to contain only the pieces for seven or eight voices, and volume 2 would have contained the larger scorings. At a late stage in the production of the volume (after the signatures had been planned, the title-page printed, and perhaps also several other gatherings finished), these were added to the present book, possibly because there were not enough works to justify a separate title. After all, the pieces for 9-12 voices take up only a little over two gatherings.

One other feature of the book's production is interesting. For Choir 2, the Alto is signed after the Basso. This allows the additional gatherings in the Alto to be signed with continuing letters, rather than with repetitions of the third letter of the book. (Much the same pattern can be found elsewhere, in, for example, Falcidio's first book of masses—RISM F68, or Asola's Missae tres—RISM A2505, both printed by Gardano's sons in 1570.)[End Page 220]

This signing pattern can also be extended to longer gatherings, each comprising more than one sheet of paper. The following example exactly follows the pattern, although the results look even more bizarre:

This manner of describing the signing patterns does make it more clumsy to indicate errors: here, for the vocal parts it is necessary to show not only the error and its position, but also the "correct" version, since that will not always be immediately obvious to the reader.

Unfortunately, the emergence of two new features mentioned here—a more lavish display of the title and dedication (involving a separate gathering and sometimes blank pages), and the retention of the same signature letter for a whole book—did not happen simply and cleanly. Thus there are many books where the patterns are not consistent, or where we can see intermediate stages in the process:

The first two folios in all partbooks carry the same material: title-page, a blank page, a dedication, and a letter A chi vorrà servirsi deli presenti concerti, all of which could be left standing for each book, with the simple change of the part-name on the first recto. All voices except Canto Primo have a short first gathering, merely a bifolio, to carry this material: all also follow the modified signing pattern that I have been describing above, with the exception of a single error in Canto Secondo. The Canto Primo presents all the material in one gathering, but signs the folios as if in two gatherings, and following a more traditional pattern of beginning each from 1:

[C1:] A, -, A1, A2, A3, A4, A5, A6, A7, A8, [10 unsigned]
[C2:] B, -;
B3, B3 [recte B4], B5, B6, B7, B8, B9, B10, [8 unsigned]
[A:] C, -;
C3, C4, C5, C6, C7, C8, [6 unsigned]
[B:] D, -;
D3, D4, D5, D6, D7, D8, [6 unsigned]
[Organo:] E, -;
E3, E4, E5, E6 [4 unsigned]
[F and G signed normally]

The following is a more complex illustration of the same point, one in which we can safely assert that two craftsmen were involved, one preparing only five partbooks, and the second preparing the other eight. I have arranged the parts to highlight the pattern:

There is no easy way to describe the signature patterns of the first of these two examples or to lay out the collational patterns of the second. Yet the details are important, and not only for the bibliographer and printing historian. The musical scholar is given clear evidence of exactly those details [End Page 222] of presentation which (in manuscript study) are frequently indicative of differences in presentation or indeed content.

Continuo Parts

With the emergence of basso continuo, the continuo part was often signed last in sequence, as in Examples 9 and 10. (In this context, the placing of the organ in the sequence of signatures in Example 8 is of interest.) However, increasingly, it would be signed from A, independently of the signing of the partbooks, as in Example 54, or the following:

Sometimes this is simple to handle, for the book can be included in a general description, using the processes employed above: entries such as "A:C2v" or "Part:C2v" are enough to indicate which page is intended. But there are many cases where the Continuo book not only is signed with letters used elsewhere, but is also different in other ways. Perhaps it is in folio while the other parts are in quarto; perhaps it has a different treatment of the direction line; probably it has a different style of pagination; certainly it is likely to use a different manner of treating the textual incipits, the dedication, the attributions, details of scoring, and so on. Here is an example:

There are enough differences in treatment that the organ part needs to be listed and discussed separately. The next example is slightly more complicated, though in fact more typical of many volumes from the seventeenth century:

A similar pattern of change can occur between vocal and instrumental parts. I have already cited Cazzati's Messa e Salmi per li defonti . . . Op. 31 (Example 48) for its unusual pattern of signatures. But is also relevant here. While the principal voices and the continuo parts were all printed with normal quarto gatherings, the ripieno voice and instrumental parts were each printed in one large gathering—still in quarto format, but with more sheets to the gathering. This may be another example of work being divided between craftsmen: it is likely though that the simpler nature of the ripieno and instrumental part-writing allowed a house foreman to cast off copy more accurately.

Many of Cazzati's editions from Bolognese presses (not least those where no printer is named) contain bibliographical anomalies of one sort or another. While some may reflect differing patterns of behaviour within the printing shops, others suggest that Cazzati's music was not always provided in a straightforward or timely manner, resulting in changes and problems in planning the printing of the edition. This lends some credence to the idea that Cazzati was himself sponsoring many of these editions, and arranging for their printing. [End Page 224]

Anomalous Signings

While several gatherings can have the same signature, the reverse can also happen, or at least appear to happen, with two sets of signatures in one gathering. One example, not very confusing, has already been mentioned, with the signing of the Cantus Primus in Example 55. A very similar pattern appears in another Roman edition:

Given the signatures, one would expect the book to be constructed as a2 A18. However, it is clearly one continuous gathering of ten bifolios. There are few plausible reasons why a single gathering should follow this pattern. One, and the best, may be simply that the first bifolium, containing only preliminary matter, was originally to have been a free-standing bifolium. In this case, it may have been incorporated into the gathering so that it could be conjugate with added folios at the back. This is to some extent supported by the contents as listed at the end of the volume. After fourteen compositions ascribed to various Roman composers, which end on the verso of folio 18, there is added a setting of Quem vidistis pastores, which is ascribed as a Melos Rusticum. If Sylvestris had wanted to end with the previous piece, the printer would have been left with no room for a table of contents. It seems possible therefore that the original preliminary bifolium was converted at this stage into four folios, wrapped around the rest, initially to contain the table. As a result, Sylvestris added one more piece, this one anomalous in that it is not collected Ab Excellentissimis Musices Auctoribus, as the title-page says (with headlines naming the composers), but is anonymous.

There are other problems with signatures and signing patterns, not all of them related to the collation. It is notable how many of the following instances, and also how many of the earlier problem cases, come from printers working in Rome.

Some printers signed each sheet, rather than each folio, in the first half of each gathering. In these cases, the gathering structure, and hence the collation is likely to be straightforward:

Here, the plan was apparently to sign each partbook with the same letter, and number each sheet consecutively. This pattern was not followed very successfully although the intention is clear. The only completely systematic part is the Canto Primo: the Tenore and the Continuo books are correct, although the actual signing is anomalous, while the other three books contain errors. The evidence is as if the compositor was more used to signing folios and kept forgetting that he had been instructed to sign each sheet only once.

If this example showed signing by sheets with the same signature letter in each gathering, another Roman, Paolo Masotti, could occasionally sign by the sheet, and give each sheet a different letter, even when two or more were in the same gathering:

There are problems here with creating a collational formula, and also with referring to individual folios. The collation probably has to be written with editorial symbols for each gathering (as I have done). It would not be possible to describe the Canto as A-D16, for that would imply four large gatherings. The only viable alternative is to call the Canto A16 (signed A-D), the Tenor E16 (signed E-H), etc., which is itself clumsy. It seems clearer to give a separate list of signings and as simple as possible a collation line, using editorial collational signs (here roman numerals).

In the same way, there is no easy way to use the signature letters to refer to individual folios: the fourth folio is part of the sheet B (B2), and so are the thirteenth and fourteenth (which are actually B3 and B4), while the fifteenth is part of A (A3)! Again, the editorial signing (the roman numerals) allows one to cite folios correctly, with or without adding the signatures found on the page. In effect the book is best treated as though it were unsigned. [End Page 226] Books like this provide an exception to my suggestion (presented below) that signatures, rather than pagination, should normally be used for citation. Fortunately, it is also one where the pagination is correct throughout.

Faulty Signings

Frequently a printer did not sign a folio or two which would normally have been signed. In most cases, especially when the title-page is involved, he allowed for it in his numbering system, for this was important for his own methodology as much as for the binder. Then, the modern scholar can assign the number editorially, in the customary manner. Sometimes, however, the folio is not allowed for in the sequence of numbers, and here one has to examine the source more carefully.

A1r contains a title, A1v is blank, A2r contains the dedication, and the music begins on A2v. There are several possible explanations for this signing pattern. One is that the printer originally intended the first bifolio to be separate, but found that he had more music than would fit into the resulting smaller volume. The first bifolio was then converted into two bifolios wrapped around the rest of the book. 32 An alternative, and more probable, suggestion is that the music was always intended to start on A1v, so that the present folio signed A2 would indeed be the second of the book. The decision to insert the Dedicatory folio would then have been made after typesetting began, but presumably before reaching the 8th folio of music.

Fairly often, however, the "extra" folio is not matched by a conjugate in the other half of the gathering, and is truly surplus. This folio may then have been tipped in:[End Page 227]

There is another possible solution, especially during the seventeenth century. When the unsigned, extra sheet is a title-page or a half-title, one should suspect the loss of a blank first leaf: the original may have followed the pattern of Example 33, above.

More extreme is the practice of publishing a title without any signatures at all. For books like this, Bowers 33 argues for the use of a greek pi or chi for all unsigned gatherings: and I have used pi regularly in earlier examples. Normally, of course, he is assuming that the unsigned gatherings are part of a book which also contains signatures. There are cases, however, such as the following, where each partbook is completely unsigned. It seems more sensible to assign editorial collational signatures, using either letters or roman numerals within brackets:

This also avoids a further complication in formulation if Bowers' formulae were followed: the Continuo part would have been described as π6 π8. While Bowers suggests the use of arabic numerals for sequential unsigned gatherings, this seems unsatisfactory for music editions, where arabic numerals are on occasion used for the signatures themselves.

Finally, all printers may at times make errors or produce structures and signing patterns which are certainly erroneous. In some cases, the signatures are clearly correctable, and probably indicate something of the procedures in the printing shop.

The book of Messa e Salmi, Corr. Op. 1 by Cozzi published in 1649 (Example 54) has a clear indication that the Tenor book was prepared after the Altus: two folios in the Tenor book, C5 and C13 (that is, at the beginning of the 2nd and 4th gatherings) are actually signed B5 and B13. In the [End Page 228] case of Has quatuor missas (Example 8), there are three wrong signings in gathering B, the Altus book: the three folios B3-5 are all signed with the letter A, and corrected in manuscript (in the same hand) in both the Bologna and Lucca copies. This must surely imply that these two sheets were prepared after those of the Cantus part, and presumably placed in the same formes. In the same way, the first three folios of the Bassus of the Quatto Libri delle Villotte (Example 26) are signed with the signature from the Tenor book. There is a surprisingly large number of such errors, many of which are most easily explained as representing the retention of a signature in a forme during the insertion of material from a different gathering.

However, we can not make this assumption as a matter of course. There are occasions when it raises impossible solutions:

It is wishful thinking to propose a sequence of preparing the partbooks on the assumption that so many signatures were erroneously retained in the forme. Such an interpretation suggests that gathering B was prepared before E, and that before C; also that gathering D was earlier than F, itself earlier than G. But then G appears to be earlier than D, and both propositions are clearly not tenable.

This seems a relatively simple error: the most likely explanation is that the two half-sheets (A9-10 and B9-10) were set up in the same forme. Indeed most signing errors seem to be explicable through an understanding of printing-house procedures. But there are some printers, among them Soldi, who seem more prone to nonsensical patterns:

II. Pagination and Piece Numbering

The signature pattern has implications for unsigned sheets in ordinary gatherings, whereby bibliographers can refer to the second half of a gathering in terms of its "signature". This has produced a simple and easy means, even in most of the books discussed above, for referring to any folio or page in a printed book. Even if a book, such as that in Example 1, is only signed to the second folio in every gathering, a reference to E4r is clear.

However, the second means of referring to each page, a sequence of page numbers, is unfortunately often less reliable than the folio numbers. One reason is that page numbers do not refer in any way to the physical structure of the book, as folio numbers do (or should do). As a result, there is little need for any consistent pattern, or for any real attempt at accuracy. The numbers are unlikely to be used by the compositor or pressman to check for accurate imposition in the forme, nor are they likely to be useful to the binder (especially since many will be hidden within folds of the paper before it is trimmed).

The result is not only that pagination patterns could often be erroneous and the printer feel no need to correct them, but also that a number of different patterns for numbering (not always by pages) could co-exist, even in one printing shop, sometimes in one volume.

Among them are the following:

(i) Pagination from the beginning of the musical content, beginning with number 1, often on folio 2r. This follows normal book-printing practice, in which the preliminary matter lay on pages without numbers, or with a different sequence of numbers:

(ii) Pagination from the start of the musical content, but numbered as if earlier pages were numbered: this could involve starting on folio 2r with the number 3, or on 2v with a number 4:

(iii) Pagination from the start of the content, but implying a page number 1 somewhere other than on 1r or 2r. A compositor accustomed to paginating from 2r with number 3 might well assign that number to 2v if an extra page of preliminary material were required:

(iv) Different pagination patterns in different partbooks, depending on the presence or absence of a dedication or other preliminary matter. This is particularly prevalent when there are organ books, often paginated according to a different pattern (see Examples 59 and 60).

(v) Piece numbering: this is already found before 1520, for example in Petrucci's Motetti de la Corona Libro II of 1519. It is most easily detected in books where the number of voices increases in later compositions, but the additional voices are printed in the same few partbooks. This results in some books having two voice-parts for the same piece on an opening, and the same number at the head of each page. A well-known example is the I Musica de diversi auttori à 7-12 (Venice: Giacomo Vincenti & Ricciardo Amadino, 1584: RISM 15844).

But other books, with consistent scoring, also have piece numbering: they include works from a wide range of printers. In some cases, it can seem to be indistinguishable from pagination, given the possibility (common with villanellas, madrigals and much late sixteenth-century music) of making each piece or pars fit onto one page. Then the index page, as well as the preliminary pages, will be unnumbered. Possibly such a pattern should be called piece numbering if it begins with the number 1 (regardless of the page involved) and does not number the page of contents at the end of the part; it also has to be consistent for all partbooks.[End Page 232]

There are also instances where the first piece begins on 2r, but, in some partbooks, the first lines of music are already printed low on the second page (1v). In these cases, there may be an additional number "1" before the music on 1v, indicating that the numerals were intended to be piece numbers, at least at this stage of the book:

(vi) Piece numbering which turns into pagination towards the end of many volumes, as longer pieces or those with more voices take up more than one page. The pattern of inserting a number on each page can continue to the end of the book. I believe this should normally be called pagination throughout, unless there is some strong indication otherwise:

(vii) Pagination which turns into piece numbers. Numbering can begin with any of the patterns already mentioned, and continue systematically as [End Page 233] pagination, assigning new numbers to different partes if they are on different pages, but then break down as these large-scale pieces appear at the end of a volume:

Each part opens with a title-page and a Dedication on the first verso. Neither is paginated, so that number 1 appears on the second recto. Numbering proceeds normally to page number 15, even though some pieces have taken more than one page: in other words, this is true pagination. However, starting at the end of the second gathering, the volume concludes with a large-scale nine-section Canzone à S. Francesco, texted "Sacrati horrori ove la folta chioma". As is normal, some sections are reduced scoring, and other take two pages. At this point, what had been page numbering becomes piece numbering, as shown below. As a result, the contents list does not need to be re-set for each partbook:

Piece Pars Canto Tenore Alto Basso Quinto
1 A2r 1 D2r 1 G2r 1 K2r 1 N2r 1
2 A2v 2 D2v 2 G2v 2 K2v 2 N2v 2
A3r 3 D3r 3 G3r 3 K3r 3 N3r 3
3 A4r 4 D4r 4 G4r 4 K4r 4 N4r 4
12 1 B4v 14 E4v 14 H4v 14 L4v 14 O4v 14
2 C1r 15 F1r 15 I1r 15 M1r 15 P1r 15
3 C1v 16 F1v 16 I1v 16 M1v 16 P1v 16
C2r 16 F2r 16 I2r 16 M2r 17 P2r 16
4 F2v 17 I2v 17 M2v 18 P2v 17
F3r 17 I3r 17 P3r 17
5 C2v 18 F3v 18 P3v 18
C3r 18 F4r 18 P4r 19
6 C3v 19 F4v 19 I3v 18 M3r 19
C4r 19 F5r 19 I4r 19
7 C4v 20 F5v 20 I4v 20 M3v 20 P4v 20
C5r 20 F6r 20 I5r 20 P5r 20
8 C5v 21 F6v 21 I5v 21 M4r 21 P5v 21
F7r 21
9 C6r 22 F7v 22 I6r 22 M4v 22 P6r 22
F8r 22
Tavola C6v [23] F8v [23] I6v [23] M4v [22] P6v [23]

This evidently sets out to number the pieces—in this case, sections of a larger work—consistently throughout the five partbooks. The two anomalies [End Page 234] (Bassus M2v, and Quintus P4r) do not destroy the larger pattern, and are themselves easily explained as temporary reversions to pagination, Amadino's more normal practice as this time.

(viii) Numbering which is based on the numbers appearing in a different partbook. Thus if the Cantus has pagination, the Altus or Bassus may assign the same numbers to the same pieces, regardless of the number of pages they take. This is likely to be most obvious in those volumes which contain pieces in a steadily increasing number of voices or where two voices are sometimes entered in the same partbook. In practice, it is most commonly found with piece numbering:

(ix) There are also still cases of foliation, rather than pagination, usually in books that are in folio format and where the music is laid out choirbook style. Examples include Palestrina: Missarum Liber Primus (Rome: Valerio & Luigi Dorico, 1554) and Lambardi: Antiphonarium Vespertina, Seconda pars (Venice: Cenobio Sancto Spirito, 1597: RISM L366). Among examples which are not in folio format can be cited the so-called "Cancionero de Uppsala", Villancicos De diversos Autores (Venice: Girolamo Scotto, 1556: RISM 155630), 37 and Il Lauro Secco (Ferrara: Baldini, Vittorio 1582). Some keyboard and lute books are also foliated, among them Valente's Intavolatura (Napoli: Giuseppe Cacchio dall'Aquila, 1576: RISM V33).

Some seventeenth-century books of solo song, usually in a long landscape format, continue to be foliated. A late example is not truly representative, in that it is only printed, from engraved plates, on the recto of each folio: Roncalli: Capricci Armonici, Op. 1 (Bergamo: Sebastian Casetti, 1692: RISM R2447). But the practice was never so common in Italy as it was north of the Alps.

(x) In addition, of course, there can be confused situations, just as there are of signatures, in which different books follow different patterns. I have already cited the case of Bassano's Op. 18 (Example 56), and another from his output follows, interesting because the pagination patterns follow those of different gathering structures and signatures:

The generally difficult nature of numbering patterns, not to mention their frequent inconsistencies, argues strongly against using pagination or piece numbering as a basic manner of referring to bibliographical aspects of a book. Piece numbering may well be the most easy way of referring to the musical content, but it obviously cannot be used to describe the structure or placing of (for example) watermarks. In those cases (common with Gardano and Scotto) where more than two pieces lie on an opening, or in those others (common around the turn of the seventeenth century) where one piece number applies to two or three pages, such numbers become valueless.

Apart from all these situations, however, there is another reason why page numbers are not a reliable means of reference. During much of the seventeenth century they are probably the least accurate part of a printed volume of music. It is possible, of course, to say at the outset of a description that the modern writer is "using corrected page numbers" and to proceed entirely consistently. In such a situation, the description will already have cited the actual pagination, and given the corrected versions against any errors:

In a case like this, omitted numbers can be safely inferred, and there are so few actual errors that they can be corrected in the description, and the correct ones used elsewhere. With so few errors, the process of reading the description and correlating it with a copy is not hampered by this style of treatment of errors: and the same is true of a number of my earlier examples.

But these are all simple examples, with few and obvious errors. In other cases, although there may be more errors, they are worth entering, and even discussing, because they throw light on the printing patterns or because they show a remarkable consistency:

These are still fairly straightforward cases. Unfortunately, however, there are many other cases where the typesetter has made nonsense of a logical sequence of numbering. Again, it is possible to list all the numbers and to give the numbers that should have been used, and then continue with the "corrected page numbers". However, the list can become so complex that the reader has trouble using it, especially since few of the numbers will correspond to those before him in the source. In these cases it is better not to use page numbers for any references to the content. The following examples should make this point clear:

Analysis of the pagination shows that two different patterns were in operation. In one (pattern a, below), pagination begins with the title-page, thought not actually entered until a later page; in the other (pattern b), it is presumed to start with 2r, athough the first indication may again not be until later:

Sheet I 1r 2v 9r 10v 1v 2r 9v 10r
A 4 17 20 A
B 4 17 20 B 1
C 4 17 20 C
D 4 17 20 D
Pattern a [1] 4 17 20 Pattern a [2] [3] 18 19
Pattern b 2 15 18 Pattern b 1 16 17
Sheet II 3r 4v 7r 8v 3v 4r 7v 8r
A 3 6 11 14 A 6 7 14 15
B 3 6 11 12 B 4 5 14 15
C 3 7 12 C 4 3 12 11
D 3 7 11 D 6 7 12 13
Pattern a 5 8 13 16 Pattern a 6 7 14 15
Pattern b 3 6 11 14 Pattern b 4 5 12 13
Half-Sheet III 5r 6v 5v 6r
A 7 10 A 10 11
[End Page 239]
B 7 7 B 10 11
C 9 11 C 10 11
D 7 10 D 10 11
Pattern a 9 12 Pattern a 10 11
Pattern b 7 10 Pattern b 8 9

The problem is typified by the two half-formes that comprise the innermost half sheet, III. While the inner forme follows pattern a, the outer seems to follow pattern b. The outer forme of the outermost sheet, I, follows pattern a consistently, while the only number on the inner forme represents pattern b. Finally, the second sheet has no clear pattern at all, for while the tendency to pattern b on the outer forme is clear, the inner is far less clear. No single partbook shows a consistent pattern of following one pattern rather than the other: indeed one forme of both Tenor and Bassus seems to change pattern in mid-forme. It is implausible that any one partbook was set up straight through, without pauses or change of compositor. Whether this means two compositors, or one who changed his habits, I cannot say. However, it is clear that any reference to "page 7", "page 11", or "page 18" will raise doubts in the mind of any reader.

Such cases may seem extreme, although they could be multiplied many times. However, even when the numbering seems logical and relatively consistent, we can not always be sure what it stands for: the difference between piece numbering and pagination is not always clear, the significance of the different patterns of pagination is not yet understood, and the apparent anomalies are as often the result of a change of plan as they are of compositor's error. Since the numbering was not needed by either compositor or binder, but was inserted as a convenience to the user, its accuracy was seldom a matter of concern. While erroneous signatures can often be seen [End Page 241] to have been corrected in the printer's shop, this is much less frequently the case with pagination errors.

For these reasons, it seems more logical to use for analytical purposes the numbering which was needed by both printer and binder, and which therefore was more likely to be accurate: that is the pattern of signature numbers. These patterns can be quite varied, but they are more closely followed in a majority of books.

Footnotes

1. I am grateful to the readers of this paper for their comments on this issue, which have led me to clarify many details of my presentation.

2. With multiple volumes in a set of prose, brackets have not been used to surround the indicator "Vol. I:", etc. However, the letters used to indicate part-names in my musical collations could easily be confused with actual signatures, especially since some printers used various marks of punctuation, including parentheses and colons.

3. The most frequent manner of setting a series of partbooks was to complete one book at a time, before moving on to the next. Mary Lewis has adopted the term "vertical composition" for a different procedure, in which the same forme (the first of the second gathering, for example) in every partbook was set before the compositor moved on to the next forme. This was primarily useful for anthologies where each composition took only one page—the same page—in each partbook. As a result, some material—the initial letter, the caption, the composer's name—could remain in the forme for all parts. See Mary S. Lewis, Antonio Gardano, Venetian Music Printer 1538-1569: A Descriptive Bibliography and Historical Study (New York: Garland, 1988- ), vol. 1, pp. 68-75. See also Donna G. Cardamone and David L. Jackson, "Multiple Formes and Vertical Setting in Susato's First Edition of Lassus' `Opus 1' ", Notes, 46 (1989-1990), 7-24.

4. See Fredson Bowers, Principles of Bibliographical Description (Princeton: Princeton Univ. Press, 1949: re-issued New York: Russell & Russell, 1962, and Winchester: St. Paul's Bibliographies, 1986). See also Franca Brambilla Ageno, L'Edizione critica dei testi volgari, Medioeve e Umanesimo, 22 (Padova: Antenore, 1975); Stanley Boorman, "Glossary", Music Printing and Publishing, ed. D. W. Krummel and Stanley Sadie, Grove Handbooks in Music (London: Macmillan, 1990), 489-550; Conor Fahy, Saggi di Bibliografia Testuale, Medioevo e Umanesimo, 66 (Padova: Antenore, 1988); and G. Thomas Tanselle, "Title-Page Transcription and Signature Collation Reconsidered", Studies in Bibliography, 38 (1985), 45-81.

The present article would have been impossible without the kind assistance of the Istituo per la Storia di Musica, at the Fondazione Giogio Cini, Venice, its Director, Professore Giovanni Morelli, and its assistant, Doctor David Bryant. Evidently, while I have tried to cite only copies that I have examined myself, in the library or (rarely) on microfilm, some of the examples cited below have been drawn to my attention by other scholars working in the field.

5. This abbreviated title appears on the direction line, usually on the first folio of each sheet—therefore on the first of each gathering in quarto—with the exception of the sheets containing title-pages. For example, the early editions of the most frequently-reprinted music book of the sixteenth century, the first book of four-voiced madrigals by Arcadelt, have slightly different titles. Two editions were published by Antonio Gardano in 1541 (both listed in RISM as 15419): one has the line "Primo Libro d'Archadelt" (with a final period only on folio R1r); the other has the same two versions, though in italic. The third edition, put out by Ottaviano Scotto (RISM 154110), has a similar line, though in a mix of italic and roman type: "Primo libro d'Archadelte." Gardano's edition of 1554 (RISM A1323 = c.155112) has a line reading Archadelt Primo a 4.

6. For descriptions of the books produced by these printers or publishers, see Stanley Boorman, Ottaviano Petrucci: Catalogue Raisonné (New York: Oxford Univ. Press, 2005); Daniel Heartz, Pierre Attaingnant, Royal Painter of Music: A Historical Study and Bibliographical Catalogue (Berkeley: Univ. of California Press, 1969); and Ute Meissner, Der Antwerpener Notendrucker Tylman Susato: eine bibliographische Studie zur niederländischen Chansonpublikationen in der ersten Hälfte des 16. Jahrhunderts, Berliner Studien zur Musikwissenschaft, 11 (Berlin: Merseburger, 1967), with Kristine Forney, Tielman Susato, Sixteenth-Century Printer: An Archival and Typographical Investigation (Ph.D. dissertation, Univ. of Kentucky, 1978).

7. This example is taken from Jane A. Bernstein, Music Printing in Renaissance Venice: The Scotto Press (1539-1572) (New York: Oxford Univ. Press, 1999), which contains considerable evidence from which one could build up a detailed picture of house-practice in Scotto's printing shop.

8. A curious example from his work concerns an edition of Arcadelt's first book of       madrigals (RISM A1319=154415), put out by Scotto in 1544 and signed as follows: [T:] A-F4; [A:] G-M4; [B:] N-S4; [C:] T-Z4 AA1. The book had already gone through several editions, and, although Scotto did rearrange the contents for this edition, it is difficult to see why the Cantus part should have been signed last.

9. Similar evidence can be found in a number of seventeenth-century editions, even when the gatherings are not particularly large. For example, Califabri's Scelta de'Motetti da cantarsi à 2-3 (Rome: Jacomo Fei, 1665: RISM 16651) has a Basso book, in quarto, which collates C12 D2. The Canto and Organo have single gatherings, both larger than the Basso.

10. Most printers tended to have a standard manner of dealing with a final half-gathering of music. Gardano always added a further half-gathering, with its own signature. Scotto did more often than not, though it is notable that there were two periods when he ended with a six-folio gathering—the year 1541 and the years 1549 to early 1554 (during which the only exception was a new edition of an earlier book). Scotto's heirs, however, preferred to end with a six-folio gathering, as did Angelo Gardano, when succeeding to Antonio's business. At the same time, some of their contemporaries were still signing with short final gatherings. By the end of the century, the longer gatherings had become more popular, partly because more printers were in any case using single long gatherings for each partbook. Even so, a number of publishers put out books with both solutions, during the period 1580-ca. 1660, just as a number were inconsistent in presenting quarto in gatherings of a single sheet or in large gatherings.

11. Bernstein, Music Printing, No. 163, points out that the contents correspond to those found in the Gardano edition of 1557 (described in Lewis, Antonio Gardano, vol. 2, No. 237), with the addition of three pieces at the end of the book. The first of these is attributed to Costanzo Porta, who (as Lewis suggests in Antonio Gardano, vol. 2, p. 37) was still relatively unknown: he may therefore have had something to do with the addition of music to the end of this often reprinted collection.

12. This is noted in Bernstein, Music Printing, No. 14.

13. RISM 154018, described in Bernstein, Music Printing, No. 15.

14. A more detailed discussion of this instance can be found in Stanley Boorman, "Printed Music Books of the Italian Renaissance from the Point of View of Manuscript Study", Revista de Musicologia, 16 (1993 [=1997]), 2587-2602.

15. This is also an argument that this book was prepared with "vertical setting", so that the first sheets of the lower voice-parts had already been set before the problem with the Cantus was discovered

16. A few general conventions need to be noted for the descriptions. A colon followed by a closing bracket, thus, :] is used to indicate that what precedes is comment, and what follows is a quotation from the specific document or source. If further comment is needed after the quotation, the bracket is reopened. For example:

Signatures:] A II [$4

(In the example, A II is an exemplary signature from the item in question.) This formulation also allows me to indicate when there is punctuation at the end of the quoted material. The manner of describing the signing pattern in particular is discussed in detail below, at Example 32.

17. This book is another example of the same situation described in Example 18 above: here again, following the Sesto book, there was no subsequent partbook to use BB-EE.

18. An amusing case of faulty imposition occurs in the 1512 edition of Reuchlin's Scenica progymnasmata (printed in Tübingen by Thomas Anshelm), one of the editions which contains music by Megel. Printed in the middle of folio C4r, an otherwise empty page, is the note] Erratus est hic in supputatione positionis & nihil omissum. | Verte paginam & mox sequitur Loco uix eredito &c. [thereby also indicating the first three words found on the verso.

19. I will return to this book later, in discussing its collation and signature pattern.

20. For Susato, see above, note 5. Sengenwald printed in Jena during the mid-seventeenth century. He published his first music book in 1649, and another eleven appeared during the next twenty years. Several were occasional books, commemorating a local death, with the musical composition merely an appendix to a longer literary text. Others comprise the usual mix of sacred and instrumental works by local composers.

21. The data here are taken from Bernstein, Music Printing, No. 11. The book was apparently constructed so that each gathering of each partbook would contain just one mass.

22. A similar structure appears in the companion edition of five-voiced masses by Morales and Jachet (RISM M3575 = 15403). Again, each gathering contains one mass, and the first in each book (here intact) contains six folios, including the title-page and a table of contents. Since the Tenor parts are much shorter than those of the other voices, the Tenor collation is unusual: a6 b4 c2 d4 e2.

23. As will appear, I believe one should use signatures to refer to virtually all features of early printed music. Foliation is often more accurate than not, but pagination patterns are regularly eccentric or incomprehensible. However, signatures are always more reliable, for printers apparently felt the need to correct signatures more often than either foliation or pagination.

24. A brief example of this practice has already been mentioned, with Example 28. The earliest musical instances known to me are in two books of 1549, put out by Antonio Gardano. They are Buus: II Recercari (RISM B5196), and Werrecore: La Bataglia Taliana (RISM M1404). I have consulted the copy of the first at the British Library, and take details of both from Lewis, Antonio Gardano, vol. 1, Nos. 128 and 133.

25. The collation of a normal book may reveal a great deal about the signature patterns, but only when all gatherings begin with an implicit number 1 in the signature. The present case, and others like it, highlight the problems that can arise when a different practice prevails. Further, the collation does not allow one to assume that a presumed signature will refer only to one folio in the book: therefore, in this as in many other musical volumes, it is not enough merely to refer to a given folio by its presumed signature, without indicating in which partbook it may be found.

26. One could refer to individual folios in this book by using a complex reflection of the signing patterns, in which the third gathering would open with "A5-1", "A5-2", and so on. But this in incompatible with other practices, as well as failing to represent what is found in the book.

27. See example 15, above.

28. This book was also used as Example 33, when discussing initial gatherings without music.

29. In fact, the first signed folio in the Tenor book, 1A1r, is erroneously signed A3.

30. Indeed, I know of no instances from sixteenth-century Italy where the pattern of repeating signature letters can be certainly said to have been planned from the beginning: all those I have seen could plausibly be explained as the result of error, or of a change of plan.

31. This copy has been published in facsimile in Solo Motets from the Seventeenth Century: Facsimiles of Prints from the Italian Baroque, vol. 6, ed. Anne Schnoebelen (New York: Garland, 1988).

32. This is not the only example of this practice: in 1567, Antonio Barrè printed the third book Delle Muse à 4. Madrigali ariosi (RISM 15627), for which three partbooks survive. For each part, a single bifolio is wrapped around three quarto gatherings, A-C. The bifolio contains the title-page, a dedication, the privilege, and a contents list, each taking one page: each partbook shows the same setting of all pages, with the simple change of the name of the voice-part. According to Mauren Buja, Antonio Barrè and Music Printing in Mid-Sixteenth Century Rome (Ph.D. dissertation, Univ. of North Carolina, Chapel Hill, 1996), 349-358, from whence these details are taken, the bifolio is signed on both rectos, A1 and C5: the normal gathering A is signed A2 and A3, B is signed normally, and the four leaves of C are signed C1-C4. The contents give no reason for this pattern, other than the ease of printing the complete run of the wrap-around bifolio apart from the musical content.

33. Bowers, Principles, pp. 213-219.

34. This copy has been reproduced as the first volume in the series Archivum Musicum: La cantata barocca (Florence: SPES, 1980).

35. Bernstein, Music Printing, p. 404.

36. This copy is reproduced as Edizione Anastatica delle fonti Palestriniane: Prima Serie, vol 2, ed. Giuliana Gialdroni (Palestrina: Fondazione Giovanni Pierluigi da Palestrina, 1989), pp. 151-181. This volume reproduces other editions of the same Palestrina book, and demonstrates some of the different procedures being discussed here.

37. This has been published in facsimile from the copy at Uppsala, Universitetsbiblioteket (Peer: Alamire, c. 1984).

38. The distinction between those editorial page numbers which are given as if in sequence, [85-86] for example, and those which are truly editorial, such as [vi], is a useful one. In the former, we are assuming that the number was merely omitted but borne in mind by the compositor, an assumption which can be made because the numbers continue in order after the omission. In the latter, it is not possible to assign a number to the omission, for the sequence does not continue in correct order. The editorial roman numeral, customarily given to the page, is useful for drawing attention to what may be a significant feature of the preparation of the book.