Two early layers of Sanctus melodies

Abstract

This article examines Sanctus melodies from the tenth and eleventh centuries with special attention to the division of the first verbal phrase. The melodies with circulation in all regions of the Roman rite fall into two groups, an earlier one with ternary division and a later one with binary division. This picture is further enriched by the analysis of melodies connected in some way with these widespread melodies, by the simple melodies of the Sundays and weekdays and by an overview over northern French melodies of the eleventh and twelfth centuries.

Some chant repertories, especially the proper chants of the Mass, have a large core of texts and melodies that remains virtually unchanged from the earliest manuscripts up to printed books around 1500. The repertory of ordinary chants, on the other hand, seems to be in constant change, at least at first glance. Leaving aside differentiations that will become necessary later, it seems reasonable to sort this huge repertory according to geographical and chronological distribution in the preserved manuscripts (from the tenth century onwards) and to use this as a base for conclusions about the origins of the particular melodies.¹ The present study attempts to identify two early layers of Sanctus melodies or (expressed differently) one early change in the way of shaping the melody for an unchanging text. The central argument concerns a small group of melodies (out of more than 200)² that have found reception in practically all regions of the Roman rite. In the case of these widely disseminated melodies, the relative small number of early manuscripts that survive will constitute a smaller obstacle to drawing conclusions about origins than in the case of melodies of restricted circulation.

Melodies of international reception

The text of the first phrase of the Sanctus can be divided in two different ways.³ Either the third invocation sanctus is connected to the following word Dominus, resulting in a ternary division: Sanctus sanctus – sanctus Dominus – Deus Sabaoth. Or the three invocations are separated from the rest of the sentence, resulting in a binary division: Sanctus sanctus sanctus – Dominus Deus Sabaoth.

The ternary division is clear if the third sanctus is given a contrasting melodic shape against two similar melodies for the first two (Example 1b) and/or if the third sanctus is given an ascending melodic motion leading to the accent of the following word (Example 1b) or if Dominus is given a cadential gesture comparable to that on Sabaoth (Example 1a).

Example 1. Sanctus 216, 154, 32, 49 (F-Pn lat. 10508).

The binary division is clear if the three sanctus are connected by a melodic pattern ABA or ABA′ (Examples 1c and d) and/or if Dominus is given an ascending melodic motion leading to the accent on the first syllable of the following word (Example 1c) or if it is included in the melodic preparation of the cadential gesture on Sabaoth (Example 1d).

Example 1 shows the first phrases of four of the eight Sanctus melodies that have found reception in practically all regions of the Roman rite.⁴ The geographically broad transmission allows a relatively precise chronology. Sanctus 216 (not in Vatican Edition (hereafter Vat.)) and 154 (Vat. I) are present in the three manuscripts that can be dated to the first half of the tenth century (Table 1). In the St Martial manuscript (BnF lat. 1240), the Sanctus melody connected with the introductory trope Sanctus Deus omnipotens remains without notation. The later Aquitanian manuscripts, however, connect this trope consistently with Sanctus 154. Besides the two common melodies, St Martial and St Gall each have one further melody. Sanctus 111 (not in Vat.) can be found in later Western manuscripts, Sanctus 153 (not in Vat.) in later Eastern manuscripts only. Thus the oldest layer of international (festal) melodies consists of exactly the two melodies 216 and 154.

Table 1. Sanctus melodies in the three oldest manuscripts

Sanctus 111 and 153 (Example 2) show a ternary division of the first phrase, too. In both cases the third sanctus contrasts with the first and second, and it leads to the following accent on the first syllable of Dominus with an ascending melodic motion. A significant difference may be seen in the relationship between the first and second sanctus: in the Western melody 111 they are identical (or variations as in most Aquitanian manuscripts), in the Eastern melody 153 they seem to be independent.

Example 2. Sanctus 111 (E-Mn 289), 153 (A-M 109), 56 and 70 (E-Mn 289).

Sanctus 32 (Vat. XVII) and 49 (Vat. IV), on the other hand, appear almost at the same time in the eleventh century. Another contemporaneous melody should be included in the following considerations; it appears in two versions – Sanctus 56 (Vat. III) transmitted only in the West, and Sanctus 70 transmitted in the East – and additionally in one Norman-Sicilian manuscript (E-Mn 289) and one late French manuscript (F-LG 2 from Fontevrault). The three sanctus invocations, with a clear ABA pattern, are identical in both versions. On Dominus, however, melody 56 forms a sort of cadential gesture similar to that on Sabaoth, as we would expect in melodies with ternary division. In melody 70 this ambivalence is avoided by giving Dominus an ascending melodic motion that obviously forms the beginning of a melodic arch for Dominus Deus Sabaoth. There are further observations (see next section) that strengthen the probability that Sanctus 70 is in fact a reworking of Sanctus 56.

The distribution of these three melodies shows significant differences among the regions (Table 2). In Germany, the chronological order seems to be clear: first came Sanctus 70, then 32, then 49.⁵ It is rather improbable, however, that any of these melodies originated in Germany. Besides the general assumption that melodies common to East and West are normally of Western origin,⁶ those Sanctus melodies that are restricted to the East – 153, 17 (Vat. VI) and 208 (Vat. a.l. II) – show remarkably little interest in melodic interrelations between the three sanctus invocations. Melodies with clear interrelations, therefore, are probably imported from elsewhere. In Lorraine and Burgundy, the situation seems to be similar: first Sanctus 70 or 56, then 32, then 49. In Italy, Sanctus 56/70 is unknown,⁷ and Sanctus 49 appears much later than 32.⁸

Table 2. Early manuscripts for Sanctus 56, 70, 32 and 49

In Aquitania, however, things are different: while some tenth- and eleventh-century manuscripts do not contain any of these melodies⁹ and Sanctus 56 is attested only marginally,¹⁰ Sanctus 49 is found in the manuscripts from St Martial from about the middle of the eleventh century onwards; in earlier manuscripts it is sometimes added.¹¹ The St Yrieix manuscript, whose precise date is not clear, is the first to include both Sanctus 49 and 32 as do most twelfth-century manuscripts. The assumption of an origin of Sanctus 49 at St Martial, however, is not probable, since the first witness, from Regensburg, is earlier.

The relevant manuscripts from northern France are few and they do not agree among themselves. St Vaast joins some other manuscripts from the eastern parts of modern France (Metz, Cluny), containing Sanctus 32, but not 49. St Denis, on the other hand, joins St Martial, containing Sanctus 49, but not 32. This testimony is confirmed by the Durham manuscript, probably from Canterbury; it represents the Anglo-Saxon tradition in England that depends, as it seems, on monastic traditions of northern France close to that of St Denis.¹²

This situation is probably best understood if we assume for the two famous melodies the same time of origin, but different places. Sanctus 49 seems to come from northwestern France, while Sanctus 32 might come from the eastern parts of modern France. Sanctus 56 seems to be a bit older than these, but perhaps geographically close to Sanctus 32.

From these seven melodies (216, 154, 111, 153, 56/70, 32, 49) a provisional picture might be drawn: the earliest layer of Sanctus melodies (visible for us from the first half of the tenth century on) follows the ternary division of the first phrase, a second layer, beginning near the end of the tenth century, develops the binary division together with the ABA scheme for the sanctus invocations.

Interlude: melodic comparisons

The relationship of Sanctus 56 and 70 requires clarification. Some further melodies seem to be derived from or influenced by the melodies cited earlier. Therefore, in the following section these Sanctus melodies are discussed on the basis of complete transcriptions. They will enlarge the picture, especially regarding the earlier layer.

Example 3 shows Sanctus 56 and 70 from the only manuscript that includes them both. Both melodies have identical sections: Sanctus sanctus sanctus; Sabaoth; in nomine Domini. Furthermore, (o)-sanna in excelsis in Sanctus 56 corresponds to in excelsis in Sanctus 70. The most important differences concern internal melodic repetitions: in both melodies the textual repetition of osanna in excelsis is connected with a melodic repetition. In Sanctus 70, the repetition extends to Pleni sunt caeli et terra – Benedictus qui uenit in nomine Domini. Parallelisms of this sort between the second and the third part of the chant are common in melodies of all layers of the repertory, but they are not obligatory. An extension of the parallelism to the first part is less common. It is short in Sanctus 56, where the first syllable of the second half of each part (Do-, o-) has the same characteristic melisma; it is extended to the whole words Dominus – osanna in Sanctus 70. The only melodic repetition outside these parallelisms concerns the words gloria tua. These could have been integrated into the parallelism between the second and third parts, but here they are set apart. In Sanctus 56 they correspond to Sanctus, in melody 70 they correspond to in excelsis.

Example 3. Sanctus 56 and 70 (E-Mn 289).

Thus, if we construe one of these melodies as the result of a reworking of the other one (this can hardly be avoided), it seems clear that Sanctus 70 extends the scale of melodic parallelism to be found in Sanctus 56 and should be seen as the reworked melody. This agrees with the observation made earlier that Sanctus 70 replaces the ambiguous division of the first part by a clear binary division.

Sanctus 51 (=13a in Hiley's catalogue, not in Vat.) has a restricted area of transmission: St Vaast and some of its neighbours (F-DOU 124 from Anchin), Canterbury (GB-DRu Cosin V.V.6) and parts of the Norman traditions (RUS-SPsc O v I 6 from Meulan, Norman-Sicilian manuscripts). The standard (and presumably original) version is found with few internal variants at St Vaast, Canterbury and Meulan (Example 4a). Anchin and the Norman-Sicilian manuscripts (Examples 4b and c) present versions with melodic variants at several places.

Example 4. Sanctus 51 from F-AS 437, F-DOU 124, E-Mn 289.

The phrase osanna in excelsis has two different melodic renderings in the standard version (this can be judged as the lectio difficilior); in the Sicilian manuscripts in excelsis becomes identical by assimilation of the second rendering to the first; in Anchin the whole phrase becomes identical, mostly by assimilation of the first to the second, except for the assimilation in the opposite direction on the syllables in ex-. On gloria tua, the Anchin version is assimilated to Sanctus 154 (Example 8).¹³

The first phrase shows a ternary division: the first two sanctus invocations end on the final pitch, the third one, however, ends a fourth above and seems to open a melodic phrase that comprises the rest of the text (‘ternary’ is less appropriate here, as in the Sunday melody 223 presented later). The melody begins with an ornamentation of the third degree above the final; in Anchin and Sicily this ornamentation is enlarged, becoming thereby identical or almost identical to the beginning of Sanctus 56/70. The assumption that there was some influence from Sanctus 56/70 is strengthened by the Sicilian version, whose third sanctus has become similar to the second sanctus of melody 56/70.

Sanctus 202 (Vat. XI) is well known since its inclusion in the Franciscan Kyriale. Before the thirteenth century, however, it was restricted to two areas: the three Norman traditions (Normandy itself, Norman-Sicilian and Norman-English) combined with a manuscript of unclear provenance (Fleury?),¹⁴ and Lorraine, represented by manuscripts from Prüm and Metz. The Norman manuscripts combine it with the trope Pater ex quo omnia (CT VII 94); the Prüm and Fleury(?) manuscripts have no musical notation for the Sanctus, but the same trope (which is not connected with other Sanctus melodies); the Metz manuscript has no trope, but provides musical notation for the Sanctus. Another melody, Sanctus 204 (not in Vat.), shares these two areas of transmission and the connection with a trope (Qui regnas sine fine, CT VII 128).¹⁵ It lacks the later success of Sanctus 202, but is present in a third area, adiastematic southern German manuscripts, beginning with the late tenth-century troper from Eichstätt(?).¹⁶ Both melodies seem to be connected with the old Sanctus 216. Examples 5, 6 and 7 show all three melodies.

Example 5. Sanctus 216 (F-Pn lat. 10508).

Example 6. Sanctus 202 (F-Pn lat. 10508).

Example 7. Sanctus 204 from D-Mbs clm 14083, F-Pn lat. 10508, E-Mn 19421, F-Pn lat. 10756 (incipit only).

Melodic correspondences with Sanctus 216 are marked in Examples 6 and 7. Some of these may be ascribed to accident or typical behaviour of the mode, but the combination is significant – as well as the extension of the parallel passages in Sanctus 204.

Sanctus 202 (Example 6) can easily be construed as a conscious reworking of Sanctus 216, aiming at modal clarity: the first and second sanctus invocations begin and end on the final pitch, the first sanctus presents the characteristic species of the fourth of the second mode, pleni and osanna the species of the fifth. This fits well into a view of mode based on interval species that found its most radical expression in the compositions of Hermannus Contractus and exerted considerable influence on musical composition from the eleventh century onwards.¹⁷

Sanctus 204 (Example 7) is less easy to understand, especially due to the assimilation of the last part of the melody (from benedictus) to Sanctus 154 (Example 8) that changes the modal orientation of the melody from the final D to the final E.¹⁸ The adiastematic notations from Metz, Echternach and southern Germany point to a different melodic version of this part that cannot be reconstructed, but ends obviously with the same melodic phrase on (o-)sanna in excelsis as the foregoing part. It seems that this version, ending probably on D, is the original one and that the Norman version is the result of an accident in transmission.¹⁹

Example 8. Sanctus 154 (F-Pn lat. 10508).

The adiastematic manuscripts unfortunately give no help regarding the melodic variants within the Norman tradition in the first part of Sanctus 204. These concern mostly the final pitches of the subphrases that change between D and E. It seems impossible to distinguish here between attempts towards modal unification according to the old or the new final of the piece and possible reassimilations to Sanctus 216. For the rest of the melody one may assume the intention to avoid the descent to the plagal space below the final that is characteristic for the osanna in Sanctus 216.

Two further melodies connected with Sanctus 216 can be found in Aquitanian manuscripts: Sanctus 227 and 228 (Example 9). Their use of material from the older melody is in some ways complementary to Sanctus 202 and 204: it is restricted to the characteristic figure on osanna, while the equally characteristic shape of the first phrase is retained only in the first sanctus invocation, which is then repeated in the second. Both melodies have in common the additional emphasis on the species of the fourth on gloria tua and in nomine Domini, and the cadential figure on in excelsis. Sanctus 228 might be the older one, since it found wider distribution (including Apt, Narbonne and Catalonia), whereas Sanctus 227 is known to us only from manuscripts from St Martial. Hence there were probably two steps of reworking. The intention behind the first step might have been the concentration on the final pitch as the exclusive cadential pitch; the intention behind the second step remains unclear.

Example 9. Sanctus 227 (F-Pn lat. 909) and 228 (F-Pn lat. 887).

The basic layer

The foregoing discussion dealt with festal melodies that are regularly notated in early tropers, often together with tropes. Melodies for normal Sundays and weekdays, on the other hand, show different patterns of transmission. They are often omitted in early manuscripts, but one can find them regularly in manuscripts from the liturgically unified Orders (from the twelfth century onwards) and more or less regularly in late medieval manuscripts from all regions and institutions. Traditional chant scholarship has tended to regard the simplest melodies of ordinary chants, that is the melodies for weekdays and Sundays, as the oldest ones. In the second half of the twentieth century, however, this view has been challenged from the point of view of the earliest manuscripts.²⁰ This is not the place for a comprehensive discussion of this topic, but it seems reasonable to study the position of the Sanctus melodies for Sundays and weekdays within our provisional picture of the early layers.

The Sunday melody (223, Vat. XV, see Example 10) is rather unproblematic. On the one hand there are some early witnesses pairing it with the trope Ante saecula (CT VII 6, common to East and West), beginning with the late tenth-century manuscripts from Prüm and Eichstätt(?) already cited.²¹ On the other hand the melody belongs clearly to the group with a ternary division of the first phrase. To be more precise, though, ‘ternary’ is not a very appropriate description, since in this case sanctus Dominus Deus Sabaoth forms a single melodic arch that clearly begins with the ascending motion on the third sanctus. The melodic gestures of the first two sanctus are not connected by repetition or variation, but they both lead to the final pitch D, one from below, the other from above.

Example 10. Sanctus 223 (F-Pn lat. 10508) and 41 (D-DS 868 (OPraem), E-Bbc 888 (OCarth)).

The weekday melody (41, Vat. XVIII, see Example 10) is not attested before the eleventh century. The ramifications of its melodic transmission require a study of its own;²² in this context, only the most obvious points will be discussed. The majority version of the beginning, here represented by a late twelfth-century Premonstratensian manuscript, belongs to the ternary group, forming a single melodic arch for sanctus Dominus Deus Sabaoth as in Sanctus 223. The minority version, here represented by a twelfth-century Carthusian manuscript, forms an AAA pattern for the three sanctus invocations and seems, therefore, to belong to the binary group, although it marks no strong divisions at all.

There are arguments in favour of the priority of the minority version. This would probably require separating it from the binary group and putting it at the beginning of a third group that includes further melodies with an AAA pattern restricted to Italy.²³ In any case the majority version fits well into the oldest layer as defined earlier. This observation will, however, not bear the weight of an argument against a scenario of ‘late’ origins of the weekday melodies.

Further melodies from France and England

To gain an idea of what happened outside the orbit of the international melodies cited earlier, it may be useful to list all new melodies appearing in northern French (including Norman-Sicilian) and English manuscripts of the eleventh and twelfth century (Table 3).²⁴ In these cases, it is much more difficult to speculate about the origin of the melodies from the preserved manuscripts due to their uneven distribution over time and space. Nevertheless a roughly chronological order of the manuscripts in Table 3 will give the best available approximation.

Table 3. Further Sanctus melodies in northern French and English manuscripts of the eleventh and twelfth centuries

Note. Characterization of the first phrase: III means a ternary division; III* a division after the second sanctus (as in melody 223); II a binary division after the third sanctus; AA′ means a melodic correspondence between the first two sanctus; ABA or ABA′ a melodic correspondence between the first and third sanctus.

Some of the melodies in Table 3 can be found earlier or at the same time in other regions. Sanctus 204 and 202 (also at Prüm near the end of the tenth century) have been commented upon earlier. Sanctus 58 and 74 are clearly Italian melodies, present in numerous manuscripts from around 1100.²⁵ In addition to the manuscripts cited in Table 3, Sanctus 155 is found in St Martial (BnF lat. 909 and 1137) and in BnF lat. 887, but not in the other Aquitanian manuscripts. It seems to be a northern French melody that fell out of use in its home region before reaching manuscripts with diastematic notation.

The bottom of the table provides a short characterization of the first phrase. A quick overview of these melodies in roughly chronological order shows a gradual shift from ternary to binary division. The combination, however, of binary division and ABA scheme, as in melodies 32 and 49, does not reappear until the late twelfth century with the internationally successful melody 203 (Vat. II). Before that one can find a few binary melodies without the ABA (melody 116 is a special case due to its derivation from the antiphon O Christi pietas), and melodies with the ABA scheme, but without a clear division after the third sanctus. These cases are shown in Example 11.

Example 11. Sanctus 213, 162, 71, 58 (E-Mn 19421).

Sanctus 213 needs no further commentary. Sanctus 162 is strange in its use of tonal space: the melody eventually ends on D, as one might expect from the first sanctus invocation, but not from the rest of the melody. Dominus Deus Sabaoth is given one musical phrase, but the division before that is rather weak, since the third sanctus neither corresponds to the first nor ends on the final pitch, and the porrectus figure on the last syllable seems to imply a close continuation.

Sanctus 71 lacks the decisive word Dominus. Furthermore the connection between the three sanctus invocations that clearly imply the third mode and the rest of the melody that clearly displays the first mode is weak.

In Sanctus 58 the three words Dominus Deus Sabaoth are set as single subphrases. The cadence on the final pitch on Dominus seems stronger than the contrasting cadence on Deus. This may be construed as another ABA scheme on Dominus Deus Sabaoth, implying a binary division, or as a strong division after Dominus, implying a ternary division of the verbal phrase. Ambiguities of this sort seem to be typical for Italian Sanctus melodies.

Conclusion

The first layer of Sanctus melodies (attested before the end of the tenth century) is rather consistent regarding the ternary division of the first verbal phrase. Internal variations concern the melodic relationship between the first two sanctus invocations (perhaps indicating different preferences in East and West) and the presence of a clear division after Dominus (depending on the degree of melodic simplicity).

The second layer begins with three rather consistent melodies that appear in manuscripts c.1000. They present a binary division of the first verbal phrase and an ABA scheme for the three sanctus invocations. They do not, however, prevent the composition and circulation of new melodies that continue the melodic features of the first layer or mix features of both layers or do not fit into these categories at all.

Most later repertories of ordinary chants include representatives of both layers and both types of Sanctus melodies. This is due to the great success of some melodies of the second layer (32, 49), to the perseverance of the melodies of the basic layer (223, 41), and to the different orientation of some popular later melodies (202, 203).