Carol.

(i) Popular or courtly dance-song.

1. Origins and social setting.

(iii) The carol and the liturgy.

(iv) Household music.

(i) Popular or courtly dance-song.

The English carol is connected in name and nature with the medieval French Carole, of which the essential features are that it was a true dance-song, that it took various choreographic forms and that it was extremely popular from the mid-12th century to the mid-14th. While the carole is best documented as a courtly dance-song, popular caroles also existed. The English court tradition up to the end of the 14th century was French, and a large number of English literary references are to the courtly carole; these reflect, if in idealized form, the festivities and amusements of English courts. (Arthur and Merlin, c1335, l.1714: ‘damisels carols ledeth’; Gower, Confessio amantis, c1390, viii, 2679: ‘The hovedance and the carole / … A softe pas thei daunce and trede’; Merlin, c1450: ‘Whereas dawnsyng many maidenis were with many karoles and ryht mery song’; see Carter, Kurath, Greene 1935, Sahlin.) Many of these references specifically mention dancing with singing.

Other English references are to non-courtly carols: one of the earliest, from about 1300, is in Cursor mundi – ‘ther caroled wives be the way’; another of about the same period uses the word ‘carol’ to retell the well-known legend of the dancers of Kölbigk, condemned to dance for a whole year without stopping (Greene, 1935), and includes a phrase relating to the movement of the dance: ‘why stond we? why go we noght?’. Although a large proportion of the surviving literary references refer to the carol as a species of dance-song, courtly or otherwise, the surviving dance-song carols of medieval England have left few traces. A burden such as ‘Honnd by honnd we schulle ous take’ (Greene, 1935, no.12) suggests dancing ‘carol-wise’, but such hints are rare.

There are indications from the later 14th century that the carol could be simply a festival song, sung perhaps to the movement of a procession: ‘At the soper and after, mony athel songes / As coundutes of Krystmasse and caroles newe’ (Sir Gawain and the Green Knight, c1380). The Boar’s Head carols are surviving examples. In addition to the celebrated Caput apri defero (Greene, 1935, nos.132–5), a ‘spiritual’ version with music survives in the Ritson Manuscript (MB, iv, no.79). (For the relation between carol and conductus, see §iii below.) Another type of festival carol, associated, like the carole, with amorous games, is that of the Holly and the Ivy. The early 16th-century carol by Henry VIII, Grene growith the holy (MB, xviii, no.33, with music), is centuries removed musically from the carole but preserves the spirit of courtly game: man Holly/woman Ivy.

Both the principal early Tudor court songbooks (GB-Lbl Add.5465 and Add.31922) contain courtly carols that are more in the tradition of the earlier caroles and carols than are the compositions of the central 15th-century musical repertory: for example, the amorous Where be ye, my love, my love(MB, xviii, no.104), and the courtly-ceremonial This day day dawes (MB, xxxvi, no.65) celebrating the union of the White and Red Roses (perhaps the marriage of Henry VII to Elizabeth of York). A type of carol making use of dramatic possibilities (i.e. the circle, the alternating leader and chorus, etc.) is represented by Go day, go day: Go day syre Cristemas (MB, iv, no.18), although there is no actual verbal or musical dialogue.

(ii) Popular religious song: the monophonic carol.

About five out of every six surviving carol texts treat wholly religious or morally didactic subjects in accord with Christian precepts (Greene, 1935, p.cxi). One text serves to illustrate the principal features of the popular religious carol:

To blis God bryng us all and sum
Christe Redemptor omnium.
1. In Bedlem, in that fayer cyte,
A chyld was born of Owr Lady,
Lord and Prynce that he shuld be
A solis ortus cardine.
2. Chyldren were slayn grett plente,
Jhesu, for the love of the;
Lett us neuer dampned be.
Hostes Herodes ympie.
3. He was born of Owr Lady
Without wembe of her body,
Godes Son that syttyth on hye
Jhesu salvator seculi.
4. As the son shynyth thorow the glas,
So Jhesu in her body was;
To serve hym he geve us grace
O lux beata Trinitas.
5. Now ys born owr Lord Jhesus,
That mad mery all us;
Be all mery in thys howse;
Exultet celum laudibus.

This text is typical of the carol as a popular religious song: (i) it pertains to the Christmas season and honours the Blessed Virgin with special reference to the mystery of virgin birth; (ii) it is simple, direct and unpretentious in style and uses stock phrases; (iii) it is macaronic, employing Latin as well as English, the Latin lines being taken at random from liturgical hymns and antiphons but woven with evident care into the structure of sense; (iv) it employs traditional imagery (for example, the sun shining through the glass is a theological commonplace); (v) it is in the standard form: a two-line burden alternates with a four-line verse, the lines of four stresses each, rhythmically somewhat rough, rhyming aaab.

That this is popular poetry and not folk poetry is obvious. The oblique, terse and often deeply imaginative poetry of such a religious folksong as The Bitter Withy (in its authentic form totally un-Christian) is remote from the rough-and-ready, prosaic, orthodox assurance of carols like To blis God bryng: In Bedlem. The genuine ‘folk’ touch is rare in the huge repertory of the medieval carol, with one striking exception – the well-known, mysterious Corpus Christi carol (Greene, 1935, no.322; no contemporary setting) which defies all attempts at precise interpretation. In religious feeling the 15th-century English carol belongs with the popular religious drama of the same period (see Medieval drama, §II). Drama and carol are similar in that both were written anonymously by men of some education and word-craft but no great intellectual pretensions (possibly laymen or minor clerics) for the enjoyment and edification of ordinary people with whom they shared unquestioning assurance in the Catholic faith. Specific links between carol and drama are apparent in such carols as Marvel not, Joseph (MB, iv, no.81, an extended dialogue between the worried Joseph and a reassuring angel) and perhaps With al the reverens (Greene, 1935, no.108, a vivid depiction of the slaughter of the Innocents).

To describe any carol as a popular religious song is not to imply any simple, single history for it. Four different texts exist for To blis God bryng: In Bedlem (Greene, 1935, no.21A–D), and a comparison strongly suggests that the same song was regarded as having different functions – the variations make it appropriate to a festive secular occasion, a popular religious procession or private devotional use. Furthermore, the two contemporary musical settings are different and unrelated.

Two important facets of the medieval religious carol are its relation to religious dance and popular song, and its connection with the activities of the Franciscan friars. It is not as absurd as it may sound to imagine festive religious songs being danced to. At Sens Cathedral the clergy were permitted by regulation to dance, provided they did not lift their feet off the ground (‘non tamen saliendo’; Office de Pierre Corbeil, 13th century, ed. H. Villetard, Bibliothèque musicologique, iv, Paris, 1907). An attractive group of monophonic Latin songs in a 13th-century French manuscript (I-Fl Plut.29.1) has been described as consisting of ‘danses ecclésiastiques’ (Rokseth). Some refer to particular feasts, others are liturgically less specific. Apart from the general tone of communal joy that they share with the contemporary English carol, they are of interest for their forms: they are rondelli or rotundelli, taking the form of the simplest French rondeau – aAbB or aAabAB. They thus relate to the carole-derived complex of forms already described. Another collection of festive Latin songs, the Red Book of Ossory, compiled for his clergy by the 14th-century Irish bishop Richard de Ledrede, includes ‘cantilenae’ to be sung at great feasts and on occasions of relaxation ‘lest the throats and mouths [of the brethren] sanctified by God should be defiled by theatrical and worldly songs’. In both form and content their texts correspond closely to those of the ‘danses ecclésiastiques’ and of the carol repertory.

The Red Book of Ossory has further interest for the history of the carol, for the bishop was a Franciscan friar. There is much evidence that the Franciscans fostered vernacular religious song on the Continent, especially in Italy (see Lauda spirituale); the connection between the Franciscans and the English carol is less well documented, though firmly established. It depends on the evidence of particular names and manuscripts rather than on a continuous tradition. These include: Friar William Herebert (d 1333; two of his translated liturgical pieces are in carol form); the Kildare collection of Anglo-Irish poems (c1300); the commonplace-book of Friar Johan de Grimestone (compiled 1372); and the voluminous versifyings of Friar James Ryman (copied by 1492; see Greene, 1935, introduction; Robbins, 1938). The proposed link between the English carol and the Franciscan community is supported by the fact that three of the ten fully surviving monophonic carols have texts that appear in Franciscan manuscripts: Lullay, lullay: As I lay (MB, iv, no.1A), for example, exists in a longer version in Grimestone’s commonplace-book, and two others appear in the manuscript of Ryman’s verse. The musical nature of the Franciscan song tradition in England can be inferred from its essential purpose, which was to edify by entertainment and to take some of the Devil’s good songs away and give them back to God. Thus the Franciscans fostered a tradition of sacred contrafacta, rendering secular songs into sacred simply by changing their texts; the marginalia of the Red Book of Ossory bear witness to this tradition, for they name lost and presumably popular songs such as ‘Have godday my lemon’, ‘Maiden in the mor lay’ and ‘Hey how the chelvaldoures wokes al nyght’, for the singing of the festive Latin verses.

There is ample testimony to the existence of a large repertory of popular melodies to which carols could be sung. Two carols in a manuscript at Gonville and Caius College, Cambridge, for example, are labelled with the names of tunes – ‘Bryd on brere’ and ‘Le bon l.don’; another carol, Hey now now now (Greene, 1935, no.93) is headed ‘A song to the tune of and I were a mayd’ (the early 16th-century court songbook GB-Lbl Add.31922 contains a five-part song based on And I war a maiden, possibly the same tune). But the only surviving musical witnesses to the repertory of popular religious song are six manuscripts (all miscellanies including some learned matter) containing a total of ten monophonic carols (MB, iv, nos.1A–10A). The carol melodies in these manuscripts are simple, rhythmically balanced, syllabic and restricted in range; most are crudely notated. Some are lively and possibly secular in origin (e.g. MB, iv, no.5A, Nova nova); others have the movement of plainchant (e.g. no.8A, Of thy mercy). It has been suggested that the tenors of some polyphonic carols (like those of Burgundian chansons) were adapted by 15th-century singers as monophonic tunes in their own right, but application of this theory to the carol rests rather on analogy than on evidence (Bukofzer, 1950). Only one of the ten surviving carols (Salve sancta parens, MB, iv, no.6A) is clearly a voice part isolated from its proper context.

This scrappy picture of the relation between popular song and the carol may be filled in in several ways. Very occasionally, as seen above, a particular tune is mentioned for singing a carol. Alternatively, a later religious partsong setting may embody a pre-existing carol melody (see Greene, 1935, no.150D; music in GB-Lbl Roy.App.58 for three voices, the ‘tune’ being in triple metre, the polyphonic setting in duple). The unison passages in early 15th-century polyphonic carols also may indicate the style and perhaps the actual melodies of lost monophonic carols (see Nowell: Out of your sleep, MB, iv, no.25, burden; Alleluia: A newë work, no.30, verse). Finally, the music of the liturgy itself, particularly hymns and litanies, may have contributed the missing melodies. Since carol writers often borrowed their Latin lines from hymns, and since most hymns, like carols, are written in four-line stanzas with four stresses to the line, it seems possible that some carols were sung to plainchant hymn melodies. Evidence of the singing of the popular religious carol is tantalizingly incomplete; but the existence of such a genre cannot be doubted.

(iii) The carol and the liturgy.

A number of carols (notably those of the Ritson Manuscript) have rubrics such as ‘in die nativitatis’, ‘de sancta Maria’ etc. What relation, if any, did carols have to the official liturgy? Could they, for example, have been used as processional hymns, or were they vernacular substitutes for some other part of the liturgy? One of the principal arguments for regarding the carol (whether a monophonic song for the laity or a polyphonic piece for professional singers) as a processional piece is a formal one. The structure of the carol is like that of the processional hymn with repetenda(repeated sections), such as the Palm Sunday hymn Gloria laus et honor(Bukofzer, 1954 [see Stevens, 1952]). Other evidence for the use of the carol as a processional song consists in the use of burdens and refrains taken from the Processional; for example, the carol for St Stephen, Pray for us that we saved be: Protomartir Stephane (MB, iv, no.92) ‘incorporates the exact metre and a line of the response sung in the procession to his altar after Vespers on Christmas Day’ (Robbins, Studies in Philology, 1959). It has also been argued that when the vernacular was introduced into the Latin service the most natural place to put it was into a processional hymn, because processions took place around the church, not ‘in choir’, and direct English translations of liturgical processions exist from the mid-15th century. Further, there may be a special connection between the popular religious carol and the litany (Sahlin). Among the carols described as ‘popular litanies for use in liturgical processions’ are Greene, 1935, nos.91, 103, 220 and 311. The characteristic people’s response is a burden like ‘Jhesu fili virginis / Miserere nobis’.

Apart from the formal resemblance to the processional hymn, several other factors combine to suggest that the 15th-century polyphonic carol was a processional form: three of the four more important musical manuscripts containing carols also contain pieces from the processional repertory; the carol resembles the Conductus (itself a processional form) in its note-against-note style, its syllabic treatment of the text, its presentation in score, the absence of cantus firmi, etc.; the headings like ‘in die nativitatis’, or ‘de sancto thoma’ appearing with certain carols are not necessarily prescriptive rubrics but indicate at least a strong sense of liturgical season. If the religious and didactic carols of the 14th and 15th centuries were adapted by the church from secular usage as an ornament of the liturgy’s processional rites, then the various forms and manifestations of the carole and carol can be resolved into a coherent relationship: all share an association with movement, whether dance or procession, a division into burden and verse (chorus and soloist, people and priest) and a use of burdens and refrains of an ejaculatory kind.

The investigation of liturgical service books, processionals in particular, has not yet, however, produced a single instance of a carol or a vernacular song being expressly required or permitted in processional rites; and the ordinals do in fact prescribe the chants to be sung in processions during the whole year. This has led to a different hypothesis about the relation of the carol to the liturgy (HarrisonMMB). Harrison has argued that the carol was a permissible substitute for the second Benedicamus at the Offices on the three days after Christmas (feasts of St Stephen, St John and the Holy Innocents), Circumcision and Epiphany; Christmas was traditionally a season of liturgical licence, especially in English secular (i.e. non-monastic) cathedrals. Harrison’s thesis is supported by the discovery in the mid-20th century of four previously unknown Latin carols together with a group of Benedicamus settings in a Gradual at Aosta (Fischer; Harrison, 1965). Further, a large number of carols celebrate the events and saints of Christmas, and many have burdens incorporating the ‘Deo gratias’ response to the Benedicamus. However, the proposed relation between carol and Benedicamus in itself implies a close association between the carol and Christmas processions, since, for example, the procession to the altar of St Stephen after Vespers on Christmas Day was preceded and followed by the singing of the Benedicamus. Until more evidence becomes available the question must remain open as to whether the medieval carol was admitted into the liturgy or kept peripheral.

(iv) Household music.

The categories so far described do not cover all the surviving carols from before 1550; there remains a number of pieces from the late 15th and early 16th centuries whose function is not clear, the most important being those of the Fayrfax Manuscript (GB-Lbl Add.5465). The most impressive group musically, far surpassing earlier carols in scope and power, are the Passion carols of this manuscript, by Browne, William Cornysh, Banaster and Davy (MB, xxxvi, nos.49–58, 68–9). Their texts are devotional and meditative, not joyous and celebratory – monologues or dialogues, spoken by Christ or to him by a penitent, questioning sinner. Even the apparent lullaby carol (MB, xxxvi, no.50, A my dere a my dere son) soon turns into a discourse by the infant Christ on his Passion to come – ‘Many a wound / suffer shall I’. The very fact that they are centred on the Passion and Crucifixion distinguishes these carols from those of the main 15th-century repertory, which honour the Virgin, the saints of Christmas and the Incarnation. Although it is not easy to conceive the precise occasions on which they were sung, they seem to be devotional music for household use at court and were probably sung by professionals. The other carols of the manuscript support this hypothesis; they are mainly courtly-political (e.g. I love, I love, and whom love ye?, on the union of White and Red Roses, MB, xxxvi, no.47, and From stormy wyndes, a prayer for the safety of Arthur, Prince of Wales, MB, xxxvi, no.64) and courtly-satirical (e.g. Jhoon is sike and ill at ease, MB, xxxvi, no.60). It is also conceivable that the religious carols of this songbook were used as vernacular substitutes for Latin votive antiphons, to be sung before the altars of private chapels. The carols of GB-Lbl Add.31922, lighter in literary and musical style than those of the Fayrfax Manuscript, fall into a similar pattern. Only one religious carol is included, Richard Pygott’s fine Quid petis o fili (MB, xviii, no.105). The courtly carols include the stridently patriotic England be glad: Pluk up(MB, xviii, no.96) and the lightly amorous What remedy, what remedy (MB, xviii, no.103).

Many carols both from the 15th-century repertory and from early Tudor songbooks are appropriate to secular ceremonies, in which the harsh realities of social life were obscured and its high aspirations expressed. One particular occasion for an exchange of courtesies between a great man and his servants was New Year’s Day. In early Tudor times, for instance, it was customary for the Chapel Royal to receive £13 6s. 8d. on that day and for prominent members to give the king presents. A salutation from musicians at the chamber door was the rule in the Northumberland household: the steward paid ‘his Lordshipis vj Trompettes when they doo play at my Lords Chaumbre Dour the said Newe Yersday in the morninge, xx.s’; they played at the doors of other members of his family as well. The chapel also may have had this privilege in some households (MB, iv, no.62). Medieval feasts, by no means the hearty, convivial affairs of popular imagination, also provided opportunities for carol performance. On these occasions, the highest nobility, retainers on horseback and trumpeters, all took part in the ‘honourable service’ which was the due of their royal or lordly host, and it seems quite possible that the moralized Boar’s Head carol (MB, iv, no.79) may have been used on such a solemn occasion. The Egerton Manuscript contains, in the carol section, a highly sophisticated setting of the goliard song O potores exquisiti; a carol (Comedentes convenite, MB, iv, no.71), a companion-piece to it, is apparently an invocation to women feasters to make themselves ready.

These are obvious examples of banquet music. Other carols are appropriate to the entertainments or to the ‘void’ (a light refreshment of wine and spices) which habitually followed a formal banquet, for example, on Twelfth Night (see the regulations for the Royal Household, 1494). There were doubtless ‘good songs’ written specially for this ceremonial occasion. Other occasions of a similar kind may well have been served by such semi-dramatic carols as Nowell (MB, iv, no.80), which welcomes ‘Sire Christemasse’ and exhorts those assembled to ‘Buvez bien par toute la compagnie’. Earlier carols are susceptible of dramatic presentation, as for instance Go day, go day: Go day syre Cristemas (MB, iv, no.18) and What tydynges bryngest thou messanger? (MB, iv, nos.11 and 27). Such a carol as Ivy ys good and glad to se (MB, iv, no.55) is less easy to place in a precise social context; possibly it belonged to the music for an elaborate courtly game related to a folk custom (like courtly May games).

There remains a large group of political carols. The best-known is also the earliest, the Agincourt carol (Deo gracias Anglia), popularly believed to have been sung by soldiers on the battlefield, but in fact an elaborate and sophisticated piece of responsorial music. A ceremonial occasion that may have provided a particularly apt setting for it was the lavish civic reception given to Henry V on his return to London. This ‘royal entry’ called forth all the glamorous and expensive ‘sights’ the City could devise, including a chorus of beautiful virgins singing from a castle ‘Welcome Henry the Fifte, kynge of Englond and of Fraunce’. It was common for the corporation to borrow skilled singers and players from the royal household or important churches in order to augment their ‘triumph’; but this carol might well have been sung by the Chapel Royal in the king’s procession. 130 years later Edward VI was welcomed at the Little Conduit in Cheap with a carol, Sing up heart (Greene, 1935, no.438). Other political carols include Anglia, tibi turbidas (MB, iv, no.56), Enfors we us with alle our myght (MB, iv, no.60), England be glad (MB, xviii, no.96) and From stormy wyndis (MB, xxxvi, no.64).

A strict division of the carol repertory into sacred and secular is impossible and inappropriate, for people in the Middle Ages did not feel such a distinction. Enfors we us may just as well have been sung before the altar of St George in a cathedral as in an official procession; and the carols of moral and political counsel would not be out of place in a solemn service. Moreover, those ‘of his lordschipes chapell’ who played ‘the play of the nativite uppon Christynmes-day in the morninge in my lord’s chapell before his lordship’ could have used, say, What tydynges bryngest thou messanger?

Like the religious carol, the household carol can be described as processional music. In both its ecclesiastical and its aristocratic milieu the carol seems to have retained its traditional association with bodily movement. When it was not danced to it could be processed to; and on the many occasions when a procession was the nucleus of a civic, aristocratic or clerkly ceremony carols may have been sung. The carol, like much other medieval music, was a ceremonial agent. Whatever its origins or the purpose of its adoption by the church, the polyphonic carol was a highly polished and sophisticated ornament of ecclesiastical and aristocratic ceremonies.

(i) Sources and repertory.

2. The pre-Reformation carol.

(ii) Form.

(iii) Rhythm, melody and harmony.

(iv) Underlay and word-setting.

(i) Sources and repertory.

There are six principal sources of polyphonic carols. In chronological order they are: GB-Ctc 0.3.58, the ‘Trinity Roll’, early 15th century; Ob Selden b.26, early 15th century; Lbl Eg.3307, c1440; Lbl Add.5665, the ‘Ritson Manuscript’, late 15th–early 16th century; Lbl Add.5465, the ‘Fayrfax Manuscript’, c1500; Lbl Add.31922, ‘Henry VIII’s Manuscript’, c1515. (For descriptions of these manuscripts see Sources, ms, §IX, 3 and 4.) All the manuscripts of polyphonic carols are associated with large choral establishments. The carols in the earlier manuscripts demand for their performance a choir of perhaps nine or ten adult male voices, with special strength in the middle register. The carols of the Fayrfax Manuscript sometimes require trebles (e.g. Affraid alas and whi so sodenli?, MB, xxxvi, no.52); those of Henry VIII’s Manuscript have varied requirements, and some could be domestic music for amateurs. The non-carol music in the Selden, Egerton and Ritson manuscripts confirms the need for a large body of trained singers, not all adult; the rubrics of the Sarum processional, from which many of the texts are taken, require that the pieces (even the plainchant settings) be sung by ‘tres clerici de superiori gradu’, ‘tres pueri’ and so on.

The first four of these manuscripts contain a carol repertory with internal connections, but one that becomes progressively isolated from the tradition of the 15th-century carol as a popular religious song. A close connection between the surviving monophonic carols (see §1(ii)) and the ‘literary’ repertory (i.e. carol texts without written music) is suggested by the fact that six out of ten monophonic pieces had an independent literary existence, five of them in more than two manuscripts. There is a comparable connection between the earliest polyphonic source, the Trinity Roll, and the literary tradition, for six of the 13 carols survive in literary as well as musical sources. But the link became more tenuous as the century progressed: only a quarter of the carols in the Selden Manuscript have a literary counterpart, only one in eight of all the carols in the Egerton and Ritson manuscripts; and the two early Tudor songbooks, the Fayrfax Manuscript and Henry VIII’s Manuscript, are totally unconnected with the main carol repertory, either in its popular ‘literary’ or in its specialized professional musical development. Indeed, the lack of integral connection between the 15th-century carol and the early Tudor repertory is evident in virtually every literary and musical aspect, and even in their manner of presentation in the manuscripts. The earlier carols are invariably written in score, with the words placed under only the lowest part, the tenor. In contrast, the carols of the Fayrfax Manuscript and Henry VIII’s Manuscript are notated exactly like the other songs in those manuscripts, in a small choirbook format: each part is written separately with its words, and the parts are on facing pages of the book.

(ii) Form.

Nothing could be more straightforward than the form of the earliest polyphonic carols, those in the Trinity Roll. Their alternation of burden and verse clearly reflects the division of the medieval carole (the dance-song) into chorus and leader. But the next stage of development is already apparent in the seventh carol of the roll (the Agincourt carol), which is the earliest surviving carol with two distinct burdens – one for soloists, one for chorus. 11 out of 28 carols in the Selden Manuscript have double burdens, including six of the last nine. The Egerton Manuscript has a slightly higher proportion, half and half, and the Ritson Manuscript consists entirely of carols with two burdens. In the Ritson Manuscript the dramatic possibilities of dividing a long burden into alternating phrases for chorus and soloists are clearly realized. The burden of MB, iv, no.80, for instance, has four sections:

soloists: Nowell, nowell
chorus: Who is there that singeth so: Nowell?
soloists: I am here, Sire Christesmas.
chorus: Welcome my lord, Sire Christesmas! Welcome to us all, both more and less! Come near, Nowell.

The traditional division of labour between chorus and leader was obscured also in the verse, since a feature of the polyphonic carol was the interpolation by the chorus of short three-part phrases into the two-part solo writing of the verse. These phrases usually repeat or slightly vary the words and music of the solo phrases immediately preceding them. They do not occur, at least not in their written-out form, in the Trinity Roll, but the Selden Manuscript has five instances, of which Alleluia: A newë work (MB, iv, no.30) is the most complex: the verse is divided into four sections, the first three of which are repeated by the chorus (VC VC VC V). Examples do occur of the last phrase being repeated by the chorus (e.g. Syng we to this mery companey, MB, iv, no.76), but more usually it is the penultimate phrase, or an earlier one that is so repeated.

One of the problems facing the student and singer of carols with two burdens is to decide in what order the sections should be sung. Two questions must be asked: did the carol start with one burden or with both? Were both burdens repeated between the verses? The essence of carol form is the alternation of burden and verse, and just as any verse may be extended by the insertion of a chorus section, so any burden may be extended by doubling it with a second or ‘chorus-burden’. There are therefore only two permissible standard orders: BV¹BV²BV³… B; and B¹B²V¹B¹B²V² … B¹B². The order for the more complex carols (B¹B²V etc.) seems the final rejection of the basic chorus–leader–chorus arrangement; but it is the natural and right way of performing this essentially responsorial music, and it is the order of the processional hymn. What now appears an excessive amount of repetition is characteristic of the medieval formes fixes and may have seemed both normal and necessary in the 15th century.

A comprehensive formal description of the 15th-century carol would have to take in the musical relationships, varied and subtle, between verse and burden, burden and chorus-burden, verse and chorus-section. The most important single unifying device is the use of musical rhyme between the burden and the end of the verse (e.g. Hayl Godys Sone, MB, iv, no.33, bars 13–15 with 38–40). When there are two burdens, they may be linked through the melody of the upper voice (see Ave Maria, MB, iv, no.36, where the key phrase is given first by a single voice, then repeated by the upper voice at the opening of a two-part burden and again at the opening of a three-part chorus-burden) or the melody of the lower voice (Almyghty Jhesu, MB, iv, no.64). Connections between verse and chorus-section on the other hand are usually very close: in Almyghty Jhesu, for example, bars 26–30 simply add a middle voice to 21–5; and in Nowell, nowell: The boarës head (MB, iv, no.79) bars 45–7 add a lower voice and vary the treble cadence. (For a full analysis of carol form based on the Egerton Manuscript see Bukofzer, 1950.)

The later history of the polyphonic carol, as witnessed by the early Tudor manuscripts, shows radical modifications of 15th-century carol form. These are of three main types: (i) the composition of new music for each verse, abandoning the strophic principle; (ii) the alteration, and sometimes shortening, of the burden when it is repeated at the end of the verse; and (iii) the complete suppression of the burden after its initial statement.

The first type of modification occurs frequently in the carols of the Fayrfax Manuscript. Sheryngham’s A gentill Jhesu (MB, xxxvi, no.54) follows the traditional form precisely, but neither Browne’s nor Cornysh’s setting of Woffully araid (MB, xxxvi, nos.55 and 53) is strophic; the verses are through-composed in each case. The same is true of Pygott’s Quid petis (MB, xviii, no.105). Banaster’s My feerfull dreme (MB, xxxvi, no.56) is particularly interesting because it comprises the two procedures: the first and second verses are identical musically, as are the third and fourth, but the two pairs differ slightly. The impulse behind this modification may have been the growing feeling that different words somehow required different music – then a revolutionary idea.

The second type of modification, alteration of the burden on repetition, occurs in many carols of the Fayrfax Manuscript and Henry VIII’s Manuscript. Thomas Phelyppis’s I love, I love, and whom love ye?(MB, xxxvi, no.47) repeats the whole burden and preserves its formal entity, but its texture is altered (ex.1). The long Passion carols are the most heavily modified: in Browne’s Jhesu mercy (MB, xxxvi, no.51) the first 15 bars of the initial burden are never repeated; the through-composed verses merge without pause or cadence into the music of the burden, picking it up at the 16th bar and indicating the repeat of bars 16–31 without writing it out in full (see also England be glad, MB, xviii, no.96). The third type of modification, the suppression of the repeated burden, advances the process one step. Instead of a return to a shortened burden, a musical and textual refrain is integrated into the end of the verse; the refrain usually consists of music from the initial burden. Clear examples are the two settings of Woffully araid, mentioned above as being through-composed. Browne’s three-voice setting uses both the words and the precise music of the last six bars of a 26-bar burden as the unchanging refrain of each new verse; Cornysh’s four-voice setting goes even further and makes a variable refrain, without precise repetition, out of music which is first heard at the end of the initial burden (cf the descending phrase to the words Woffully araid, beginning at bar 27, tenor). Small-scale examples of the same procedure are found in Henry VIII’s Manuscript (e.g. Where be ye, my love?, MB, xviii, no.104, with precise repetition of six bars, and What remedy?, MB, xviii, no.103, with only a musical reminiscence of two or three bars but a genuine textual refrain).

(iii) Rhythm, melody and harmony.

If form distinguished the carol as a genre in the late Middle Ages, a certain style of rhythm and melody was the common trait of 15th-century carols distinguishing them from other music of the time. Bukofzer (1950) described the essence of carol style as the ‘interaction between angular design and rhythmic vigour’; it is this that gives the carols their characteristic brisk gaiety, freshness and lilt.

The 15th-century carols are, with few exceptions, written either in ‘major prolation’ (C, transcribed as 6/8 or 3/8) or in ‘perfect time’ (, transcribed as 3/4; see Notation, §III, 3). Major prolation is the rule in the Trinity Roll; perfect time is increasingly represented in the Selden and Egerton manuscripts and is the sole metre of the Ritson Manuscript. The significance of this mensural change can be seen in a comparison of the terse and vigorous carols of the Trinity Roll with the sedate and comparatively turgid carols of the Ritson Manuscript. The solemnity of many later carols is partly due to the greater rhythmic complexities encouraged by perfect time. The choice of notational metre cannot, however, have been thought a matter of decisive importance in the middle of the century because the Agincourt carol is found in both. Furthermore, the characteristic cross-rhythms of carol music were as easily expressed in major prolation as in perfect time. These cross-rhythms are of two kinds: the first (shown as ex.2a) is usually, but not always, expressed by coloration; the second (ex.2b) does not require coloured notation. These rhythmic shifts of emphasis are also found in monophonic carols. Contrasting patterns of minim–crotchet and crotchet–minim rhythms are another characteristic of both early and late carol music.

It is not easy to improve on Bukofzer’s description of carol melodies as ‘paradoxically both smooth and angular’. Some distinctions can, however, be observed between treble and tenor melodic style. The first are more highly figured, inclined to use syncopations and repetitions, while the second are sturdier and more firmly based on alternating patterns of minim–crotchet and crotchet–minim. There is a tendency, particularly in the verse, for the tenor to start at the top of the compass and to fall; this undoubtedly gives a characteristic flavour to some melodies. In both voices rising phrases tend to be swift and abruptly disjunct, descending phrases less so.

The myth that the tenors of polyphonic carols are folktunes has been thoroughly demolished. It has been suggested at the other extreme that the melodies had an independent existence only after their composition for the polyphonic carols with which they survive (Bukofzer, 1950). But the presence in the earliest carols of monophonic or unison passages is a reminder that in the first decades of the 15th century the carol tradition was still comparatively homogeneous, and that there was still a link between the polyphonic carol and the popular singing tradition. It is possible that tunes were composed ‘according to the rules of art’ for monophonic performance and later used for polyphonic settings. The tenors of the polyphonic Nowell: Out of your sleep(MB, iv, no.25), and of the refrain-song Omnes una gaudeamus (MB, iv, no.15A), and the upper part shared by the three settings of Ecce quod natura (MB, iv, nos.37, 43, 63) could have evolved thus. Whether or not this is so, most 15th-century carols probably had tunes that were neither folksongs nor tenors of partsongs.

The harmonic basis of the 15th-century carol is the gymel or cantus gemellus. In this style, common even as late as the Ritson Manuscript, two equal voices move and cross to weave a texture of unisons, 3rds, 6ths and 10ths. But many carols, particularly early ones, use parallel 6ths, separating at cadences to octave and unison, so consistently that it is only natural to assume that instructed musicians would sometimes have sung appropriate sections with an improvised middle part in ‘English discant’ or fauxbourdon style (but see Bukofzer’s review of Stevens, 1952). The alternation of monophonic and ‘discant’ sections in Alleluia: A newë work suggests that the monophonic carols also may have been the subject of improvisation. The harmonic system of parallel 3rds and 6ths based on this extempore technique was a familiar basis for three-part composition in the 15th century; and even in the Ritson Manuscript, where the harmonic and rhythmic freedom of the medius(or countertenor) is most marked, passages of strict ‘discant’ are not unknown (e.g. Alleluia: Now may we myrthis make, MB, iv, no.105). The occasional addition of a middle voice in the earlier two-part carols seems reasonable, even when the manuscripts give no indication of it, because one of the essentials of carol-style in the mid- and late 15th century was the frequent contrast of two- and three-part writing and of soloists and chorus. Furthermore, the direction ‘Fa-burden’ occurring in Te Deum laudamus: O blessed God in the Ritson Manuscript (MB, iv, no.95) clearly indicates that parts were in fact improvised.

The growing freedom of the medius to form a harmonic bass has already been noted. Perfect cadences are found in both the Egerton and Ritson manuscripts. But so much was the carol style felt to have marked harmonic and rhythmic features, that even the latest carols have the traditional tenor and discant cadence; the Landini or ‘under-3rd’ cadence was still also common.

A full examination of the use of modes in the 15th-century carol cannot be undertaken here: the following remarks are based solely on an analysis of the cadences of burdens (where there are two burdens the finals are almost invariably the same). As the verse frequently gravitates into a different mode or ‘air’, it is the burden that may be regarded as setting the ‘air’ for the piece. The most striking observation is the popularity of the C mode, the modus lascivus. Over half the carols of the Selden and Ritson manuscripts close on C or, transposed, on F. Next most popular was the D mode, which appeared twice as often in its untransposed form as in its transposed. The composers of the Trinity Roll and the Egerton Manuscript had a particular liking for this mode and used it as much as the C mode. The first carol of the Trinity Roll is the only one to employ A as final; the A mode is moderately popular only in its transposed form, occurring in about one tenth of the carols. The E mode frequently appears, particularly transposed, as an intermediate cadence, but never as a final cadence. The remaining tenth of the carols are in the G mode, equally divided between those actually ending on G and those transposed to C.

Even the fairly compact, self-contained repertory of the 15th-century carol did not exist in a musical vacuum. Many of its stylistic traits are found in other small polyphonic forms of the century – in particular, in antiphons and in the responds and hymns of the Selden and Egerton manuscripts. For example, the English version (in Selden) of the sequence Laetabundus exultet fidelis chorus, Glad and blithe mote thou be, shares with the carol such stylistic features as a combination of iambic and trochaic rhythms, a harmonic style based essentially on parallel 6-3 progressions with unison and octave cadences, a delicate rhythmic interplay between discant and tenor making much use of extended or compact cross-rhythms, and a vigorous verbal pulse, particularly at the beginnings of phrases. (For the small-scale non-carol music of the mid-century, see the editions of the Selden and Egerton manuscripts by Hughes and McPeek respectively.)

The carols of the two songbooks from the Tudor court (the Fayrfax Manuscript and Henry VIII’s Manuscript) give a much stronger impression that the carol form is, so to speak, accident rather than substance. Throughout the 15th century the form of carol always implied a certain style of rhythm and melody as well, but those of the early Tudor court were court music first and carols thereafter. Between the earliest 15th-century carols of the Trinity Roll and the later elaborated carols of the Ritson Manuscript there is an organic stylistic connection, between the Ritson carols and those of the Tudor songbooks none at all. A simple example of the Tudor carol is Alone, alone: As I me walkyd(MB, xxxvi, no.49); it is for three voices, yet often only two are singing: in earlier carols such changes in texture were purely formal and distinguished the sections, but here the principle is simply that of musical variety (ex.3). Most striking of all the new features is the great increase in rhythmic flexibility conferred by duple metre, a flexibility essential to the new style of word-setting. The cadences no longer conform to stereotyped patterns, nor are the internal cadences of sections so clearly defined. The harmonic style is not so advanced here as in the massive four-part pieces of the manuscript with their firmly conceived bass parts, and the style is still melodically conceived, but the 6-3 progression has been abandoned.

The three amorous carols towards the end of Henry VIII’s Manuscript are typical of the light ‘courtly-popular’ song – chordal in conception with a clearly defined bass and firm tonality, and consisting of lightly patterned conventional rhythms (ex.4). The only individual carol in the manuscript is the lone religious one, Pygott’s Quid petis, a curious amalgam of styles. Its burden is in well-organized, open, imitative counterpoint with two clearly distinguishable points; the verses have some characteristics more frequently found in contemporary church music, especially the second verse, an extended and rhythmically complex duet for treble and bass, like a solo section from a big polyphonic antiphon.

(iv) Underlay and word-setting.

The words of 15th-century carols raise two separate problems: first, the way composers treated the text in their music; second, the way scribes presented it in the manuscripts. Carol scribes did not, fortunately, adopt the tantalizing practice common in some manuscripts of noting down only the first words of each poem. On the contrary, with one exception (MB, iv, no.74) each piece of music is amply provided with text; normally only the first verse is underlaid, to the tenor (the lowest part in the score). The detail of fitting the words to the music was left to the singer. In the Ritson Manuscript, for instance, he had more to do than in earlier manuscripts, partly because the manuscript is less tidily written than the others, partly because the musical style is more complicated. On the whole a general agreement of verbal with musical phrase is all that most scribes attempted, and even that cannot be relied on. There are, however, several places where, quite inexplicably, the exact underlay is indicated by thin lines drawn from the words to the notes. The carol sources present the full range of problems and inconsistencies common to most 15th-century manuscripts.

In the earlier and simpler carols the singer’s task was quite straightforward: the words were set syllabically and forced, without much regard for natural stress, into the metrical straitjacket of the music. Sometimes a short melisma was reserved for the end of a phrase or for a conventional word like ‘Good-day’ or ‘Alleluia’. Carols of the middle period, especially those written in perfect time, show a slightly different treatment which may best be described as ‘metrical’. The first five or six syllables are generally set to as many notes, but after that each syllable, regardless of just accent, is set to a strong beat in the music (usually the beginning of a bar in transcription). In a long phrase the later syllables may be spread at intervals of two bars or more (e.g. MB, iv, no.53). This use of words to underline the rhythmic vigour of the music is exactly what one would expect in the carol. Each syllable seems to administer, as it were, a little punch to the melodic line. Only in the later carols of the Ritson Manuscript is the principle modified. There, after the syllabic beginning which remains characteristic, the increasingly florid melodies carry the syllables along with comparatively little regard for the rhythmic lilt of the music, which is, partly for this reason, less emphatic. Change of syllable is introduced to mark a new phrase in the melody or a change in its direction. Late though these carols are, there is little more attempt in them than in the earlier ones to draw out the inner meaning of the words. Only occasionally, as in the penitential carol To many a will have Y gone (MB, iv, no.114), does the composer seem to consider the text he is setting. Word-painting is unknown in the 15th-century carol; simple declamatory passages are rare but not ruled out, at least in the Ritson Manuscript (see Mervele noght Joseph on Mary mylde, MB, iv, no.81, especially the first verse).

Once again, the carols of the Tudor songbooks can be sharply distinguished from the earlier carols and not at all sharply distinguished from the songs that occur alongside them in the same manuscripts. The Fayrfax Manuscript is, indeed, from the point of view of underlay and word-setting, a truly remarkable document. For the first time in English song, to judge from the surviving manuscripts, serious and detailed attention was given by both composers and scribes to the way words and notes were to be related. The most striking symptom of this is the composer’s careful representation of physical word-sounds in melodic and rhythmic shapes (ex.5). Even this short passage shows that there are other purely musical considerations which prevent the words being sensitively represented at every occurrence; but the contrast between such a passage and the dispassionate metrical procedures of the 15th-century carol is obvious.

In a number of instances the kind of attention given seems to go beyond the aural – the end of the burden of A gentill Jhesu (ex.6) carries the full force of impassioned speech in the words ‘Ah, I will, I will’ with its agitated quavers. Other passages, again, seem to earn the epithet ‘expressive’ for their evocative power. Thus in Cornysh’s Woffully araid (MB, xxxvi, no.53), bar 75, the setting of ‘Thus nakyd am I nailid’ is enhanced by the rare use of a diminished 4th in the uppermost voice. But essentially the novelty of the Tudor carol is its very careful attention to the sounds of words and the shape of short word-groups, attention so detailed that it was thought necessary to write out at length the verses even of strophic carols, with the most minute notational variations required for each different set of words. In Henry VIII’s Manuscript the underlay is far less careful and the word-setting arbitrary, or at best conforming to recognized conventions. In the more syllabic carols (as in the other court songs of the manuscript), the composer’s intentions are not often in serious doubt, not because the scribes took more care, but simply because the musical styles are straightforward.

3. The post-Reformation carol.

After 1550 the changes brought about in Britain by the Reformation and by the rise of new styles in music, poetry and dance resulted in a decline and transformation of the carol. The monks and friars who had contributed so much to the religious carol were gone, and although pieces called carols were still occasionally produced by leading composers, their reflection of the increasing influence of continental music has caused them to be distinguished from the older polyphonic carol as carol-motets. Among several examples by Byrd for Christmas or New Year are From virgin's womb and An earthly tree (Songs of Sundrie Natures, 1589), both in several strophes of repeated music for one or two voices and viols with multi-voice refrain after each strophe, and This day Christ was born (Psalmes, Songs and Sonnets, 1611), a through-composed motet for six voices. Herrick's poetic collection Noble Numbers (1648) gives evidence of the continued use of the carol at court. Among its several carols, including two for New Year, is What sweeter music can we bring, identified as ‘a Christmas Caroll, sung to the king in the Presence at White-Hall’ and composed by Henry Lawes. Carols, or references to them, also occur in Tudor and Stuart drama, beginning with ‘Back and syde go bare’ in Gammer Gurton's Needle (1566), and they continued in popular general use, although nothing is known about their oral transmission in this period.

During the 17th century, religious festivities, and therefore the carols forming part of them, came under increasing pressure from Puritan reformers. In 1642, for example, William Slatyer published his Psalmes or Songs of Sion, Turned into the Language and Set to the Tunes of a Strange Land, Intended for Christmas Carols, and Fitted to Divers of the Most Noted and Common but Solemne Tunes, Everywhere in this Land Familiarly Used and Known. It is not clear whether this work was an attempt to entice reluctant psalm singers by taking advantage of the popularity of carols and ballad tunes (the ‘solemne tunes’ are not printed or named, but some titles have been added in ink to the British Library copy, including Jane Shore, Garden Green and Walsingham) or whether it was intended to replace the carols as something to be suppressed, as the Presbyterians had largely done in Scotland. Under the Commonwealth the old religious festivities, especially those of Christmas, were banned or strongly discouraged, but Christmas customs could not be rooted out, especially in the country, and such literary sources as The Vindication of Christmas(1653) describe the continued singing and dancing of carols in rural celebrations. On the Restoration carols began immediately to be published once more (e.g. New Carols, or The Merry Time of Christmas, 1661).

The art carol of aristocratic or courtly circles did not revive after the Restoration, but the popular tradition continued, with carols, like ballads, circulating orally or in broadsheets with carol texts and decorative woodcuts; these were published annually for the Christmas trade, a practice that survived until the 20th century, although its heyday was the 18th and early 19th. Many new ‘carols’ were in fact Christmas hymns, like O come all ye faithful, Watt's Joy to the World or Wesley's Hark the herald angels sing, later set to Mendelssohn's tune.

With Percy's Reliques (1765) and Ritson's Ancient Songs (1790) some antiquarian and scholarly attention began to be given to early carol texts, but intensive interest in the carol as it still existed among the people was not aroused until the early 19th century, at a time when the practice was on the wane. William Hone (Ancient Mysteries Described, Especially the English Miracle Plays, 1823) wrote that ‘Carols begin to be spoken of as not belonging to this century’, but he also listed 89 carol titles still printed annually in broadsheets, not including ‘any of the numerous compositions printed by religious societies under the denomination of carols’. According to Hone, carols were then still being sung in Ireland, but not in Scotland, and in England less than in Wales, where

after the turn of midnight at Christmas eve, service is performed in the churches, followed by the singing of carols to the harp. Whilst the Christmas holidays continue, they are sung in like manner in the houses, and there are carols especially adapted to be sung at the door of the houses by visitors before they enter.

Also in 1823 Gilbert Davies (Some Ancient Christmas Carols) described a similar festive and ecclesiastical use of carols that had existed in the west of England at the end of the 18th century:

The day of Christmas Eve was passed in an ordinary manner; but at seven or eight o'clock in the evening, cakes were drawn hot from the oven; cyder or beer exhilarated the spirits in every house; and the singing of Carols was continued late into the night. On Christmas Day these Carols took the place of Psalms in all the Churches, especially at afternoon service, the whole congregation joining.

Davies's collection of eight carols with tunes drawn from oral tradition, the first of its kind, must have been a success, since a much enlarged second edition appeared the next year. In subsequent years interest in carols grew, principally in collecting and publishing texts and (to a lesser extent) tunes from the living tradition, and in editing early texts (and, later, music) from manuscript sources (Thomas Wright's work in the 1840s for the Percy Society was notable). At the same time, writers continued to describe carols as dying out among the people. William Sandys (in the important preface to his Christmas Carols Ancient and Modern, 1833) wrote that ‘In the Northern counties, and in some of the Midland, carol-singing is still preserved. In the metropolis a solitary itinerant may be occasionally heard in the streets, croaking out God Rest You Merry, Gentlemen, or some other old carol, to an ancient and simple tune’. W.H. Husk (Songs of the Nativity, ?1868; one of the first collections to draw in an important way on the broadsheets) also wrote of carolling as ‘a departing Christmas custom’ and observed that broadsheet publishers ‘find the taste of their customers rather inclined towards hymns, mostly those in use amongst dissenting congregations … The old festive carol seems to have grown into almost total neglect’. Some of the clergy, Husk noted, were trying to revivify the carol, but their endeavours had as yet met with little success, and Husk doubted the efficacy of trying artificially to stimulate a dying folk custom. These efforts partly took the form of expanding the existing body of carols with a variety of new material: setting early carol texts to new or adapted music; transferring and adapting pieces from other repertories related to the carol, like the Noël, Weihnachtslied, Pastoral and Latin or Lutheran hymns (two of the earliest successful examples of this procedure were J.M. Neale's and Thomas Helmore's Carols for Christmastide, 1853, and Carols for Eastertide, 1854, both adapted from Theodoricus Petri's Piae cantiones of 1582); writing new texts for traditional and popular tunes; and composing entirely new carols (notable additions to the ‘traditional’ repertory included Good King Wenceslas and O come, O come Emanuel). Whatever its source, the music was usually moulded to the prevailing taste of Victorian church music: four-part hymn-like textures with a 19th-century harmonic vocabulary. The most successful example of this synthetic approach was Christmas Carols New and Old (1871) by H.R. Bramley and John Stainer. It held the field until replaced in 1928 by The Oxford Book of Carols, edited by Percy Dearmer, Vaughan Williams and Martin Shaw, which revised the carol arrangements in a style in keeping with the taste of an age reacting against Victorian church music. Some later collections, like The Penguin Book of Christmas Carols (1965), edited by Elizabeth Poston, attempted to go beyond The Oxford Book of Carols in reflecting increased public sensitivity to the individual qualities of medieval and folk music.

While Bramley and Stainer hit upon a formula that gave new impetus to carol singing, the revival took place in a form less spontaneous and more institutionalized than before, as the preface to one of the expanded later editions of their collection makes clear: ‘Instead of the itinerant ballad-singer or the little bands of wandering children, the practice of singing Carols in Divine Service, or by a full choir at some fixed meeting is becoming prevalent’. However, the practice of singing carols from door to door is still kept up, and the word is also applied to hymns sung in church at several important times besides Christmas: Advent, Lent, Passion Week and Easter. The new tradition of services consisting of nine lessons and carols for Advent and Christmas, which began in the Victorian period and received its present form at King's College, Cambridge, in 1918, has had a great influence on carol singing in Britain, leading to the revival of medieval carols and to the composition and performance of new ones by leading composers (e.g. Warlock, Holst, Gardner and Rutter).