ii. Variants: Revision or Correction; Mistakes and Intentionality

Machaut's complete-works manuscripts demonstrate a remarkable uniformity despite being complex in their production and content and having been produced over a period of around fifty years. Nevertheless, these manuscripts include a large number of local variants of different kinds which are on a more modest scale than the different combinations of voices discussed on the previous pages.

Since this is a large topic, it has been divided into a set of sub-discussions. These start with variants arising from the recopying of early, non-standardised versions of works into later exemplars that enjoyed a more stable circulation. This is followed by an examination of specific locations where variants cloud Machaut's putative original intention, resulting in problematic places with different readings and solutions. Next will be considered a number of small and most-likely unintentional variants whose reading, nonetheless, has a significant effect on the resulting versions. Then, the issue of variants arising from clear mistakes is discussed, including corrections made to versions within single source. Finally, thoughts are offered on the interpretation of variants: can we always tell whether they were made in error or deliberately?

This section does not include a separate discussion of variants in underlay and their effect. Examples relating to this topic can be found elsewhere on the website.


Uri Smilansky

1. Early, Non-standardised Versions

It is possible to interpret some variants as the result of a process of revision between subsequent sources. This, of course, is easier to perceive when new voices were added to the setting, and as with the discussion of that topic, most of these cases occur in the transition from the early manuscript MS C to the other collections. Such revisions are by no means universal. For example, Ballade 30, written after this breakpoint, is copied with the same mistake in its tenor voice in all four of its concordances. When revisions do occur, their reasoning does not always seem corrective.

The B-part of B12 can be taken as an example here.

MS C:     Sound and Score (MS C)     ¦     Facsimile

MS A:     Sound and Score (MS A)     ¦     Facsimile

It is possible that the prime reason for revising this section was the disposal of the parallel octaves between measures 51 and 52 of the older MS C version. Such progressions, though, appear elsewhere in Machaut’s two-part compositions where they did not cause revision, and the new reading only displaces the skeletal parallels by inserting a dissonance here rather than ‘improving’ the counterpoint. Indeed, the more remarkable changes, appearing at the beginning of the form-part, subvert the more standard counterpoint of the old version, inserting strong and protracted dissonances in b. 30-32. It is hard to believe contrapuntal probity was the reason for the changes here.

The new beginning of this form part is also a measure longer than its predecessor. Technically, one has to apply perfect Modus in order to read this song correctly. The longer reading, though, forces the tenor to contradict the perfection rules and keep the first longa in this section perfect, even though it is followed by a string of four breves. Editorially, modus grouping can be represented as follows:

MS C:     Score     ¦     Facsimile

MS A:     Score     ¦     Facsimile

The barring here makes it clear that the extra measure inserted in the later versions does not resolve a deficiency in the rhythmic behaviour of original, but destabilizes the regularity of the Modus groupings. To make the section work, b. 19 becomes too long. It is, of course, possible to interpret some rests as separation lines and force the music into regular Modus measures, but this requires inconsistent interpretation. When maintaining a regular interpretation of these lines as rests (for this song, if not for the entire oeuvre), the addition occurs at a cadence point, making it easy to incorporate the resulting irregularity in performance. Other problems in the later MS A score – an omitted note in b. 21 of the cantus and too long a rest (not represented in the edition) in b. 22 of the tenor – appear only in this MS and are unique errors of its scribe. They cannot be seen as an outcome of the revision process.

This example suggests that while some revisions may have occurred due to problems in the older version, many others are aesthetic rather than corrective, often replacing one usable version with another. Indeed, the alterations in B12 offer the analyst (if not the performer or audience) more rather than fewer difficulties.


Uri Smilansky

2. Problematic Places with Different Readings and Solutions

Within the central manuscript tradition, one encounters locations where no manuscript presents a clear and unproblematic reading. Such places are especially hard to resolve in a monophonic context, where voice-alignment and harmonic considerations are not present.

One such case is the beginning of strophe X of L19. The following diagram presents the readings in the different sources:

The only reading that can be fitted into the overall mensural pattern can be found in the musical repetition marking the middle of the strophe, where MS A omits the first rest (in brackets above), probably mistakenly. As it stands, this beginning requires either a mensural or conceptual adjustment. A strict reading of MS C requires the third bar of this section to be shortened.

Sound and Score     ¦     Facsimile

A strict reading of the other sources – exemplified through the version in MS E requires the first bar to be extended to include four beats.

Sound and Score     ¦     Facsimile

The conceptual adjustment adopted for the edition – applied here on MS A – understands the first note of the strophe as an upbeat even though it is not notated as such. To compensate for this, the rest at the end of the previous section is shortened. This arrangement allows for the rest of the strophe to be read without problem.

Sound and Score     ¦     Facsimile

This kind of reading is further discussed in connection with un-notated upbeats here. For another case of all sources presenting unclear versions (rather than actual mistakes), see here.


Uri Smilansky

3. Unintentional Variants

Sometimes, it is hard to determine whether a variant is intentional or accidental, or even whether it would have been acted upon or ironed out in performance. Such questions are especially relevant when the changes are minute, but nevertheless crucial to the reading.

The actual notes of the A-part of V16, for example, are not disputed. Still, MSS C, A and G have a dot after this song’s second note, which is missing in MSS Vg, B and E. Other parameters such as spacing and underlay may suggest that the intention of at least some of the latter sources was to follow the reading of the former group. Still, strict adherence to the notational conventions would cause this small detail to result in very different readings.

MS A:     Sound and Score     ¦     Facsimile

MS E:     Sound and Score     ¦     Facsimile

Interestingly, as the settings of the last two lines of text of this form-part incorporate both ‘long-short-long-short’ and ‘short-long-short-long’ rhythmic combinations, both readings of the song’s beginning can be naturally integrated into the rest of the work. Furthermore, the reading which is more likely to be erroneous (the one without the dot) offers better alignment between mensural strong-points and poetic stresses, as well as a more conventional starting point for line four of the text (b. 7, at the beginning of a brevis unit). While less likely to have been Machaut’s intention, a reader of this version would not have had any reason to suspect it was wrong. It is also worth remembering that while the odd historical occasion enabled two Machaut manuscripts to be found in the same library, these books were in all likelihood consulted independently and singularly, hence a reader would have been unlikely to have the opportunity to form an opinion as to their preferred version based on a variety of sources in the manner that we are today.

For a number of reasons, MS E is considered closer to actual musical practice than the other sources. This raises a number of interesting questions with regard to the reading of V16. It is possible that the scribe knew the intended musical result before noting this song down. In this case, the correct imperfection may have seemed so obvious that the appearance or otherwise of a dot would not have influenced this song’s reading. On the other hand, the tentative links of Vg and MS E to subsequent performance, combined with the likely intended use of MS B as an exemplar in the further circulation of Machaut’s work as written rather than heard artefacts, might together suggest the possibility of the incorrect version being performed more widely than the correct one.


Uri Smilansky

4. Mistakes and Corrections

As in the case of the dissemination of unintentional variants, it is clear that the availability of sources to compare readings would also be a central issue in the transmission of errors and their influence on practice. While a mistake in the transmission may be clear to us through an examination of all surviving sources, it would not have been so easy for medieval users to identify them (we can better appreciate this when we consider that we are faced with a similar situation when working with compositions transmitted in one source only).

An interestingly layered case of erroneous transmission is the B-part of V10. It is clear that for this song MSS Vg, B and E are directly linked, each source using its closest predecessor as an exemplar. Still, each source presents a markedly different reading.

Vg reproduces the standard and correct reading familiar from all the manuscripts outside this group.

Sound and Score     ¦     Facsimile

MS B made a number of copying errors, seemingly conflating the intended melody with that of the B-part of V7 (copied in the same position in the previous opening of Vg). The result is a rather messy hybrid, truncating the longer form part of V7 to fit more or less syllabically above the pre-copied text.

Sound and Score     ¦     Facsimile:     Vg (V10)   ¦   Vg (V7)   ¦   MS B

Perhaps due to this lack of clarity, when the scribe of MS E came to copy his version, a few more changes were inserted, probably unintentionally.

Sound and Score     ¦     Facsimile:     MS B   ¦   MS E

These three versions attest to the degree of variety possible even with direct copying. Interestingly, some reviewing evidently took place of the MS E version, as the second variant between it and MS B was subsequently removed (this involved the crossing out of a stem attached to the fourth note before the end of this section). This resulted in a more symmetrical division of this line into two mirroring rhythmic patterns, even though this does not match the text structure.

Sound and Score     ¦     Facsimile

Both error and correction come together to produce a workable version, but one which differs markedly from Machaut’s original intention.

Pages are available with discussion on clear or interpreted errors.


Uri Smilansky

5. Error or Deliberate?

In comparison with his contemporaries, Machaut's works seem to suggest he was rather liberal in his use of dissonance. However, since his well-preserved oeuvre constitutes by far the most comprehensive source material we possess from his time, such a comparison is difficult: we cannot be certain as to how far Machaut really was exceptional in these matters, or whether the generally accepted conventions and the latitude they left for dissonance are insufficiently known to us. Alternatively, his exemplar may have contained errors that were taken over in all the manuscripts copied from it. In several of Machaut’s works such grave contrapuntal conflicts are found that they must be considered either as errors or as deliberate decisions to transgress the conventions, perhaps in some cases for the sake of text expression. A few problematic passages from the motets may serve as examples.

Let’s begin with a doubtful case:     Score 

In M3, bar 42, a dissonance d-c between triplum and motetus lasts for a brevis. It could be a mistake since collisions like this one are not very frequent in Machaut’s works; on the other hand, all the manuscripts contain it. Although for the modern ear such a long-lasting dissonance is hard to accept (and easy to “remedy”), theoretically, the clash could have been acceptable since both the upper voices are concordant with the tenor. The author of Quatuor principalia (John of Tewkesbury, active during the second half of the 14th century) permits dissonance in three-voice counterpoint under the condition that two of the possible three voice pairs are consonant.

M18 is an early work, probably composed in 1324-5 for the enthronement of Guillaume de Trie as archbishop of Reims in that year:     Score

In talea III, bars 51-62 present a very strange sound. The unexpected augmented sixth between tenor and motetus in bar 56-7 is particularly hard to explain, the more so since, just before, the upper voices are in unison above a rest in the tenor: nothing prepares us for this sudden tension. In the earliest manuscript, C, the g of the motetus has no mi-sign so that there the problem does not occur. Could the sharpening be a later addition, or did the scribe of C forget the sign; or is it a mistake that crept in later? If it is not a mistake, was the combination of g-sharp and b-flat meant as a double leading tone for a strong cadence on a? In that case the continuation is curious since right after it, in bars 59-62, an equally strong cadence on G follows, and then a weaker one on F. If one compares the passage with the opening of the piece (which has the same tenor notes) the cadences in the first talea are much more the expected ones, the last one on F being the strongest; the cadence succession in talea III is much more surprising. Clearly, then, the dissonance in bars 56-7 is a deliberate choice and must have been meant to shock. But for what reason? In order to make the following cadence on G all the more satisfying after the dissonant delay? The c-sharp in the triplum in bar 59 as a leading note for the next cadence also comes as a surprise. The tension caused by these sharpened notes contrasts strongly with the dominating motif of the piece, the triad c-e-g which sounds so vigorously at the opening of the motet, crosswise: ascending in the triplum, descending in the motetus and both times on the words Bone pastor. That same ascending motif is repeated at the second color entry in the motetus (bars 51-2), on the name of the ‘Good shepherd’, Guillerme, but its harmonic stability is immediately disrupted by the g-sharp in bar 56. Could it be explained as an attack on the bishop’s steadfastness which is praised in the motet and which would then be expressed by the energetic triadic motif? Far-fetched as this explanation seems, there may be a textual argument, since at the end of the triplum text the speaker prays God to give Guillermus ‘a stable dominion in place of this unsteady one’ (‘Stabile dominium Pro labili.’).

Even more problematic are the collisions in M5, a four-part composition:     Score

This motet is often supposed to be an early work written in emulation of a motet by Philippe de Vitry;[1] however, since the work is very subtle in its use of both musical and textual quotations, in its deceptive notation and its text expression, it might very well be a later work.[2]  Several clashes occur between the voices; was four-part counterpoint more permissive? Take for example bars 27-31: all the manuscripts have a c in the motetus which is dissonant with the contratenor’s D for a length of no less than two perfect breves; an emendation of the motetus to d (Ludwig’s solution) or to a (my suggestion) seems preferable. A little later (bar 30) the G of the tenor is overlaid with a G-sharp in bar 31 in the motetus (found in all the manuscripts except E); the severe dissonance, lasting for only a semibrevis, could only be solved by not holding the tenor note until the end. The tenor’s G cannot be sharpened as this borrowed voice is never to be altered to agree with an upper voice, it is always the other way round: the upper voices must adapt to the tenor. Such chromaticism as found here is extreme, however. A miscalculation is a real possibility but one could also think of deliberate transgression for the sake of text expression; the motetus text speaks about the ‘false lover who does not take heed of the god of Love’ whereas the lover in the triplum vows to do what is pleasing Good Love. Moreover two successive plicas in bars 31-2 of the motetus (not in all the manuscripts) emphasize this moment in the text.

Text expression, therefore, may sometimes serve as an explanation of otherwise unjustifyable dissonances, although this is difficult to prove, since no theorist ever dropped a word on such devices.

Some other cases are dealt with elsewhere on this website. M11 and the possible solutions for its contrapuntal problems appear in a number of discussions, but most extensively in relation to the application of editorial musica ficta.

M1, a model motet which not for nothing was placed first in the motet collections in all manuscripts also contains a model contrapuntal problem discussed under Voice hierarchy and structure.


Jacques Boogaart


[1] See Daniel Leech-Wilkinson, Compositional Techniques in the Four-Part Isorhythmic Motets of Philippe de Vitry and his Contemporaries. New York: Garland, 1989, pp. 88-107.

[2] See Jacques Boogaart, ‘Encompassing past and present: quotations and their function in Machaut’s motets.’ Early Music History 20 (2001), pp. 56-86.