Versions & Variants of the Tunes of "Barbara Allen"- Seeger
[This excellent article by Charles Seeger and the additional article on the US texts by Ed Cray will take some time to edit and add music examples and charts. Not sure when I'll have time to do this. The original article may be viewed here:
http://www.loc.gov/folklife/LP/BarbaraAllenAFS_L54_sm.pdf
Footnotes moved to the end of the article. The album cover (below) and variant's on the album will be included in the individual versions along with the MP3's. Moses Platt's version is also included in the article: Moses Platt and the Regeneration of Barbara Allen.
R. Matteson 2012]
Versions and Variants of the Tunes of "Barbara Allen" [1]
by Charles Seeger
As sung in traditional singing styles in the United States and recorded by field collectors who have deposited their discs and tapes in the Archive of American Folk Song in the Library of Congress, Washington, D.C.
AFS L 54 Edited by Charles Seeger
PROBABLY IT IS safe to say that most English-speaking people in the United States know at least one ballad- tune or a derivative of one. If it is not "The Two Sisters," it will surely be "When Johnny Comes Marching Home"; or not "The Derby Ram," then the old Broadway hit "Oh Didn't He Ramble." If the title is given or the song sung to them, they will say "Oh yes, I know that tune." And probably that tune, more or less as they know it, is to them, the tune of the song. If they hear it sung differently, as may be the case, they are as likely to protest as to ignore or even not notice the difference.
Afterward, in their recognition or singing of it, they are as likely to incorporate some of the differences as not to do so. If they do, they are as likely to be aware as to be entirely unconscious of having done so. But if they admit the difference yet grant that both singings are of "that" tune, they have taken the first step toward the study of the ballad- tune. They have acknowledged that there are enough resemblances between the two to allow both to be called by the same name. Observe, however, that this nameness refers not to anyone thing but to notion regarding two separate things- A's singing and B's singing. This notion is a third thing, namely, a class of things to which both be long, which is to say that what is sung is distinguishable from the singing of it.
Suppose , then, one hears three, four, or more separate singings of that tune, some times with different titles or words, or both, by as many different singers. The name finally adopted for the lot will be found to cover an increasing number of differences. Probably each singer has sung the tune in his individual way; some, perhaps, slightly differently with each singing. The more separate singings one has heard, the more differences may be noted among both words and tunes. These may be so marked the tone may discover that more than one tune has been used to deliver almost identical texts and that almost identical tunes have been used to deliver the words of entirely different songs- children's songs, political songs , and even hymns.
Folklorists agree that printing has tended to stabilize variation of the words of folk songs, although at the same time it tends to vary them by elimination or addition of detail and by "correction" and "improvement" to suit the tastes of various classes of buyers of printed materials. Thus, new variants become current and in turn be come varied by faulty or capricious memory and by fresh invention. But the printing of tunes has been, by comparison, rare. Few people can read notes.
It is much easier to identify a set of words than a set of notes as "be longing to" a certain ballad. Incidents and plot, names of persons, places, and event scan be collated accurately enough to permit a consensus among students and laymen alike that two separate sets of words are enough alike to be grouped together along with others sufficiently like them as a unit. It is then easy to select a name for the lot. And the name is a word or group of words, often as not selected from the words of the song. We name things in the art of speech. We do not, indeed, cannot distinguish, designate, much less name in our art of music. Consequently, there is, at present, no consensus among students of tunes such as that among students of words of ballads. There is no recognized musical or musicological way of naming a tune except by the title of whatever text or group of texts with which one thinks it is most commonly associated. The question whether any of these associations of words, tunes , and titles ha s any other validity than mere convenience becomes, therefore, a prime consideration of their study.
The work of Bayard[2] and Bronson[3] points to a negative answer to the question. Both, though working separately and independently, have classified hundreds of ballad-tunes, regardless of their associations with particular texts, in enormous "tune- families," of which there would seem to be comparatively few in the total repertory of British-American folk-song.
On the other hand, the aggregate tunes of anyone of the more widely sung ballads seems to give a positive answer to the question. Granted that we accept, as do Bayard and Bronson, the concept of "what is the tune"--that is, a skeleton notation- the resemblances among the majority of the singings of anyone ballad are impressive. See, for examples, "Lady Isabel and the Elf Knight" (Child No.4) , "The Two Sisters " (No. 10) , "Lord Randal" (No. 12), "Edward" (No. 13) , "Young Beichan" (No. 53) , "Lord Thomas and Fair Eleanor " (No. 73) , "Fair Margaret and Sweet William" (No. 74) , "Lord Lovel " (No. 75) , "Little Musgrave and Lady Barnard" (No. 81) , "Bonny Barbara Allan" (No. 84) . Bronson 3 groups the tunes of each of these ten ballads under the letters, A, 13, and C, with occasional recourse to D and E, breaking them down into subgroups Aa , Bb, and so on. The average is two to three versions per ballad. "Sports, " or unique instances of marked difference from any group are found; but the number is, usually, very low.
The tune- family approach, then, leaves us with the impression that tunes and words are utterly promiscuous in their relationships and that marriages of any permanence between them are illusory. The majority-us age approach, on the other hand, leaves us with the impression that although both spouses are frequently unfaithful to their common- law kind of union, yet it has a permanence of a sort not lightly to be regarded.
Of course, small and even large variations among performances of concert and operatic music occur. But anchored, as all usually are, to a single written text, any variation can be referred to this text for purposes of identification. Thus, the consensus among adepts of the written tradition respecting the identity of individual items can be almost absolute, as is shown in such ensemble performance as symphonic and chamber music or in opera. It is the comparative invariance of the usually single printed or written text that one can see and hold in one's hand that defines the identity of the item.
But with the folk song that is sung in English in the United States the situation is opposite. The lack of any printed or written anchor not only encourages but enforces variance of performance. Thus, it is in the relationships of the resemblances and differences in the singings that the identity attributed to what is sung must be sought. In a nut shell, the question is: how much can two singings differ and still be singings of the same tune? Or, conversely, how little can they vary and still be singings of different tunes?
Now, identity is a word- a construct in the art of speech in general, in the English language, in particular. On the one hand, it may refer to an abstract concept on a high, universal level. Its relevance can be sought in many connections other than music. When sought as a tune- family, as by Bayard and Bronson, the identity of the ballad- tune is a generalization dependent upon the resemblances among skeleton notations of separate singings in terms of a nest of descriptive concepts in accord with which hundreds - even thousand -of tunes are ordered, their relationships defined, their history investigated, and so on. On the other hand, the word may refer to the perception of a very concrete sensory phenomenon, for example, Dock Bogg's singing of "Pretty Polly" as recorded on a Brunswick ten- inch record. Almost any musical ear can identify it on the basis of other Pretty Pollies previously heard, or, for lack of such, distinguish it from Harry Belafonte's singing of "Water Boy. " There are enough differences and similarities in the considerable are a between singings to invite extensive study.
Ideally, musicologists should be able to step in here and show that short melodies such as ballad- tune s could be classified by purely musicological methods and designated by purely formal names quite as arbitrary a s "Symphony No. 5" or "Opus 30, No.6. " Elsewhere,[5] I have shown one possible base upon which such a classificatory and naming procedure might operate. It would, I believe , have to pay equal attention to the twin concepts of tune family and majority usage. The former might provide a counterpart to the Arne -Thompson index of motifs in the words;[6] the latter, a counterpart of the Child canon.[7] Since the preliminary investigation of the tune- family has already been competently made and that of majority usage still waits, I shall essay here the latter only, leaving to another occasion outline of a melodic motif index.
To investigate this presentation of the problem of the identity of the ballad tune, it is necessary to select a ballad for which there is available for examination a large number of specimens that are representative of the ma in stream of oral transmission as well in the British Isles as in North America. For as far as this kind of material is concerned, the two areas have formed one singing community for more than three centuries, maintained since 1620 by constant westward migration and now, by the eastward migration of the folk music- revival movement. The ballad of "Barbara Allen," as it is known in the United States, seems to fulfill the conditions set for the above. Folklorists have frequently attested to the fact of its being the best and most widely known of all the "Child" ballads on both sides of the Atlantic.
[Map of the US- missing]
Distribution of the "field collection" of published variants of the ballad "Barbara Allen" in the United States as of December 1962: 243 notations, about one- third transcribed from sound- recordings. (Commercial and revivalist sources are not included in this roster nor are those of British and Canadian field collection amounting to 93 items as of the same date.)
The holdings of the Archive of American Folk Song (AAFS) in the Library of Congress, Washington, D. C. are eminently suited to use for such a purpose. Of the total of seventy-six dubbings made available for the present study, most were recorded at or near the residences of the informants during the years 1933 to 1940 with what must be considered, at this writing, primitive equipment. For the most part, recording was made on aluminum blanks. Collectors were rarely experienced in field collection and few had had special training. Few could make the adjustments and repairs of the machines incident to wear and tear, rough handling, variation of electrical current , and obsolescence. Some of the original discs show blemishes of such basic character that subsequent sound- engineering could not modify them without loss of essential features of the singing. In spite of these hazards, some of the sound-tracks still project a luminous quality through the veil of imperfections. Recordings not only of exceptionally talented but even of ordinary carriers of the tradition of ten possess this quality, though to varying degrees.
It cannot be too strongly emphasized that what is sung and the singing of it are not, musically speaking, two things, but one. Abstraction of the song from its singing is a necessary procedure in talking about music that makes two things out of the original one. This is especially evident when the song is envisaged as printed words and notes. A singing of the song in a singing style other than that of its own tradition is likely to be a distortion comparable to the translation of the words into a foreign language.
Two sets of complete notations of the whole AAFS collection were made by two transcribers, working separately and independently. Of these, thirty were chosen for (1) pressing on a 12" LP disc (AAFS L54) and (2) publication in skeletal notat ions in the present study (Tables I and II). For lack of space and uniformity, no complete notations - that is, of all stanzas of a singing- are included here. Of those selected for pressing, nine (Nos. A1 , 6, 7, 8, 11, and B1 , 13, 17, 19) are given in their entirety as on the original recordings. For the rest, there is space for one or two stanzas only, as illustrative of the pattern or style already presented in the complete singings. Selection was made with reference to musical and musicological criteria alone.
The states represented by place of collection and/or location of the home of the informant (so far as known) are:
Virginia 3
Mississippi 2
North Carolina 4
Florida 2
Kentucky 3 (1 recorded in New York)
Missouri 1
Tennessee 3 (1 recorded in Washington, D. C. )
Arkansas 1
Texas 4
Indiana 2
Michigan 1
Wisconsin 1 (migrant from Kentucky)
California 2 (1 migrant from Oklahoma )
TABLE I (missing)
TABLE II (missing)
Fifteen of the singers were men; fifteen, women. One recording (B1) was of "a family." This is less representative of the Archive's holdings, in which by rough count women outnumber men about four to three .
It is not possible to speak with certainty of the ages of more than about half the total number of singers. For the rest, documentation other than that appearing in the Checklist of Recorded Songs in the English Language in the Archive of American Folk Song to July, 1940[8] is lacking. We can be reasonably sure that most, if not all, of the first twenty strips on the disc were recorded from the singing of informants of middle or advanced age and represent variants of what seems to be the older, more archaic singing style, without instrumental accompaniment or noticeably influenced by such accompaniment. The last ten strips were, for the most part, recorded from the singing of younger men and women, some with guitar accompaniment. No banjo- accompanied performance of the ballad was available for dubbing. There are entries for two singings with dulcimer in the Checklist. It is not known how widely ballads have been sung with fiddle as on s trips A4 and B6. The last two items (B18 and 19) are not, strictly speaking, typical of traditional country singing. Miss Tarwater learned the ballad from an elderly relative who carried the traditional singing style; but being an educated young woman, she reproduced it to a large extent by deliberate cultivation of its excellencies. The result is probably the nearest to what the average urban concert-goer would consider "good singing."
Study of these materials was made in the frame of a roster of more than 200 additional variants of the ballad assembled from printed sources, recordings and microfilms of hand notations, both British and American, publicly available to students in the United States. No attempt was made to cover private collect ions. For convenience, a set of transcriptions of all recordings was made in skeletal form to match the notations in printed sources. The whole 300-odd were then transposed so that their stanza finals fell upon Q. above middle Q. and the phrases we relined to match the lines of the words. The fact that the recorded tunes revealed far more notes and more tonal and rhythmic subtleties than were found in published skeletal transcriptions will not be considered here in detail.
The material from the Southern United States shows, on the whole, more homogeneity than does that from Great Britain or New England, where, among a smaller number of variants, individual variation seems to have been more extensive. The number of defective items from any area is negligible. Some resulted from two successive singings of one half of the tune without rendition of the other. Momentary failure of memory some times resulted in isolated successions such as these. A few singers sang the words of the ballad to well-known tunes such as "Won't Be Home Till Morning," "Auld Lang Syne," and "Polly Wally Doodle," or to a tune more commonly associated with another ballad.
The task of putting such a large number of diverse materials into such order that their study can be pursued most conveniently and economically is primarily one of classification. Its first requirement is a set of criteria in terms of which the materials may be described in the simplest but most precise and comprehensive manner possible; its second, a uniform taxonomy or series of categories or boxes, into which each individual variant may be grouped with others sufficiently like it in a class of variants in a box to be placed, along with other boxes containing classes of variants less resembling it, in a box of versions, boxes of versions in a larger box of classes of versions , eventually, in the boxes of families, families in the context of the repertory of the tradition as a whole, the tradition in its musicological, social, and cultural context , and so on. If this is done shrewdly and with common sense, comparative studies may be engaged in conveniently at any point a long the way, not only within the confines of one particular idiom, but cross - culturally, in accord with the aims of modern ethnomusicology.
The great majority of American ballads in the English language are cast in a single basic or gross form. It is strophic (stanzaic), in contrast to the older verse or line form still found, among other places, in the epics of southeastern Europe and among the Spanish romances (ballads). The words of many, including those of "Barbara Allen," usually written as four lines, the second and fourth rhyming, are twinned in the singing with four melodic phrases, the ends of the second and fourth of t en, but by no means always , comprising melodic patterns that might be called "musical rhymes," that is, identical or closely resembling tonal or rhythmic devices or both. Sometimes the last couplet or just the last line is repeated. This is less common in southeastern than in northeastern United States and in Nova Scotia, England, Scotland, and Ireland. The words of the ballad are invariably arranged in iambic (v, short long; or vi, weak strong) poetic feet. The number of syllables in the lines conforms to the general pattern 8-7-8-7. As in other forms of English poetry, additional syllables may be crowded into the two-syllable foot. Two or more shorts of ten fill the place of one long; several shorts may occur in the place of one short. The accented syllable may not be long in duration but short, some times very short. Long syllables are not always accented. The normal accent pattern is 4-3-4-3. In reading the stanza, there is usually a pause after the end of the second line and a longer one after the fourth, separating the stanza from its successor. This combination of rhyme and short lines plus pauses tends to give the feeling of a basic couplet 8-8.
Normally, each syllable in the singing is accommodated by a note. We may refer to the union of the two arts upon this level as the "sung syllable." [9] The sung syllable must be regarded as an elastic concept. It simplications now be come our immediate concern.
The actual sound of the sung syllable can be described most economically and conveniently in terms of the variance and non-variance of the six essential functions or resources of the singing voice. These are essential because without anyone of them there can be no tune, while with them all, there can be, viz.:
tonal functions: loudness, pitch
tone-quality
rhythmic functions: tempo proportion
accent
Pitch, loudness, tempo, and proportion are simple functions; they may vary in only two ways or not vary. By varying (variance) is meant that a melody may rise or fall in pitch, may be come louder or softer , its tempo (speed of beat) may be come faster or slower, and the beats may be grouped in patterns , regular or irregular, of long, short, and so on. By not varying (non-variance) is meant that the function, once it is set at the beginning, is maintained fairly steadily. The set and the variance of the simple functions can be measured in cycles per second, decibels, seconds, and relative durations in terms of some given unit.
Tone-quality (timbre, or quality, for short) and accent (quality, in prosodic theory) are both compounds of two simple functions. The sound that we call a "note" is actually a complex of many simultaneously sounding pitches -most of them higher than the pitch we hear as the note--of varying degrees of loudness, the loudest among them being perceived by us as the note. Keen and practiced ears can hear some, but not all of them. The total effect is referred to as tone-quality or timbre. Viewed in these terms, this function is one of each single note or fundamental pitch that we perceive in a song. There is a slightly different tone -quality to each of the notes, and the singer has only limited control of the variance of the function as far as separate notes are concerned. The singer trained in the bell canto of Italian opera or in the tradition of the German Lied tries to produce a desired quality in the stream of sound rather than in each separate note (barring, of course, exceptional not es to which they wish to give special effect). The British-American folk singer, on the other hand, seems to give no special thought to quality of sound, but sings in as "natural" a voice as that in which he talks. Consequently, the singing voice varies greatly according to sex, age , and an infinite number of psychological and physiological factors. The lack of any preconception of what it ought to be gives, however, to the quality of the traditional singer's singing a clearly recognizable character that can be instantly recognized by other carriers and by connoisseurs of the tradition.
The case of accentuation in this tradition of ballad-singing is similar, except that what matters is not the number of simultaneous pitches and loudnesses that are heard at once but the number of consecutive pitches and loudnesses that are heard in the course of a tune. It has been again and again demonstrated that a succession of exactly even or equally strong, loud, stressed, emphasized (or whatever) beats can neither be delivered nor felt. Thus, just as we hear a predominant fundamental or note, in the complex of simultaneous sounds in a tune, so it is that we hear an accent or, more precisely, an accent pattern. This pattern depends to varying extents upon the pitch and loudness of the "notes" but even more upon the rate of tempo beat and the relative durations of the pitches.
No scales of measurement of the compound functions have as yet been proposed. Their set and variance can be estimated upon the basis of the simple functions of which they are compounded, but their variance in kind is infinite. In every tradition of song, each of the six functions or resources is exploited in a manner characteristic of that tradition. The carriers of the tradition, as well as knowledgeable students, can perceive immediately whether a song is or is not sung in the traditional singing style. In the music of the Occident, the tonal functions are organized quite independently of each other. That of pitch is the most highly organized. Continual variance is the rule. But singers and instrumentalists are conscious that they employ a limited number of principal pitches in a melody such as a ballad tune and that there is a hierarchy among these in which one serves as the "key note," known to students as the "final, II "tonic," or "tonal center," around which the others move in diverse relationships that are fixed in the tradition and with respect to which their relationships among themselves are defined. Thus, the continual variance upon a lower level is ordered with respect to a non-variance upon a higher. A singer "keeps the pitch" or "sings in the key. "
Loudness is comparatively unorganized. It is customarily regarded as a characteristic of the way a whole piece or substantial part of it is performed. The sounding of a certain individual note slightly louder than its neighbors is usually referred to as accent: Even by the elite, the professional or fine art musician, loudness is not articulated beyond the somewhat vague steps of pp, p, mp, mf, f, ff and so on. None of these has any significance in traditional ballad singing. Thus, while continual variance in loudness is the rule in twentieth century concert music (and in the singing of folk songs by professional or professionally influenced performers), in the folk art the tendency is to invariance.
Although there is no musical measurement of tone-quality and no notation for it, certain standards and rough classifications are in use in the fine and popular idioms, where, in addition, wide variance-some of it controlled, some random-is the rule. In folk song, there seems to be no particular attention paid to tone-quality. To the extent that hymn-singing, concert, and the more elaborate forms of popular music have influenced folk singers, there is often an intent to produce a "pleasing, " "refined," or "good" tone-quality. But for the most part, the singer sings with his "natural" voice and does not give special attention to enunciation, voice placement, projection, and the other requirements of platform exhibitionism.
While the tonal functions operate more or less independently of each other, the rhythmic functions, although susceptible to analysis as separate items, are so mutually interdependent that the tendency in all idioms is to regard them simply as one thing-"the rhythm." As in the tonal, so in the rhythmic functions there is a hierarchy. Both singers and instrumentalists tend to present them in "one package," to which all three contribute. The role of axis or center lies, then, partly in the tempo beat, partly in the proportional pattern, and partly in the accent. Change in any of them tends to change "the rhythm."
The tempo beat, often referred to as "pulse, " tends to center in the range of the heart-beat, step, foot, or hand movement, roughly 40-200 per minute. In British-American folk song, the range may perhaps be set more narrowly60-180. In this tradition, proportion is usually in two's and three's; that is, two or three tempo beats are likely to be grouped together by an accent either of the words, the tune, or both. When the tempo beat dominates, the rate of movement tends to be in the 120-180 range; when proportion dominates, half those value~0-100. Domination is by accentuation. We have, thus, three kinds or levels of accent: (1) of tempo beat, where each sung syllable is rendered in approximately equal strength; (2) proportional or metrical, where the first beats of groups of two's and three's is stronger; (3) "agogic" or rhythmic. Anyone may dominate the others; but in theory, at least, the first gives way to the second and the second to the third. Accent, then, is not only a rather complicated function in a strictly music-technical sense, but is also largely a matter of individual psychological disposition, moulded by the socio-cultural context in which the singer is brought up and lives. Loudness of attack is only one way in which accent can be produced and is not by any means necessary. Tempo accent, or pulse, can be felt on a rest. Long notes, especially if integral parts of a proportional pattern, tend to receive, or to give, accentuation. Anticipation and delay of attack and a change of tone-quality can produce accent. Ornamentation of a pitch level by a slide, by over-or under-shooting of a skip, or by quick little accessory notes, and especially the prosodic accent, felt through the musical fabric, can produce it.
The potentials of the singing voice are about equal with respect to three different attitudes toward the song and the singing of it in the socio-cultural context in which the tradition is current. They can best be understood in reference to the management of the six melodic functions or resources through -out the procession of sound from the beginning to the end. They are:
(1) a firm initial broaching of the essential stuff of the tune, tapering off toward the end-athetic, trochaic, strong-weak, tense-detense kind of delivery. It is typical of much Amerindian singing.
(2) a gradual picking up of the tune, as if directing it toward a preconceived end-an anacrustic, iambic, weak-strong, detense-tense kind of delivery. It is typical of European-style concert music.
(3) establishment of an initial level and maintenance of it throughoutspondaic, finn between the alternatives of weak and strong, balanced between the alternatives of tension and detension, typical of Plainsong. Among the most admired carriers of the tradition of British-American ballad singing, the configuration of the basic melodic functions is effected as in (3) above, that is, the level of each is set at the beginning and none is varied intentionally, excepting only that of pitch (but pitch within the higher level of mode and key conforms to the rule.) The degree of loudness and the characteristic tone-quality with which the singing begins is maintained throughout except when incidentally or involuntarily compelled by high or low notes; and the less variance, the better. The rhythmic level is set at the beginning by a steady beat and a clear proportional pattern that are adhered to, within the freedom characteristic of the tradition, throughout the rendition. Some singers adhere to a uniform proportional and accentual pattern; others, to an irregular one. The loudness and tone-quality remain steady, then, except unde.r compulsion of physiological factors in the vocal mechanism, and are never varied on accOlmt of any word or detail of content or to produce a special effect upon listeners.
It is difficult for urban Occidentals to recognize the fact that although variance may be the spice of life, invariance may be the meat. Tough meat, sometimes, to be sure. But the spice is not worth much without the meat. Rather, it is through the teaming together of the variances of some and nonvariances of others of the six basic melodic functions that the diverse balances characteristic of each tradition are achieved. Indeed, the prediction might be ventured that if only its characteristic balances be maintained,9 the tradition of British-American folk song might, in spite of more pessimistic views, successfully survive not only the assaults upon it by social and cultural change, technology, and commercialism, but also the defenses of it by the mass folk music revival movement.
The attitude, then, typical of the most admired traditional singer toward the song tends to the serene and detached, however much force or gentleness may impel the line from its beginning to its end. Singing seems to be a natural thing for one to occupy himself with if he wishes. It requires of him no special preparation, effort, or pretense of an organized sort. Difficulty or ease in execution seem not to be factors. It is not a vehicle for pathos, but seems to meet accepted requirements of an ethos. In spite of the often romantic words, an almost classic reserve is maintained. There is no assumption by the singer of a position above or apart from the tune' or text, but, rather, an implicit recognition that "life is that way." He keeps a straight face and unchanging posture, without attempting to win an audience by smiles or gestures, Yet there is nothing "dead pan" about the singing. There is usually a difference between singing to oneself and to another person or persons. Comparatively few singers, I understand, are unaffected by the presence of a microphone.
[missing para]
[chart]
In the use of conventional notation to illustrate these resemblances and differences, we must remember that this notation was developed for the purposes of composition and performance in the tradition of the professional, "high, "elegant, elite, or fine art of Occidental music, not for performance of the folk art. There are, of course, enough resemblances between the two idioms to render the notation useful in the study of folk song; but it is deceptive because it can represent only these, not the differences between the idioms. And there are substantial differences. Furthermore, singers vary with respect to the degrees of contact they have had with the idioms of popular and fine art; and this shows in their employment of each of the six melodic functions.
Rhythmic modality, though long abandoned in the theory and practice of the fine and popular arts, is still significant in the analysis of folk song. This is not the place to resume the transformation of the tight mediaeval theory of six rhythmic modes based upon the theory of Greek prosody, in which relative length in "perfect" (triple) meter was the sole criterion. Suffice it to say that in the more highly diversified and experimental idioms of both poetry and music of the Western World, duration (quantity) and accent (quality) function together, with tempo beat, in such complicated interweaving that no one has yet been able to set up any general theory comprehending the rhythmic practice of either art-much less, their association in song-that has been able to achieve a wide consensus of experts.to Fortunately, the singing of the American ballad in the English language is simple enough to warrant classification of the rhythmic norms in terms of both proportion and accent with some plausibility, granted, of course, that the steady tempo beat can be determined.
In terms of proportion, variant classes la, Ib and Ic can be seen to be distinguished, among other ways, by even duration; classes ITa, lIb and IIc, by uneven, but in a patterned unevenness. (The random, unpatterned, long notes in strips A1, All and B4, for examples, are not considered to alter the basic even pattern.) Class Ix shows the tonal configuration of Version I with the uneven proportion of Version II; class IIx the tonal configuration of Version II with the even proportion of Version I. Curiously enough, neither of these two common music-proportional patterns approximates the spoken patterns. A reading of a stanza in the manner of either (a) or (b) in Example 1 would be in poor taste. Pattern (c), a variant of Version I, is rarely found in the United States, though more often in England, and resembles much more closely the rhythm of speech. And the latter wbuld never show strict regularity, but would vary the pattern with doublets, both even (J J ) and uneven ( ~. ;, and ;, ~. ). A competent reader of the words alone would never conceive these divisions of the foot as countable in terms of an arithmetic unit as musicians count the beats and divisions of beats in measures. Some singers approximate the spoken patterns of delivery, but sporadically. Observe, especially, the more detailed notation of the first stanza of Mrs. McCord's singing (A6) where the ornamentation of the sung syllables appears as a doublet, triplet or quadruplet, but rarely with much more than near accuracy. (See Table I).
The singer may take a breath after each phrase, or if the tempo is lively or the lungs capacious, only after the second and fourth phrases, as in variant B10 (Ford), in which case the quatrain verges still more closely upon the couplet. Incidentally, the word "Barbara" is invariably shortened to "Barbra," "Barbry, " "Bobby," or some such.
The three most common musical proportions associated with the poetic meter of our ballad are shown in Example 1. The two former, (a) and (b)
Example 1
[missing]
were not among the six accepted by mediaeval rhythmic theory. The last, (c), conforms to rhythmic mode II. Neither of the former has been recognized as forming a poetic meter by classic prosodic theory.ll Both the even (pyrrhic,vv, or spondee, --, depending upon the speed of delivery when sung) and the uneven (fourth paeon,vvv -) were, however, recognized as prosodic feet that could be introduced occasionally in any accepted poetic metrical sequence as a variation of the over-all poetic rhythm. in the singing of this and other ballads of similar gross form, the undeniably iambic rhythm of the words is, then, customarily sung, in the great majority of cases, to two very different musical meters, the prosodic counterparts of which are considered illegitimate for poetic use. The result is a rhythmic counterpoint between text and tune. The musical rhythm dominates most of the singings. But the poetical is always evident and serves a basic function in the inner organization of tonal and formal resources in the conventional theory, notation, and practice of the elite and popular idioms of European music for the last three centuries or so, musical meter has not been reckoned in terms of long and short (quantity) but as the grouping of even tempo beats in measures, that is, between bars. Among these beats a hierarchy of accents, strong and weak, had inevitably to be found, simply because a chain of equally accented or equally unaccented notes could be sustained only for a short time and with special effort. Random or chance accentuation was considered bad taste, unworthy of the professionally trained musician. But in spite of the dominance of relative accentuation (strong, weak) over relative duration (long, short), the latter has remained an important factor both in English poetic practice and in European music in general, and often dominates the former. Musical meter, then, as primarily a mere measurement of the number of tempo beats in a measure, came to imply an accentual hierarchy within it, strictly bound to the counts in the measure, the first being given precedence over the others regardless whether there was a sound to deliver it, subordinate beats, if any, being given secondary or less emphasis according to the theoretical or affective disposition of the performer.
This conception of musical meter seems not to be a necessary feature of the most admired ballad singing. It often intrudes and is evident, as in variant B3 (Harmon) and others on Table I. Hymn-singing encourages and instrumental accompaniment compels it. But judgment of its presence or absencelike that of "in tune-ness"-is largely subjective and may be "read into" the notation of a singing where it does not belong, as for example, wherever the tempo beat is very strongly marked. Variant B13 (Beeker) is a case in point. Each syllable is "punched out" with such nearly equal emphasis that some students might claim it should be notated with a metrical signature of 1. This can easily be disproved by counting it in a triple meter. If in anyas surely it seems to be-it must be in a duple meter; but whether a simple a or compound 1still remains a question. Barring, then, likely to be a problem in the notation of any music idiom, poses an especially serious one in that of British-American folk song because conventions such as rule in the fine and popular arts are, more often than not, lacking. Many professional musicians and, probably, most amateurs seem to understand that the first beat of every measure should receive a strong accent, which is to say: it should be sung louder than other beats. In some written music, this is intended by the writer. But competent musicians and musicologists seem to be generally agreed that in any correctly written notation of measul:ed music the metrical accent, like the tempo pulse (of which it is an elaboration), though always felt, must be surpassed in strength by phrase (agogic) accent. There are no grounds of adducing such considerations in the notation (least of all, in the skeletal notation) of British-American folk song. Prepared for, and for the most part read by, people of modest music literacy, no such hierarchy of accentual complexity as that briefly outlined above can be expected to be heeded, even if a resort to it could be justified. This is the more nearly true because it has become increasingly conventional to bar music by melodic pattern rather than by either relative duration of relative accentuation, both of which, together with the pitch function, are factors in the formation of conceptions of melodic pattern. These conceptions, the formation of which is sometimes more subjective than objective, are less atomic, more extended in nature and have led to choices of compound metric· t h 45669123 d Th h"
caI Signatures such as 4' 4: 41 8) 8l 8' 2' an so on. us, sue slnglngs as A2 and A4 (Grallam, Bryant, Gevedon) could be written in accord with conventional usage in several different ways, three of which are shown in Example 2. But if, as seems likely, most readers are going to place strong accents on first beats, distortion of the singing style and, consequently, of the song will result.
Example 2
Version II and Version I when in the uneven proportion of Version II (class Ix )--permit several different fittings of the notes given to sung syllables within the various metrical signatures chosen by transcribers. Most often, these are ~ or~. Sharp sometimes used the one; sometimes the other. It is not clear why he did so. Howard Brockway, 12 a trained professional musician, used a (see Example 3), which might have been the most accurate (as it might also be for Example All). But many people have difficulty reading an apparently triple-metered tune in a duple meter, especially if many ties
Example 3
cross bar lines. Transcribers (Sharp, among them) have sometimes begun their notation as at (a) in Example 4 and sometimes as at (b). Perhaps
Example 4
Musical notation.
Musical notation. Musical notation. Musical notation. Musical notation. Musical notation.Musical notation. Musical notation. Musical notation.Musical notation.
though one cannot be sure, an accentual difference is indicated. In the first of these cases, the triple meter can usually be maintained without change. But in the second, it is usually necessary to provide a duple signature for each of two measures, one in the second, the other in the fourth phrase, as,
for example:
Example 5
Musical notation.
It is on account of difficulties such as these that ethnomusicologists--and many editors of pre-1600 European composed music-often dispense entirely with the bar and the modern metrical signature. Some students compromise by dotting some or all bar lines; others may use the Mensul'stl'ich, or a portion only of a bar. At the bottom of this unsettled state of affairs is a real confusion of the pulse, step or beat of tempo, and the accent of metered measures, with the accent required by melodic pattern, phrasing, and by unusual features such as high notes, long notes, dissonant notes, notes rendering certain words, and many other such exceptional and often irregular items. But the main confusion is, as said above, between the sound and the procession of the sound that is marked by the tempo beat, which, once the sound has established it, is understood to "be there" whether or not there is a sound to mark it, as, for example, upon a rest. One can test this by holding an electric metronome in one hand and setting the dial, with the other, to the initial tempo beat. After running through a stanza of even such flexible renderings as those of Mr. Marlor (Ai) and Mrs. Jackson (All), one should be able to keep the instrument in time with the singer, even through the rests and in the occasional rubato passages. (This is most easily done at half-speed. )
A sung syllable may be divided-stuffed, as it were-with interpolated syllables. Thus, the usual "For love of" often becomes "For the love of." The pitch rendering the interpolated "the" may be the same as that of the preceding or succeeding syllable, a passing note (if a skip is being made), or other. Its duration may be halved or quartered or run in uneven division as in A6.
A sung syllable may be prolonged randomly without alterations of the basic modal pattern and is almost invariably effected in strict accord with the tempo beat. Often, the prolonged syllable is not a prosodically long, strong or stressed one, as in the first stanza of variant Ai (Marlor) where the last syllables of "remember" and "William" (in the repetition of the last line) are so treated. The rests (pauses) after the second and fourth lines of the quatrain are also often prolonged. Naturally, both of these practices result in distortion of the poetic meter and complicate any preconceptions of musical meter "read into" the singing of the ballad by concert-music bias. Thus, three beats are often given where the academic mind expects two, or four, where it expects three.
On the other hand, several sung syllables may be diminished so as to pack them into a single~ven a weak-beat. This is often done in the, anacrusis (up-beat) of lines, as in the first stanza of B17 (Carr), where in place of the normal
v/v/v Ivl It was all in the month of May
the singing is
vu~uu /uuu U v / 'Twas a-bout in the mId'dle of the month of May
The stresses that should have fallen upon the second syllable of "about" and upon "month" are all but lost. Similarly, the pauses after the second and
fourth lines and the long notes characteristic of Version II may be shortened, often running counter to preconceived notions of meter by taking two beats where three are expected, or three, where four are expected.
The accenting of prosodically short syllables is as common in the sung as in the spoken language. But the prolongation of them, impossible in many cases without phonetic and phonemic change in the latter, is common in the former. See again, in variant B17, the singing of the first syllable of "middle"
as t1 mere-" or "mear-. 11
In variant All (Jackson), the singer employs all these practices, with a resulting distortion of the poetic meter that is extreme even in her branch of the tradition. With a fairly normal text-syllable layout of 8-8-9-8, the stresspattern of her first stanza comes close to 2-2-4-2, instead of the normal 43-4-3, as
In some of the best singing (i. e., in what is recognized by carriers of the tradition and by connoisseurs as most representative of the main stream of the tradition) the sung syllable is not always articulated as a proper note in as definitive a manner as is traditional in the fine and popular arts. Attacks are often-in some instances, usually-made by an ascending slide, even in a descending passage. Descending slides are not so common. Detection of articulated (passing) notes in slides is often possible. Even skeletal transcriptions must show some of them. It must be borne in mind, however, that such notations serve limited ends: (1) for quick scansion; (2) for classificatory and lexicographical convenience; (3) for typological study. For in these, only the generalities of the singing appear. As Fox-Strangways put it, only the "substantive notes" are shown. To the extent to which we wish to probe more deeply into the particular musical event and to represent more thoroughly the nature of the music-communicatory substance, more notes and symbols must be added and pitches and proportional values more accurately measured. The more detailed such a notation is made, the less readily legible, even to the expert, it becomes. Recourse may be had to modern electronic music writing devices. Sometimes these support subjective judgments, but sometimes they run surprisingly counter to them. Such is the case with the first strong accent in the line quoted above from Mrs. Jackson's singing (All). The fourth syllable, "in," of the initial line "It was all in the month of May," may be seen to be sung more softly than either the preceding or succeeding sung syllable. (A full graph of this extraordinarily free and admirably traditional singing is shown in Table IV. It was made by the Model B "melograph" in the Institute of Ethnomusicology of the University of California at Los Angeles. The upper tracing shows the pitch line of the melody, the lower, the loudness. The dots on the bottom line of each frame show the lapse of time in seconds to be 25 mm. on the chart. The heavy vertical lines indicate possible metrical divisions enabling note-readers to keep their place, but not with any implication that the singer necessarily made such divisions. To permit least reduction in print, the rests at the ends of phrases have been cut. These are indicated in the transcription in Table 1. )
If we are to believe that the tempos of the recordings are not very far from the tempos taken by the informants, they vary by nearly 350 per cent. Mrs. Griffin (B9), an elderly singer with a large repertory, but obviously very nervous, for she slowed down considerably in the course of her singing, ran through the first stanza in 11 seconds; Mrs. McCord (A6), in 38. Mr. Carr ( B17), on the other hand, increased the speed of his delivery from nineteen to fourteen seconds for the stanza.
Perception of the speed of the tempo beat of these singings should not be difficult for one even slightly familiar with the tradition; and he should be able to mark it with foot or hand in a musically satisfactory manner corresponding to the flow of sung syllables of the ballad. For the changes of pitch and the duration of notes and rests conform, with rare exceptions, to a steady, undifferentiated, recurrent pulse or beat. It is not a clock beat. There is, as in most good concert singing, an ever-present rubato. There is therefore a continual interplay between the tempo beat and the flow of sung syllables. Sometimes, the syllables seem to be compelled by the beat and sometimes the beat seems to yield to their exigencies.
There are three principal ways of expressing the speed of beat in written form. First, and least accurate, is the commercial metronome: the penaulum type (Maelzel) and the electric. The former operates only upon a set scale. The latter is more versatile and can be set at any nwnber of beats per minute from forty to two hundred or so. But its calibration is usually not dependable and the amount of trial and error required to arrive at a result satisfactory at more than one session is sometimes frustrating. The second method is to time the stanza with a stop watch and divide it by the nwnber of beats-in this case, 69, as indicated in the notation in TableIVby quarter-notes and quarter-rests. By dividing the nwnber of seconds in a minute by this nroduct one obtains a more accurate metronome figure. The third method consists of recourse to a laboratory machine, or to a prototype of one such as the Model B melograph mentioned above. On a complete chart of the four frames shown in Table IV, the length of the tune line is 668 mm., or 26.72 seconds marked in 25ths of a second. Division of the former by the nwnber of beats gives a length of 9.681 mm. to a beat; of the latter, .3872 seconds. These, divided into 1500 and 60, respectively yield 154.54 and 154.95 for the metronomic speed of beat, which, for practical purposes can be regarded as 155, and accepted as supporting or correcting resuIts obtained by the first two methods of expressing musical time in terms of general time. The musical validity of this figure can be tested in two ways: (1) it may be set as a medial point on the electric metronome and raised or lowered to follow the rubato quite closely and to form some notion of the extent above and below this point; or (2) it may be held unchanged throughout and the flexibility of the melodic line noted. In a thorough study, both should be used for each stanza and for the whole song.
The variance in number of beats for first stanzas of strips on the record is 32-45 for Version I and class IIx; 44-70, for Version II and class Ibx. Variants All and B4 are highest, with 69 and 68, respectively. The beats are unevenly distributed among the phrases. In only half the thirty are first and third phrases longer than second and fourth, reflecting, thus, the prosodic lay-out. Not one of the thirty could be given a metrical signature other than ~ without suppression or addition of rests, if not during the singing of stanzas at least between them. Only two pairs of variants agree in the number of beats for the four phrases shown in the notations in Tables I-II: A7 and A9 (12-11-12-10) and B12 and B13 (8-8-8-7). Besides these four, there are seven others with only two different phrase lengths, while there are fifteen with three, and six in which all four are of different lengths, among these being All, with 18-17-19-15 (=69). In Table IV, I have marked bar lines for possible division into metrically countable measures; and it will be observed that no two successive measures carry the same metrical signature. The beat of nearly J=155, or 9.7 mm., appears as a row of small dots under the words. It may be observed that Mrs. Jackson had a tendency to anticipate the beginnings of phrases and important words.
The absence of eighth syllables at the ends of the second and fourth lines is
dealt with in two different ways. On the one hand, the spoken contractions of
"dwelling" and "Allen" into one crowded spoken "long" are disregarded, as
at (a) in Example I, and are spread over two or more beats, each of the two
constituting separate sung syllables: long-long or long-short. This is typical
of our Version I, in which the second note of each pair tends to be on the same
pitch as the first. On the other hand, the spoken contraction is accepted, as
at (b) in Example I, and is given what is sometimes called a "Scotch snap"
on two different pitches, short-short or short-long. This is typical of Version
II.
Classification of the versions and variants of tunes for "Barbara Allen" by
rhythmic CQnfiguration is, then, in terms of proportional mQdality. But it should not be~ forgotten:' that these. terms 'imply measurement in y;e"rmsof thetempobeatwhich, inthistradition,-as in"many"others, is'a steady, even, and undifferentiated pulse, hand, or foot beat, not grouped in uniform metrical patterns, but manometric. Each beat is heavily and uniformly accented by some singers; by others, scarcely at all. But it is always
present. As in good concert singing, slight rubato is the rule, medial
often met with and maximal, though common with some singers, not common
with most. The axis or centricity of the mode, therefore, is on the lowest
level the tempo beat, with whatever degree of accentuation the singer custom arily gives it; on a higher level, it is its own recurring pattern. And thus, all the tonal-rhythmic units we call "notes" (and all rests) refer to this beat, delivering it, dividing it, spanning two or more, arriving a little late or before it, as to a rhythmic "tonic" that in any competent singing of a melody is alternately affirmed, moved away from and back toward as a center.
Classification by tonal configuration must be entirely in terms of pitch, regardless of loudness and tone quality. The subtle interplay among the three rhythmic functions has a counterpart, however, in an interplay among the pitch factors themselves and upon a far more highly elaborated series of four levels: the smallest melodic unit (usually only three or four notes), the phrase, the double phrase, the single tune setting a quatrain. Development beyond this point, by refrains, repetitions of lines or fragments of them, and alternate stanza combinations in which the odd stanzas are sung to one tune
and the even to a variant or quite different are so rare in the singing of this ballad in the United States as to need only an occasional passing notice here. Principal diagnostic criteria are: (I) intonation, including vibrato; (II) total contour; (III) phrase contour, (IV) phrase levels; (V) tonal axis or center; (VI) tonal mode; (VII) prosodically and/or music-metrically accented notes; (VIII) phrase patterns; (IX) phrase endings; (X) "trademark" formulae, such as the above mentioned Scotch snap; (XI) cadential or beginning formulae. Most of these require too much technical analysis for use here. Suffice it to say:
tune,
l--Intonation, in the tradition of British-American folk song, differs in some respects from that of the concert stage. No precise or comprehensive tonometrical measurements have, to my knowledge, been made. Such as are at hand, although they encourage us to expect that this criterion could be of firstrate classificatory value, do not warrant use to that end at present. The same is true of vibrato studies. In my experience and in that of such students as I have been able to question, singing in or out of tune is not, ordinarily, mentioned among carriers of the traditional singing style. (One may observe, however, that in the four frames of Table IV, Mrs. Jackson maintains her pitch very closely in all long notes except ~ This is set at the very beginning and returned to at the end, but is raised more than a quarter-tone at the end of Phrase I and lowered by as much at the end of Phrase III. Whether there is artistic significance in this I cannot say. ).
A similar unevenness is found in the vibrato. The word "May," for example, shows clear twinning of frequency (pitch) and amplitude (loudness) vibratos of about 6-1/2 per second-eonsidered a good rate in 19th-century "good" singing. But there is a heavy secondary frequency vibrato of about 45 per second-outlawed by every canon of concert usage simply as bad tone-quality, but typical of at least one strand of the American tradition of ballad-singing. The variance of pitch of the vibrato is slight, however, and not more than a quarter-tone above or below the perceived fundamental-narrow by concert standards of the past century. But the words "was, " "died" and "with," all accented, long and high for this singer, are sung with D.O apparent primary vibrato and a very narrow secondary. On thE; words "heart, " "For the love of" and "Barbry Allen, " there is no apparent primary pitch vibrato and almost no loudness vibrato, but heavy secondary vibrato.
If we are to believe that the frequencies of the recordings are not far from those set by the informants, tessituras or average "lie" of the voices (as the Harvard Dictionary of Music puts it) on disc L54 are for the women low. Finals are usually between .E.-flat above and A-flat below middle~ with the highest notes an octave or less above and the lowest a tone or a fourth below. For the men, finals range somewhat more widely, but with a medial clustering a fourth or a fifth below middle.Q,. The musical ear that is more highly trained to pitch than to rhythm can often place the intonation very accurately; but many people, though no less musically talented or trained, may have to determine it by recourse to pitch-pipe, tuning fork, or instrument with fixed tuning. In the notating of pitches below or above the degrees of the equal-tempered, twelve-tone, octave scale, based upon A=440 c. p. s., a small arrow pointing up or down is coming to be used in preference to plus and minus signs, or other symbols. By placing the record on a dependable variable speed turntable and choosing a rate of revolution that will set the final on some convenient, predetermined pitch, one can check by ear all variances from t,h.e standard scale. The small, stem-less note-heads to the left of clefs in Tables I and IT represent the notes of the standard scale nearest the actual number of cycles per second as measured ~y a strobocoIID--ii device Widely used for tuning student band and orchestral instruments.
IT, ill-Version I shows a total over-all pitch contour that emphases the couplet qualities of the ballad stanza, viz., a high first half and a high to low second. The individual phrases tend to be cast in bow or arc form; an initial ascent is followed by a descent to, or below, the starting point. Version IT shows varieties of this form not only in the individual phrases but in the overall contour as welL
;lV'Typical of Version' ils a pattern of phras.al pitch levels: ~:~ high-high to iow-Iow.· .Typicai of Ve'rsion IT' is the pattern: medlal-high-high to lowlow. One peculiar variety is found in the third phrases of both versions: the descent of the arc to the bottom of the ambit or range of the tune. Poladian,t3 follOWing Herzog, describes this as "an ascent by the interval of a second, then a descent with a bend in the opposite direction in the middle of the phrase of many." It may be worth mentioning that in Sparp's Appalachian collection\(
13Sirvart Poladian, A Study in variants. published by the University of Cali f ornia. Department of Music, as a Report on Unit A-25 of Official Project No. 65-1-08-62, conducted under the auspices of the Works Projects Administration; Edward B. Law ton. Jr., Faculty Adviser; Sidney H. Robertson, Supervisor (Berkeley, January, 1940). Typescript, viii + 91 + 117 positive photostats of music notations, 26 pages. P.46.
the patternHH~LLis found often up to No. 45, whi1lHends the Child ballad section, but rarely after that; whereas the pattern M l.LL is common through both volumes. Classification by pitch levels of phrases, is, however, less significant in Version I than in Version II, where it provides the main criterion for distinction between variant classes IIa (variants A7 -All) and lIb (variants (BI-BI9 ).
V-Conventionally, tonal centricity is considered to be a factor of mode. Recognition of the tonal axis or center of any variant of Version II can be arrived at easily and correctly on this basis by any 20th-century amateur. In Tables I and II, they all end on G above middle C. It is felt to be the "Key note." A major triad (chord), with J:i as its root or bass can be found among the components of every variant. Indeed, whole tunes can be-and traditionally often are-harmonized ("chorded") with this chord alone, for the melody almost persistently outlines it. The mode is J:i-major.
There is no point in quarreling, here, with such self-evident reasoning in connection with our Version II, except to point out that in some cases it is an over-simplification and therefore obscures the facts (I) that there are in all these tunes other tonal axes or centers; (2) that it is upon the interrelationships of the lot that anyone of them depends, in part, for its primacy;
(3) but that equally important are the locations, durations, frequencies of occurrence and accentuation-all four of which are rhythmic functions; and, finally, (4) that although conventional major-minor musical ears will not go far wrong in this matter as far as our Version II is concerned, presupposition of self-eVidence cannot be trusted with the variants of Version I. They do not all end on the same component. No triad either major or minor can be formed of their components upon a final J:i. Secondary tonal axes are, in some cases, of such "strength" or "weight" as to challenge or weaken the status of the final. In a few instances, they have more weight and so suggest that the tune is "circular," that is, its last note lacks finality, seeming to lead back into the beginning.
Assuming that we are working only with accepted notations and not with the sound of singing, which interposes preliminary problems touched upon briefly here on other pages, the first step toward use of the factor of tonal centricity in a classification of ballad tunes is to arrange the notes found in each tune in ascending or descending order of pitch. This gives us the scale of the tonal aggregate. The second, is to observe which, if any, of the parts of the tune are confined to any particular part of the scale; for if they are so confined, they are likely to define separate tonal centers whose interplay is basic to the art of voice leading and the determination of centers upon higher levels. Such a treatment of our notations of variants Al to A5 appears in Example 6. in which it is clear that the first half of the tunes are confined to the five highest notes and the second, to the six lowest. This is typical of the majority in the roster.
The third step in this analytical process is to measure the durations of the individual notes in the aggregate of tonal materials. This is most easily done on the base of the shortest note-value in the notations. In Example 7, using variant A3 (Bryant) as the paradigm, the eighth-note is the unit of measure.
(NB. This step is undependable in all published notations in our roster; for they are given metrical signatures that do not change or are only seldom changed, so that the individual notes and rests must often have been shortened or lengthened, or, in the case of rests, even ignored; and we cannot be sure when this has been and has not been done. )
,
Example 7
Several interesting facts are shown here: in the first half, D's exceed high G's, but in the second, are exceeded by low G's. (On account of the repetit~ of its last half, the duration of Al are all higher than those of the other examples of Version Ia.) The high Q is, nevertheless important because it is high and in both first and second phrases is sounded twice with only an A in between. The low Q is important because it is the last note although it is heard only fleetingly until then. But II is clearly the axis of the ambit. It is strongly accented in both the first and second phrases and forms the semior mid-cadence.
The conclusion must be: in view of whatever the final significance of the factor of duration may be, Version I is heavily bi-centric. There would seem to be some objective support for a subjective notion that there is an impulsive movement upward from the final low Q, not only to the D but to the high Q, and consequently, a tendency toward circularity. Other criteria~maybe cited in support of a view that the version as a whole is n_ot te-1:Je regarded as outand-out circular, as may be Example 11 (f). Examples 11 (g) and (h) can be cited as haVing completely given way to the circular tendency, with radical transformation of the tune. The low G and E have been eliminated, the low ~ is merely touched upon and D is clearly the tonal axis.
Customarily, the tonal axis is regarded as having two distinct kinds of relationship to the aggregate tonal components of a tune. On the one hand, it may be located in about the middle of the ambit, when the tUne is referred to as "plagaL" Examples are variants A6, B1-B13 and B15-B19. On the other, it may be located at or very near either end, in which case it is referred to
as "authentic." Examples are variants AI-AS. Variants A7-All are "mixed."
These terms are, however, usually used in connection with tonal mode.
Vl--Tonal mode is usually conceived as a hierarchy of pitches based upon a grouping of the components of the tonal aggregate within an octave and arranged in ascending order from a tonal center called the "final" or "tonic. " The sameness but difference of two pitches an octave apart has perplexed musicologists and confused musicology throughout historical times. Any singer of a variant of our Version I would know that the sounds represented by the high and low irs, E's and A"s are different. Probably few, unless especially adept in shape-note hymn-singing, would think of them as "the same" notes. But in the tradition of British-American ballad-singing, two singers can sing the"same" tune an octave apart and believe it is the same tune, which--unlike traditional singers, say, in the Caucasus or the Balkansthey would not do and could not, even as a stunt, a second, fourth, or fifth apart. There is nothing mystical in the matter, but simply the fact that in music, as in many other things, a single object of attention can have more than one function in a particular process-as, for example, in a melody. The tendency in verbalization referent to these functions is to present them in what seem to be conflicting manners, as, for example, in the present instance when they are the same but different. The study of melody has viewed melodyespecially, sung melody-mainly in terms of modal octave-scales in which every constituent is primarily bound to the final or tonic. This context requires that any sound outside a particular octave must be considered the same as some sound within it. But any singer or student of any song that exceeds an octave knows that in the context of the singing the sound outside the octave is not substitutable by a sound within the octave. The conflict-and to some minds, the paradox-lies in the verbalization of the musical event, not in the musical event itself.
The conventional modal scaling of our two versions is shown in Example 8.
Example 8
Musical notation. Musical notation. Musical notation. Musical notation. Musical notation. Musical notation. Musical notation. Musical notation. Musical notation. Musical notation.Musical notation. Musical notation.Musical notation.Musical notation. Musical notation.Musical notation. Musical notation. Musical notation. Musical notation. Musical notation. Musical notation. Musical notation. Musical notation. Musical notation.
In the gross outlines of skeletal transcriptions, most of the variants of "Barbara Allen" in the holdings of the AAF8--as, indeed, in our whole roster --1lre, like thescales shown in Example 8, in pentatonic modes (which is to say, in scales of five degrees in the octave). Where comparison with the actual sound of the singing is pOSSible, as in sound recordings, additional degrees are often disclosed, as are also significant variations among the skeletal or more substantive sounds. The ascription of pentatonism is often, The number of pentatonic modes in the total musics of the world is probably uncountahle. British-American folk song makes use of only one type, namely, that which can be sounded with approximate accuracy by the black keys of the pianoforte, organ, or accordion, without what are semitones in the twelve-tone, equal tempered octave scale. Five different modes can be formed within this aggregate, for anyone of the five components can serve as final. Various nomenclatures have been suggested and used, but none has been universally accepted by students. Two of the most logical and useful have been proposed by Bronson15 and Kolinsky,16 and are shown above the staff in Example 9. The concern of both scholars has been the derivation and the order of the naming of these modes. Both sets of names are based upon the "circle of fifths, "in the terms of which many other fundamental musical relationships and processes are customarily explained. The circle of fifths is usually plotted by music theorists upward from F, as I-.Q-Q-~-b-~-~. Downward from F, it runs into flats; above~, into sharps, to extents unlimited in either direction except by the practibility of musical or musicological use. The advantage of thinking in terms of the circle of fifths is that any five successive steps in it comprise the components of a traditionally used pentatonic scale in the BritishAmerican folk idiom, any six, a hexatonic, and any seven, a heptatonic.
A more precise nomenclature is placed below the staves of Example 9, to designate progressions of degrees in a scale in conventional twelve-tone, equal tempered semitones. This provides a universal nomenclature for designation of scales of whatever number of components in whatever system of tuning,_ in terms of the equal tempered twelve-tone octave scale, now for some seventy years the standard among physicists and mUsicologists throughout the world. By reading the digits "2" and "3" (and "4" or more, if necessary) as "200," "300" (or more) the designation can be refined to the limits of the Ellis cent
system, which is a division of the octave into 1200 equal parts and accommodates the optimal discrimination of the ear to 1/100 of a semitone. In these terms, any particular British-American pentatonic scale can be compared with any other scale of whatever tuning of whatever number of degrees per octave.
students of this particular tradition have found it convenient to remember pentatonic scales by the absence in them of the two additional degrees in diatonic seven-tone major or minor scales plotted upward from the same tonic or final. The numbers of these absent degrees, preceded by minus signs
[chart]
scales as "gapped" merely because they comprise fewer than seven degrees is, however, to be discouraged. Some of them may already have filled in gaps in prototypes of a lesser number. Some may have had no ancestry of heptatonic relatives. And the heptatonic scale itself could be as inappropriately called "gapped" with respect to those of more than seven degrees, as, for example, of the twelve of tbe equal-tempered keyboard. According to Kolinski, the only scales of whatever number of degrees th!1t might legitimately be dubbed gapped would be those that (a) comprise the first and last of any series of successive fifths but (b) do not comprise one or more of the intervening fifths. For example, a tonal aggregate of 1'-g-~-.Q would be tetratonic; but its mode would be pentatonic because the Q between the g and the ~ in the successive series of fifths is omitted; therefore, it is gapped.
The pentatonic mode typical, then, of our Version I is P2323 (Bronson:rr 3 ; Kolinski, Penta-sol; -3-6): of Version II, P2232 (Bronson, 'J1' 1: 'Kolinski, Penta-fa; -4-7). But we should not,forget that most variants of Version I break into two somewhat opposed halves: the first giving us an aggregate of .Q-Q-1'-g-~ and a pentatonic mode Q-E-Q-~-Q, P 3223; the second, except where there is no opposition, as in Ie, a variable aggregate (E) -Q-~-Q-Q(
E) and a pentatonic mode Q:-~-Q.-Q-E, P2323. In Version 11, on the other hand, we have no such situation. The heavy durational weight of Q, though sometimes greater than that of g in the modal condensation, is actually divided, part above and part below, thus offsetting each other's pull, whereas in Version I, the weights of both Q and the Q an octave above are in only one, unopposed direction-upward.
VI-Prosodically accented beats, though often, as already pointed out, covered over by stronger random musical accents, do spell out, more in detail, the tonal phrase contour of criterion m. Classification in these terms has been admirably developed by Bronsonl! with the aid of statistical techniques too elaborate for use here. Suffice it to say: the iambic beat of the words remains evident, accent-wise, never entirely obscured by the distinctly nonverbal tempo, proportion, and additional, overlaying accents of the music. For example, the tonal pattern of the prosodically accented beats of variant A2 (Graham) is 1)-.9-.9-1), 1)-.9-.9-1); E-E-£;-E, .9-~-.9-.9. In arabic numerals, it would read: 5-8-8-5, 5-8-8-5; 7-7-4-7, 1-4-1-1.
VIII-Classification by phrase pattern is one of the most used but most abused resorts of ballad tune study. The pattern of the first phrase is designated
A. If the second is the same, or Virtually the same, as in the above accent pattern, some students will call it A or AI, others B, etc. If it is repeated in this or any other phrase exactly but upon a higher or lower pitch level, some will call it B, others, A3, A5, or A, A according to whether it is a athird or a fifth higher or lower, A', A", 30r ~ven A, and so on, or by an entirely different numeral. The method cannot express tonal identity with rhythmic diversity, or vice versa. I would be inclined to label Version I as AA' BC and Version II as ABCD, though D is often a mixture of A and B.
IX-Classification by phrase endings is one of the most useful and generally agreed upon of all criteria, also the qUickest and easiest to handle, once the octave scale and final of the mode have been established. Thus the phrase ending pattern of variant A2 conforms to that of the majority of our variant classes la and Ix, vis., 4571. The follOWing table shows the distribution in the AAFS holdings and in the total United states roster.
[charts]
X-Xl-The two criteria, "trade-mark" and cadential formulae, have already been mentioned in other context&-the characteristic third-phrase pattern of both versions and the "Scotch snap" skip of Version II. Tt may be worth noting that the United states variants show 80 per cent skips up for ends of second and fourth phrases but a total of only 20 per cent of the other six possibilities, whereas the British show 60 per cent twin skips down as over against 40 per cent of the others.
Variants of both our versions carry the words of other ballads and songs; but Version I does this far more commonly than does Version II. Participants in the current folk music revival movement in the United States will have found variants of the former most often in "Come All Ye Fair and Tender Ladies;" of the latter, in" Lord Bateman" ("Young Beichan," Child 53 ), "Geordie" (Child 209) and others. It appears with a religious text under the title "Heavenly Dove. ,,18
The chronological and geographical provenance of the tunes must, therefore, be studied independently of any particular text as well as in connection with each with which it is associated. Although my search for British sources has hardly more than begun, I can cite here as Example 10a, through the courtesy of S. P. Bayard, what seems to be the earliest printing of Version
Example 10
I, by D'Urley, in 1719,19 but titled only "A song," with the first line "There Was A Knight and He Was Young" ("The Baffled Knight, " Child 112). An introduction of four measures, that seems not to have been sung, is omitted here. In the 1750's, a variant of the version, also with a section not shown here, was printed by Oswald20 with the words of"Andro and His Cutty Gun" (later a favorite of Robert Burns and with a stanza of words for an expanded introduction). I give this, also through the kindness of Mr. Bayard, as Example 10(b). The tune appears as "The Dowie Dens of Yarrow, ,,21 shown as Example 10c, but was found by Sharp only three times in his monumental collection of nearly 3300 English tunes, as "Boyne Water, ,,22 "Navigation,,,23 and "London Pride, ,,24 a Morris tune. To the best of my knowledge, no collector has reported it with the words of "Barbara Allen" in England. But in the United states, we have it with these words in the 63 variants of our Version I, fourteen of them notated by Sharp himself25 in the Appalachians. Variants were reported from various states, among them Virginia, Maine, and Kentucky, from soon after 1900. Two informants testified they had learned it from elders in the family during the 1860's. It has been recorded recently as current in Scotland26 with the words of "Barbara Allen." (See Example
10d. )
Ewan MacColl kindly gave me the notation, 10e, of "Andro" as he heard his father sing it. The identity of its first three phrases with those of Mrs. McDowell's singing of "Barbara Allen," 10f is as striking as the differences in their final cadences. The latter, with Examples 10g2T and 10h-all three or PiHs to ""~e Me'--holy representative of the mutations in the American roster-show the progressive weakening of the lower tonal center and strengthening of the higher, with consequent loss of the bimodality of the majority of the roster.
Through the kindness of D. K. Wilgus I am able to show a variant that takes the opposite tack from that of Examples 1Of-h. This was notated from a dubbing of the singing of Professor Patrick W. Gainer28 of the University of West Virginia who tells me that tills variant has been traditional in his family, which came to the United States in the 18th century and which had for generations been a fiddling and singing family, both of secular music and of the shape-note hymns introduced after the Civil War.
Example 11
Musical notation
As can be seen in Example 11, there is an addition of two notes, G and A at the beginning, compensation being made by deletion of one or two at the end of the phrase. I can find no parallel for the addition of these or any notes at the beginning of the tune. But the same notes, one or both, sometimes begin later stanzas in the singing of Miss Stewart and Mrs. McDowell. The net result of the Gainer variant is transference of the factor of bimodality inherent·in the version from an axis Q-to-G to one of C-to-F. It is done by (1) increasing the notes outlining an~-major chord with an accented A on the beat of the first phrase, (2) accenting components of the chord in subsequent phrases, and (3) harmonizing the final Q. with a C-major triad to serve as a dominant of a new tonic sounded at the opening of the next stanza, as in Example 3. Thus, although it still preserves the circularity also inherent in the version, it weakens the original bimodality. For!, is not one of the principal components in the majority of variants. One might expect from this modal instability that a further mutation would abandon the circularity as well, by substituting a final F in a clear-cut major-mode cadence. And this is just what is reported, by a notatjon in the University of Virginia archives, to have been the singing of Mr. Ed. Davis, of Shipman.29 (See Example 12).
The modal instability of Version I is, then, one of its most interesting features, but at the same time one that makes investigation of its chronological provenance unusually difficult. One might suppose that the re-appearances of D'Urfey's final 12 in some 20th-century American recordings (10h, 14) were evidence of the marginal survival in the colonies of an older tradition that had otherwise been lost. But D' Urfey' s notation is itself open to suspicion. The Introduction, omitted in our notation here, is unashamedly minor-modal, even to a raised leading tone, Q sharp. We cannot be sure that he did not "correct" or "improve" an existent final Q and any other notes standing in the way of a fashionable D-minor ending. For until long after him, the "common people," though they did have some quaint tunes that everyone liked, did not know how to sing them-a complaint still heard even in the 1960's.
Speculation upon the relative chronological provenance of our two versions is also interesting. The somewhat sombre, non-major-minor archaism of Version I may seem to many necessarily older than the sweetly familiar major ring of Version II with its shorter history in print, (having first appeared and with the words of "Barbara Allen" in 1839 [30] as in Example 13a). Perhaps it is. But in this connection it is wise to keep three facts in mind: (1)
the identity of a folk tune is very difficult to pinpoint, so that the very notion "when it began to be itself" is metaphysical; (2) there were tunes in various modes in the 13th century that would pass today as major ("Sumer is icumen in" is an example); (3) Phillips Barry, the first and one of the most profound students of British-American folk song wrote:
It is no longer safe to say of any ballad tune, that, because it is cast in an ecclesiastical mode, it is necessarily ancient. On the contrary, it
may not be older than the singer from whom it was recorded. Nothing in
folk music is more evanescent than modality: instances are demonstrable
of the origin of modal tunes, built up from variation of a phrase of
a major air.[31]
The chronological provenance of our two versions, of course, intimately fuses with the geographical, so that historical changes in the tradition must be plotted with both in mind. With respect to the latter, there seems to be a general but untested agreement. Chappell [32] wrote, "Under this name the English and the Scotch have each a ballad, with their respective tunes and a comparison will show that there is no similarity between the tunes." Rimbault[33] and Graham[34] writing not long afterward concurred, as have others.
A second English tune is given in Example 12b.35 These two seem most widespread in England and have been reprinted with little or no alteration in many British publications, both scholarly and for popular use. Although often reprinted in the United states in cheap song-books and school textbooks, neither 13a nor 13b is found in the AAFS. In each case, when, rarely, a folklore collector has published either, there has been strong suspicion of copying from British sources by literary-minded urbanites. The resemblance of these to the norms of our Version II and the recordings on disc L54 is not very close. A few British variants do show a close resemblance, as, for example, the one sung for R. Vaughn Williams in 1905, shown as Example 13c. 36 The closest so far coming to my notice is a recent recording from Ireland,37 shown as Example 13d. The collector reports "This ballad seems to be the most widely known of the 'Child' ballads found in Ireland." Is it possible that the two versions making up over 95 per cent of the United States repertory stem from Scotland and Ireland, respectively, and least, or indirectly, from England?
The Scottish tune most Widely printed in England and Scotland as the Scotch (Scottish) version is shown as Example 13e. It appeared first with the words of "Barbara Allen" in Oswald's Pocket Companion 38 already cited. There are no recordings of it in the AAFS. It has not been found in print in the United states. For completeness, I give it here as sung by Ewan MacColl. [39]
The influence of print, so strong upon the words of the ballad, seems to be non-existent in the case of the tunes. Of the more than two hundred songsters examined by Ed Cray in connection with the present study or cited by folklorists, only four have been found with notations for" Barbara Allen, [40] although many contained the words. And these tunes have not been members of the classes of variants found in oral transmission by folklorist collectors. They are not found in the AAFS roster. Fiddlers have often filled blank-books with notations of the tunes they play or wish to learn. But although singers have as often written out compilations of the words of their repertory, notations are so rare as to be negligible as factors in the study of the tunes of our ballad.
More difficult to trace is the influence of commercial records upon country singing. From 1923 on, these inexpensive articles, often sold in five-and ten-cent stores, flooded the country from coast to coast for a full ten to twenty years before the folkloristic field recordings deposited in the AAFS were made. They formed the staple repertory of rural radio stations. Indeed, D. K. Wilgus assures me that country singers were heard over the radio even before their singing was released on commercial records. It is inconceivable that many of the informants of the folklorist collectors were not accustomed to hearing their own songs sung both from recordings and over the radio. In some cases, we are assured by the singers themselves that they learned a song by hearing it from such sources, or had actually sung it for broadcasting. Some folklorist collectors feared "contamination" by radio and record and avoided collection of songs said to have been learned or sung in connection with them. Others, we know, thought so little of the matter that they did not even question informants about it. Probably, before 1930, most of the singing that was commercially recorded or broadcast was not very different from what was normally current in rural areas. Considering the admittedly wide popularity of "Barbara Allen," it seems strange that more country singing of it is not found on commercial records. With the generous help of such specialists in the field of hillbilly music discography as Edward Cray, Joe Drochetz, Will Roy Hearne, Ed Kahn, Guthrie T. Meade, Jr., and D. K. Wilgus, this writer has been able to make a list of fifty commercial records of the ballad by ten different singers under thirty different labels. Often, a single matrix appeared under a number of different labels. Sometimes the original singer's name or the title was changed. With such hurdles to surmount, it may not be surprising that the discography of country and hillbilly music has grown slowly. From such as it is, however, we may suppose that the most Widely distributed releases were those of Marion Try Slaughter, a small-part grand and light opera singer in New York in the 1920's. He was born in Jefferson, Texas, in 1883, and sang under at least thirty known pseudonyms, the most used being Vernon Dalhart.[41] He was responsible for the sale of millions of discs from 1927 to 1930, one of which-"The Wreck of the Old 97"-is said to have netted him over $100,000 royalties. With his collaborator, guitarist, and "idea man," Carson J. Robison, he sang for commercial recording (under at least nineteen different labels) or published in cheap, wood-pulp songbooks (for purposes of copyright claim) at least four different variants of our Version II. Two of them are found in the AAFS holdings. One is'the Widely known Class lIb, which was collected, notated, and printed in both England and America, long before Dalhart or Robison began their country music careers. The other is unique. It is in the "odd-and-even" stanza or two-part form. In this there are two slightly different but related tunes of four phrases each, one for the odd and the other for the even numbered stanzas (as in the Jacobite song "Wae's Me for Prince Charlie"[42] ), (a tune, by the way, occasionally found carrying the words of other ballads, as well as those of "Barbara Allen").[43] Two closely associated singers in Florida sang identical, note perfect renditions of the even stanza tune alone of the Dalhart and Robison "copyrighted" tune. (see AAFS 987B1 and 988B1). A Mississippi singer sang the whole tune, but in g instead of a. (see AAFS 3111B1). To the best of present knowledge, only other report of this variant is from North Carolina,44 where it was sung
with considerable variance, including a confusion of odd and even stanza tunes similar to that mentioned above with respect to first and second halves of single-stanza tunes. The other two Dalhart and Robinson variants are note-for-note copies of printed British tunes that have not taken root in oral tradition in the United States. Before we conclude that the influence of the commercial record and printed songbook upon oral tradition in music in the United States is as slight as the above account would seem to indicate, a far more extensive survey of surviving discs and a sufficient number of similar studies of other ballads would have to be made. The basic difficulty involved would be to show that the singer recorded on the record did not himself learn the tune from oral tradition or that it was not already current before the recording.
No serious study of the ballad tune and its singing style in the 20th century can afford to omit consideration of the factor of instrumental accompaniment. Although Widely regarded by conservative students as an innovation-and disapproved by them-we cannot be sure that it was never traditional in the British Isles and is not in America. It seems not unlikely that in the early days of its development the ballad was accompanied by the harp, fretted zither, or the bowed lute. There are many reasons to suppose that ballads were sung where those instruments were played-and competently played.45 For ballad singing may not always have been, as it came to be in the 19th century, mainly a rural-even peasant-art. However that may be, folklorists have assured us that during the 19th and 20th centuries the unaccompanied singing of ballads has been the unexceptionable rule among the most highly esteemed carriers of the tradition in England. Therefore, when Sharp made his famous field collection trips in the Appalachian mountains, 1916-18, his schedule (we may suppose) was prepared in such a way as to have him hear chiefly unaccompanied singing. As a consequence, a belief that such singing was the only traditional one in the United States became accepted by many students. It is clear that singing style is influenced by accompaniment and that singers who sometimes use accompaniment and sometimes do not, may have, virtually, two distinct singing styles. We may prefer one to the other, or find both acceptable when competently done and preserving the basic essentials of the tradition. But when commercial recording was begun, banjo and guitar accompaniment was customary; and judging from the skill with which this was done, it must have been a common practice long before Sharp's visit. It may be significant, however, that among the holdings of the AAFS we have no accompanied renditions of our Version I from a country singer. Among commercial, or semi-commercial, recordings, I know of only the one shown as Example lI~. Accompaniment of Version II, on the other hand, is common both in field recordings of country singers and on commercial discs.
By way of summary, it may be stated that in the light of present knowledge and in view of present techniques of study, no such entity as "the Barbara Allen tune" can be set up other than for temporary convenience. The fact that with a few intermediate steps we can easily change one version into the other must be regarded in the light of the fact that we can change either version into any other tune of like length with as little, less, or more ease. Melodies are, by their very nature, infinitely changeable or interchangeable. Shown in melodic contour, there is considerable resemblance between the two versions- certainly enough to make possible, through precisely such variation as is shown by the notations in this article, transformation of a variant of one into a variant of the other. Effected little by little, as would be much ordinary variation in oral tradition, only three principal steps are required:
(1) to lower the level of the first phrase of I or raise that of II; (2) to twist the simple arcs of I ("B" type, in my table of basic contours of tonal variance46 or smooth out the zig-zag courses of the four phrases of II (all "N" or "F" types, in the table cited); (3) to blur or make ambiguous the modal functions during the process of change, leaVing to the final appearance of the variant changed into, its clear modality. With such a transformation in view, the objective can be reached almost unnoticed in the singing of five or six consecutive stanzas of the ballad. Of course, no such objective might occur to a carrier of the tradition. But we have, in a recent publication,47 a variant of Version I with "Scotch snap" of Version II at mid-and final cadences. In another,48 the whole fourth phrase is a variant of the typical form of Version
Example 14
Musical notation.
II. And about 25 per cent of the total American roster of Version I show the last two short-short or short-long sung syllables of second and fourth lines, typical of Version II. Of the opposite transformation-long-short or longlong mid' -and final cadences typical of Version I-less than 3 per cent of a total roster over twice as large are found in Version II; and some of these are suspect of notational ineptness. Poladian has perceptively provided an "intermediary class" of "Barbara Allen" tunes.49 While not accepted here on account of its small size and heterogeneous content, such a class might very well be found convenient by more mature research. In the present study, it will have been noticed, two groups of six items each-"leftovers" of the classificatory process-have been listed in the table on page 26 as"miscellaneous." But they could not be thrown into one class of twelve items.
As they stand, the two versions have such distinct characters that it is very easy to determine when one takes on a feature of the other. That one and the same ballad should be sung to two such different kinds of tune&--and, with rare exceptions, only these two-should certainly serve as a restraint upon the reckless positing of one-to-one correspondence between texts and tunes. Still, the question "why these two and only these two?" cannot fail to pique our curiosity.
Notations
Most of the transcriptions of the thirty variants in Tables I and II were made from first stanzas of singings only. In some cases, the recordings were partially defective and a later stanza was chosen for pressing on disc L54. In others, first stanzas were not typical of majority usage in the respective singings of the complete ballad. The singer was not, as it were, in his stride. Sometimes, too, there has been a slip in an otherwise acceptable first stanza. Except for excessive cost of publication, complete notations of whole singings must remain the scholarly ideal.
Not all notations are equally" skeletalized" by sung syllable. Some, by this technique, for example, variants All (Jackson), B3 (Parks), B17 (Carr) and B18 (Tarwater) -would have so misrepresented what actually occurred at the recording session that a more inclusive treatment seemed indicated. The nature of a still fuller transcription can be seen in Table IV between the upper graph of the pitch line and the lower, of the loudness.
It would be a pleasure to be able to mention the names of the many people whose advice and counsel I have sought in this excursion, not outside but by the side, of my main interest. But the list would be too long for inclusion here. Of course, the guide lines for study were laid down by the early comparative musicologists in Europe. For the United States, Phillips Barry, Helen H. Roberts, George Herzog, S. P. Bayard, and B. H. Bronson have been my main reliance. The Institute of Ethnomusicology of the University of California at Los Angeles, under the direction of Mantle Hood, has provided facilities unequalled for the work. Edward Cray has checked the transcriptions of the words. Sam Chianis has made parallel notations of the thirty strips as a check against mine; and I have often taken his alternative.
Footnotes:
1. Reprinted from Selected Reports of the Institute of Ethnomusicology, University of California, Los Angeles, Vol. I, No.1.
2. Samuel P. Bayard, "Aspects of Melodic Kinship and Variation in Br itish-Ame ri can Folk Tune s . " Papers Read at the Int e rnat ional Congress of Mus i cology , New York. 1939 (New York: Ame r i c an Musicological Soc i e ty. 1944) , pp. 122-129, "Prolegommena to a Study of the Principal Me lodic Fami l i e s of Br i t i sh-Ame r i can Folk Song . . . Journal of American FolHor e , vol . 63. (1950) . pp. 1-44.
3. Be r trand H. Bronson, "Some Observations about Melodic Variation in Br i t i sh-Ame r i can Folk Tune s ," Journal of the Amel"ican ,",uBiaologicaZ Soui e t v , vol . 3, no. 3 (1950) , pp. 120-134; "Toward the Comparat ive Analys i s of Br i t i sh-Ame r i can Folk Tune s ," Journal of American Fol k lor e , vol . 72. no. 284 (1959) , pp. 165-191.
4. Be r t r and H. Bronson. The Tradi t ional Tunes of theChi ld Bal lads , vol s . 1 and 2 (Pr inc e ton: Pr inc e ton Unive r s i ty Pr e s s . 1950 and 1962)
5. Charles Seeger, "On the Moods of a Musk: Logic. " Journal of the Ame r i can Musi cologi cal Soc i e t y , vol . 1: ) (1960) , pp. 24H-249.
6. St i th Thompson. Mot i f - Inde x of Fol k Li t emtul " e (Bloomington, Ind, : Indi ana Un versi ty Press. 19:J2-19:J6;· revised- and enlaq?;t'd. 1955-195H) , 6 vols.
7. Francis J. Chi ld. The English and Scottish Popuar Ballads (Boston: Houghton Mi f f l in Co. . 1HH2-1R9H: New York. The Folklor e Pr e s s . 1956) , 5 vol s ,
8. Check list of Recorded Songs in the English Language in the Arohiv e of American Fol k Song to July . 1940 (Washington: Libr a ry of- -Congr e s s . 1942) . 3 vol s .
9. Charles Seeger, "Singing Style," Western Folklore, vol. 17, no. 1 (1958), pp. 3-11
10. Bertrand H. Bronson, "The Murphology of Ballad Tunes." Journal of American Folklore, vol. (;7 (1954). pp. 1-1:1.
11 Edward A. Sonnenschein, What is Rhythm? (Oxford: 8. Blackwell. 1952).
12. "David S. Raven. Greek Meter (London: Faber and Faber Paul Maas. Greek Met'!'e (Oxford: Clarenden Press. 1962).
13. Loraine Wyman. Lonesome Tunes (New York: G. Schirmer. Inc. 1917) p. 1.
14. Cecil J. Sharp, English Folk Songs from the Southern Appalachians (London: Oxford University Press. 1932, 2nd edition, 1952.)
15 Bertrand H. Bronson, "Folk Music and the Modes," The Musical Quarterly, vol. 32 (1946), pp. 37 -49.
16 Mieczyslaw Kolinski, "Musicological Analysis," in Melville J. and Frances S. Herskovitz, Suriname Folklore (New York: Columbia University Press, 1936, pp. 489-740; "Classification of Tonal Structures, "in Studies in Ethnomusicology, vol. 1 (1961 l, pp. 38-76.
17 Bertrand H. Bronson. "Toward the Comparative Analysis of British-American Folk Tunes, " Journal of American Folklore, vol. 72 (1959). pp. 165-191.
"Thomas D'Urfey', Wit and Mirth: • -~ «.,,= (Lonon:.d W
Pearson. 1719-1720), vol. 5, p. 112.
20 James Oswald, The Caledonian Pooket Companion (London: James Oswald, 175-?),
Bk.6. p. 4.
2'Frank Kidson, Traditional Tunes (Oxford: Chas. Taphouse and Son, 1891), p. 22.
22Ceoil J.~ Sharp, CoHeotion of EngZish Folk Songs in tM Clare CoHege Librazoy,
microfilm 9547 in Music Library of lhe University of California at Berkeley, notations
numbers 2419 and 2420.
23 Idem, notation 1757.
2' ldi}m, notation 2597.
25Cecil J. Sharp, English Folk Songs from tM Southern Appalachians.
2. Luoy Stewart, Traditional Singer from Aberdeenshire, Sootland, collected, edited, and annotated by Kenneth S. Goldstein, Folkways Records FG 3519 (New York, 1961), vol. 1: Child BaHads, Side 2, band 4; Jilmry Stewart, collected and edited by Jean Ritchie, Collector Ltd. Ed.. LP 1201, side A, band 7.
27 Jan P. Schinhan, Frank C. Brown CoHeotion of North ClU'OZi1I4 Folklore, vol. 4,
TM Musio of tM Ballads (Durham: Duke University Press, 1957), p. 62, number 17G.
"Patrick Gainer, Folk Songs of the Alleghenny Mountains, Folk Heritage Record, DB 2122.
29Arthur K. Davis, Jr., TY'aditional Ballads of Virginia (Cambridge: Harvard University Press, 1929), p. 578, (V).
30 William Chappell, A CoHection of National English Airs (London: Chappell, 1839), vol. 2, p. 114, no. 242.
31PhillipS Barry. "American Folk Music," Southern FolkLore Quarterly, vol. I,
no. 2 (1957), pp. 29-47.
"Chapell,op. cit., Vol. 1(1840), p.183.
33Edward F. Rimbault,Musical Illustrations of Bishop Percy's Reliques.. .(London:
Cramer, Beale and Co., 1859 l, p. 35.
"George F. Graham, The Popular Songs of Scotland (Glasgow: J. M. Wood and
Co., 1887), p. 81, footnote.
"Kidson, op. cit., p. 37.
3'Journal of the Folk-Song Society, no. 7 (1905), n, second part, p. 80 (second
version), sung by Mrs. Bennefer.
31The Lark in the Morning, collected by Diane Hamilton, Tradition Records, TLP.
1004, sung by Thomas Baynes.
380swald,
op. oit., Book the Second, p. 37. -,' 38 Ewan MacColl, in The Englieh an<!. .Soottieh Popular Ballads, sung by Ewan MacColl and A. L. Lloyd, edlted and produced by Kenneth S. Goldstein. "The Amateur's Songbook (Boston, E. Howe, Jr., ca. 1843), p. 36; Heart Songs Dear to the Amerioan People (Boston: Chapple Publishing Co., 1909), p. 247; Phillips Barry, British Ballads' from Maine (New Haven: Yale University Press, 1929), bibiiography cites Charlie Fox's Minstrel Companion [Philadelphia, Turner and Fisher, 1860] and 1'rifet's Monthly Budget of Musio_ no. 15, n, p. [March, 1892
41 Jim Walsh, "Favorite Pioneer Recording Artists, "Hobbies, the Magaaine for Collectors,
vol. 65, nos. 3-10 (1960, May-December).
42 George P. Jackson. Spiritual Folk-Songs of Early America (New York: J. J. Augustin, 1937), p. 58.
43 Reed Smith, South Carolina Ballads (CantJridge: Harvard University Press, 1928), p. 13C
44 Schinhan, op. cit., p. 61, no. 27 B.
45 Chappell op. cit., "An Essay on the Ancient Minstrelsy of England." p. 2ff.
46 Charles Seeger, "On the Moods of a Music Logic.· 11
47 Archur K. Davies. Jr.. More Traditional Bal"Lads from Virginia (Chapel Hill: University of North Carolina Press, 1960). p. 191.
48George W. Boswell, "Kentucky Folksongs in the Tennessee Archives," Kentucky
Folklore Record, vol. 4, no. J (1958). p. 118.
"Poladian, op. cit., p. 85.
COMMENT ON THE WORDS
by Ed Cray
Contrary to expectation aroused by "Barbara Allen's" remarkable currency, the words of the ballad exhibit little variation in oral tradition- this lack of change is probably the result of the constant "corrective" of print.
In the North American tradition, in spite of some seven hundred collected texts, there are but four basic versions of the ballad, each readily identifiable by its first stanza. Among the four, there is virtually no borrowing of motifs or stanzas, and recovered texts tend to be fairly complete. Published fragments, at any rate, are rare.
Conjecturally, the oldest texts are those which begin: "It fell about a Martinmas time /When the green leaves were a- fallin'." These "Martinmas" versions, more specifically the traditional Scottish variants represented by Child C, may contain a legacy motif where in the dying lover leaves Barbara a series of gifts, including a bowl of his heart's blood. (Child thought the legacy mean stuff and did not print an available text which contained it. No "Martinmas" texts were found among the AAFS recordings available for this study.
Two other variants of the "Martinmas" group are less old: Child A, which in spite of a lively history in print has rarely been collected from oral tradition; and a Forget-me-not Songster text, identifiable by the hero's offer to make Barbara mistress of seven ships. This latter variant has entered oral tradition in the United States- a tribute to the popularity of the songster which reportedly had multiple press runs in the 1840's totalling one million copies. The Child A text, from Allan Ramsay's Tea- table Miscellany, seemingly has been most reprinted in those literary collections of "olden ballads", which rarely were distributed among the folk; the Forget-me-not Songster, on the other hand, was aimed at the mass, and therefore the folk, market.
The second large group of versions, most conveniently dubbed the "Early Early" group, begins: "So early, early in the Spring/When little birds were singing/A young man on his death bed lay/For the love of Barbara Allen. " (See variants A5, 7.) This group, in turn, neatly divides itself into two variants types, those which identity the dying lover as "Sweet William" and those which speak of him anonymously only as a "young man. " There is a notable geographic dichotomy between the two variants; the "Young-Man" texts are generally from the tradition of the northern woods, the "Sweet William" variants largely from the southern mountains .
While the "Martinmas" and "Early-Early" versions remain relatively pure in oral tradition, these groups are markedly less popular in the New World than the third and fourth textual types, the "Scarlet Town" and "May" versions. Only two of the variants recorded on Record L54, those sung by Kate Singleton (variant A5) and Mary F. Farmer (variant A7), are not "Scarlet Town" or "May" texts.
The third group of texts usually begins "In Scarlet Town where I was born" (see variants A2, 10; B3, 5, 8, 9, 11, 13) and is divided into a series of variants although there is less positive demarcation between these.
Closely identified with the printing houses, the "Scarlet Town" version breaks down into two variants of the parent broadside and a series of rewritten or slightly altered variants distinguished principally by shifts in the locale of the action.
Of the two "Scarlet Town" or broadside variants, one identifies the hero as "Jemmy Grove," the other as "Sweet William" again. The "Jemmy Grove" texts have a two stanza introduction, the usual "Scarlet Town" and a second concerning events in the month of May. The "Sweet William" variants employ a three stanza introduction: the usual "Scarlet Town, " a "May" and a third dealing with events in the month of June. Jemmy invariably sends a "man;" Sweet William dispatches "a servant" to seek the girl. In the "Jemmy Grove" texts, Barbara, "with scornful eyes," looks down on the hero' s corpse; at this point, the "Sweet William" texts substitute "The more she looked, the more she mourned (grieved)."
There is a series of sports in this group: texts which localize the song to Stoney Town, to Yonders Town (variants B10, 14) or in London, in which some of the language may be local dialect, for example, "howdy" and "bursted," and in-which the hero, usually unnamed, sends not a man or a servant , but a "letter." Another localized variant beginning "In Scotland I was bred and born" probably is of English derivation; some versions contain the legacy motif of Scots tradition and identify the hero as a "squire," a title more of the Old World than the new. A fourth variant in this "Scarlet Town" or broadside group is also identified by a line in the first stanza which names Barbara as one of three maids living in Yonders Town.
Although most popular in the United States in terms of the number of texts recovered, the "May" versions are divided only into two variants. (See variants A1 , 3, 4, 6, 8, 9, 11; B1, 2, 4, 6, 12, 15, 17, 18. ) In the course of the ballad, Sweet William, invariably the hero, will either beg a kiss to keep from dying or at this point will lament that he will not improve his health unless he gets Barbara Allen. Her refusal in either case need not be present .
Extreme diversity of circumstantial detail within the "May" group is probably the result of a full flowering of oral tradition without the "corrective" influences of printed texts. Of the fifty-one broadside and songster texts recovered, none is a "May" version, yet within oral tradition in the United States and Canada , fully one-half of the texts are "May" variants. This is the case, also, with the total AAFS holdings. Fifteen of the thirty recorded examples of the ballad on this record, for example, are "May" variants. (Two of the thirty field recordings on the record lack first stanzas which would permit positive identification.)
The narrative style of "Barbara Allen" is distinguished by a lack of the expected ballad commonplaces. No one mounts a milk white steed or a dapple grey; Barbara twirls at no pins, and so on. This lack of cliches is not due to any lack of opportunity within the story line, but seemingly is due to the fact that the ballad is constantly refreshed in oral tradtion by 1) cheap print , and 2) frequent rehearings of other singers' versions. There is little need for the ballad cliches to advance the story if the singer knows the song well. The song itself may expand or contract at will. Five seems to be about the minimal number of stanzas for story- telling purposes. The maximum in the LC holdings are Mr. Wilson's 17 of which two stanzas are given in variant B11 and Mr. Dartey's 18 on AAFS No. 302B. When longer texts are found, the additional material is narrative, not lyric, the added stanzas advancing the story or, at least, telling more of it. "Barbara Allen" does contain one notable commonplace in all of its ver sions and vaiants: the rose-briar motif found also in "Lord Thomas and Fair Annet" (the parent text?) and "Sweet William and Fair Margaret" - two ballads that, like "Barbara Allen," have a theme of frustrated love.
One further note on the literary style: the ballad is told in a series of two stanza building-blocks which in themselves form a self - contained incident of the narrative. The two stanzas of each block or group are of ten very similar, truncated examples of incremental repetition as it were. As narrative building-blocks, these two-stanza segments may be added or dropped freely. When used, both stanzas are most of ten present, or to put it conversely, rarely will one stanza of a narrative segment fall into disuse. The general tendency is two-or-nothing.
The origin of the ballad is unsettled. H. M. Belden (whom this grouping of texts follows in part) cites a letter written by Mrs. Fannie Hardy Eckstorm saying that she and Phillips Barry "had satisfied themselves, before Barry's death, that as sung by Mrs. Knipp to the delight of Samuel Pepys in 1666, it was not a stage song at all but a libel on Barbara Villiers and her relations with Charles II. . . " (Frank C. Brown Collection of North Carolina Folklore , II, p. 111.) There is little corroborative evidence in documentary records of the Restoration; but neither is there contradictory. Pepys was a gossip- and one notoriously fickle in his feminine favorites. If on January 2, 1666, the promiscuous Mrs. Villiers was not in his favor, why did not Pepys cite the ballad (identified as "her," that is, Mrs. Knipp's, "little Scotch song" and presumably close to Child's C text) as an attack on the King' s mistress? If, on the other hand, Mrs. Villiers were in favor, it is unlikely that the volatile Pepys would cite the ballad favorably.
It may be that the actress's little Scotch song may have later served as the model for a satirical reworking which now survives as the "Scarlet Town" group. Farmer and Henley' s Dictionary of slang and Colloquial English identifies "Scarlet Town" as an obsolete punning name for Reading, Berkshire. Perhaps it is no coincidence that Barbara Villiers received from Charles II in the early months of 1668 Berkshire House which stood conveniently only a few hundred yards from St. James. A broadside hack who wanted to identify his subject but not lay him self open to libel or to the king's wrath might so disguise the satirical parody he had in mind to fashion- in the manner of the day- from popular song. If true, this would date the "Scarlet Town" texts no earlier than 1668, two years after Pepys first noted the "little Scotch song." Charles II died in 1680, reportedly telling the heir apparent , "Take care of my Lady Cleveland" (one of Mrs. Villiers' titles). This request is strikingly close to the line in the "Scarlet Town" versions in which the dying hero asks his friends to be good to Barbara Allen.
Significantly, the "Martinmas" texts predominate in published Scots collections; the "Scarlet Town" possibly satirical texts are little known. This would be the case if the "Martinmas" texts were the older and already established in oral tradition in Scotland before the "Scarlet Town" versions received the attention of the cheap printers. If the Barry-Eckstorm theory is to be credited, it may be that the traditional ballad was remade for satirical purposes as late as 1680, only after the king had died and was unable to protect his widely disliked and already frequently libeled mistress.
Whatever the origin of the ballad, or anyone particular version of it, "Barbara Allen" remains the most frequently sung of the English and Scottish popular ballads .
Footnotes:
1. Reprinted f rom Se l e c t ed Repor t s of the Ins t i tut e of Ethnomus i cology, Unive r s i ty of Ca l i forni a , Los Angeles , Vol. I , No.1.
2. Samuel P. Bayard, "Aspe c ts of Melodic Kinship and Var iat ion in Br itish-Ame ri can Folk Tune s . " Papers Read at the Int e rnat ional Congress of Mus i cology , New York. 1939 (Ne \ \ ' York: Ame r i c an Mus i cologi c a l Soc i e ty. 1944) , pp. 122-129, "Prol egommena to a Study of the Pr inc ipal Me lodi c Fami l i e s of Br i t i sh-Ame r i can Folk Song . . . Journal ofAmerican FolHor e , vol . 63. (1950) . pp. 1-44.
3. Be r trand H. Bronson, "Some Observations about Melodic Variation in Br i t i sh-Ame r i can Folk Tune s ," Journal of the Amel"ican ,",uBiaologicaZ Soui e t v , vol . 3, no. 3 (1950) , pp. 120-134; "Toward the Comparat ive Analys i s of Br i t i sh-Ame r i can Folk Tune s ," Journal of American Fol k lor e , vol . 72. no. 284 (1959) , pp. 165-191.
4. 'Be r t r and H. Bronson. The Tradi t ional Tunes of theChi ld Bal lads , vol s . 1 and 2 (Pr inc e ton: Pr inc e ton Unive r s i ty Pr e s s . 1950 and 1962)
5. Char les Seeg"cr, "On the Moods of a Musk: Logic. " Journal of the Ame r i can Musi cologi cal Soc i e t y , vol . 1: ) (1960) , pp. 24H-249.
6. St i th Thompson. Mot i f - Inde x of Fol k Li t emtul " e (Bloomington, Ind, : Indi ana Un versi ty Press. 19:J2-19:J6;· revised- and enlaq?;t'd. 1955-195H) , 6 vols.
7. Francis J. Chi ld. The English and Scottish Popuar Ballads (Boston: Houghton Mi f f l in Co. . 1HH2-1R9H: New York. The Folklor e Pr e s s . 1956) , 5 vol s ,
8. Check list of Recorded Songs in the English Language in the Arohiv e of American Fol k Song to July . 1940 (Washington: Libr a ry of- -Congr e s s . 1942) . 3 vol s .
9. Charles Seeger, "Singing Style," Western Folklore, vol. 17, no. 1 (1958), pp. 3-11
10. Ber t rand H. Bronson, "The Murphology of Bal lad Tunes " Journal of American Folklore , vol . (;7 (1954) . pp. 1-1:1.
11. Edward A. Sonnenschein, What i s Rhythm? (Oxford: 8. Black "cl l . 1952) .
12. David S. Raven. G'!'eek Mete' ! ' (London: Fabe r and Fabe r . 1962) ; Paul Ma a s . G'!'eek Met'!'e (Oxford: Cl a r enden P, ' e s s . 1962)
13. Loraine Wyman. Lonesome Tunes (New York: G. Schi rmer, Inc . 1911) p. 1.
14. Si rva r t Pol adi an, A Study in var iant s . publ i shed by the Unive r s i ty of Ca l i f orni a . Depa r tment of Mus i c , a s a Repor t on Uni t A-25 of Of f i c i a l Proj e c t No. 65-1-08-62, conduc t ed unde r the auspi c e s of the Works Proj e c t s Admini s t r a t ion; Edwa rd B. Law ton. J r . , Fa cul ty Advi s e r ; Sidney H. Robe r t son, Supe rvi sor (Be rke l ey, J anua ry, 1940) . Type s c r ipt , vi i i + 91 + 117 pos i t ive photos t a t s of mus i c not a t ions , 26 page s . P.46.
15. Ce c i l J . Sha rp, E"{/tish FoZk SO"{/s from the Southe rn AppaZachians (London: Oxford Unive r s i ty Pr e s s . 1932, 2nd edi t ion, 1952. )
16. Be r t r and H. Bronson, "Folk Mus i c and the Mode s , " Th. Musical Quar t e r l y , vol. 32 (1946) , pp. 37 -49.
17. Mi e c zys l aw Kol inski , "Mus i cologi c a l Ana lys i s , " in Me lvi l l e J . and Fr anc e s S. He r skovi t z , Suriname Fol k lor e (New York: Columbi a Unive r s i ty Pr e s s , 1936, pp. 489-740; "Cl a s s i f i c a t ion of Tona l St ruc tur e s , " in Studi e s in Ethnomus icology, vol . 1 (1961 l, pp. 38-76.
18 Be r t r and H. Bronson. "Towa rd the Compa r a t ive Ana lys i s of Br i t i sh-Ame r i c an Folk Tune s , " Journal of American Fol k lor e , vol . 72 (1959) . pp. 165-191.
19 John G. McCur ry. The Soaial Harp (Phi l ade lphi a : S. C. Col l ins . 1868) . p. 23.
20 "Thoma s D'Ur f ey' , Wit and Mi r th: or PiHs to • ""~e -~ Me ' - -hol y «.,,= (Lonon: . d W Pe a r son. 1719-1720) , vol . 5, p. 112.
21 J ame s Oswald, The Caledonian Pook e t Companion (London: J ame s Oswa ld, 175- ? ) , Bk. 6. p. 4.
22 Fr ank Kidson, Tradi t ional Tunes (Oxford: Cha s . Taphous e and Son, 1891) , p. 22.
23Ceoil J.~ Sha rp, Collect ion of EngZish Fol k Songs in tM Clar e CoHege Librazoy, mi c rof i lm 9547 in Mus i c Libr a ry of lhe Unive r s i ty of Ca l i forni a a t Be rke l ey, not a t ions numbe r s 2419 and 2420.
24 Idem, notation 1757.
25 ldi}m, not a t ion 2597.
26 Cecil J . Sha rp, Engl i sh Fol k Songs from tM Southe rn Appalachians .
27 Luoy St ewar t , Tradi t ional Singe r from Abe rde enshi r e , Soot land, col l e c t ed, edi t ed, and annot a t ed by Kenne th S. Golds t e in, Folkways Re cords FG 3519 (New York, 1961) , vol . 1: Chi ld BaHads , Side 2, band 4; Jilmry St ewar t , col l e c t ed and edi t ed by J e an Ri t chi e , Col l e c tor Ltd. Ed. . LP 1201, s ide A, band 7.
28 Jan P. Schinhan, Frank C. Brown CoHeot ion of North ClU'OZi1I4 Fol k lor e , vol . 4, TM Musio of tM Bal lads (Durham: Duke Unive r s i ty Pr e s s , 1957) , p. 62, numbe r 17G.
29. Pa t rick Ga ine r , Fol k Songs of the Al l eghenny Mountains, Folk He r i t age Re cord, DB 2122.
30. Ar thur K. Davi s , J r . , TY'adi t ional Bal lads of Vi rginia (Cambr idge : Ha rva rd Unive r s i ty Pr e s s , 1929) , p. 578, (V) . Exampl e 12 Davis I' . I
31 Wi l l i am Chappe l l , A CoHe c t ion of National English Ai r s (London: Chappe l l , 1839) , vol. 2, p. 114, no. 242.
32 PhillipS Ba r ry. "Ame r i c an Folk Mus i c , " Southe rn FolkLore Quar t e r l y , vol. I , no. 2 (1957) , pp. 29-47.
33. Chape l l ,op. c i t . , Vol. 1(1840) , p.183.
34. Edward F. Rimbaul t ,Mus i c a l I l lus t rat ions of Bi shop Pe r c y ' s Re l ique s. . .(Lon don: Cr ame r , Beale and Co. , 1859 l, p. 35.
35. George F. Gr aham, The Popular Songs of Scot land (Gl a sgow: J . M. Wood and Co. , 1887) , p. 81, footnot e .
36 Kidson, op. c i t . , p. 37.
37 Journal of the Fol k -Song Soc i e t y , no. 7 (1905) , n, s e cond pa r t , p. 80 ( s e cond ve r s ion) , sung by Mr s . Benne f e r .
38 The Lark in the Morning, col l e c t ed by Di ane Hami l ton, Tr adi t ion Re cords , TLP. 1004, sung by Thoma s Bayne s .
39 0swa ld, op. oi t . , Book the Se cond, p. 37. - , '
40 Ewan Ma cCol l , in The Engl i eh an<!. .Soot t i eh Popular Bal lads , sung by Ewan Ma cCol l and A. L. Lloyd, edl t ed and produc ed by Kenne th S. Golds t e in.
41. The Amat eur ' s Songbook (Bos ton, E. Howe, J r . , c a . 1843) , p. 36; Heart Songs Dear to the Amerioan Peopl e (Bos ton: Chappl e Publ i shing Co. , 1909) , p. 247; Phi l l ips Ba r ry, Br i t i sh Bal lads ' from Maine (New Haven: Ya l e Unive r s i ty Pr e s s , 1929) , bibi iogr aphy c i t e s Char l i e Fox ' s Mins t r e l Companion [Phi l ade lphi a , Turne r and Fi she r , 1860] and 1' r i f e t ' s Monthly Budget of Musio_ no. 15, n, p. [Ma r ch, 18 92].
42 Jim Walsh, "Favor i t e Pione e r Re cording Ar t i s t s , "Hobbi e s , the Magaaine for Col l e c tor s , vol . 65, nos . 3-10 (1960, May-De c embe r ) .
43 George P. J a ckson. Spi r i tual Fol k -Songs of Ear l y America (New York: J . J . Augus t in, 1937) , p. 58.
44.Re ed Smi th, South Carol ina Bal lads (Cant J r idge : Ha rva rd Unive r s i ty Pr e s s , 1928) , p. 13C
45 Schinhan, op. c i t . , p. 61, no. 27 B.
46. Chapr_ll • op. c i t . , "An Es s ay on the Anc i ent Mins t r e l sy of Engl and. " p. 2ff
47. Charles Se ege r , "On the Moods of a Mus i c Logi c . · 11
48. Archur K. Davi e s . J r. . More Tradi t ional Ballads from Vi rginia (Chapel Hill: Unive r s i ty of Nor th Ca rol ina Pr e s s , 1960) . p. 191. 48George W. Boswe l l , "Kentucky Folksongs in the Tenne s s e e Ar chive s ," Kentucky Fol k lor e Re cord, vol . 4, no. J (1958) . p. 118. "Pol adi an, op. c i t . , p. 85.