The English, Scottish, and American Versions of the 'Twa Sisters'
by Archer Taylor
The Journal of American Folklore, Vol. 42, No. 165 (Jul. - Sep., 1929), pp. 238-246
[See Barry's comments on Taylor's article in The Two Sisters: Prolegomena to a Critical Study by Barry 1931.]
THE ENGLISH, SCOTTISH, AND AMERICAN VERSIONS OF THE 'TWA SISTERS'
BY ARCHER TAYLOR.
The versions of "The Twa Sisters" (Child 10) known in England and Scotland can be divided into two groups, English and Scottish, each with characteristic traits. The consequences which result from making such a division are of distinct importance both for this particular ballad's history and for ballad study in general. It will be most convenient to demonstrate the existence of these two groups before drawing any conclusions.
Many readily recognizable peculiarities of one group can be pointed out in Child B, perhaps the best representative of the Scottish tradition. We are not endeavoring to show that Child B is the oldest or the best version of the ballad. For our purpose it suffices that Child B contains in unmistakable form many, perhaps all those details which are characteristic of Scottish tradition. Whether these details are original or not is, at least for the present, beside the point. They serve to identify the Scottish tradition, which includes the following texts: B, C, D, E, F, G, H, I, J, M, N, O, P, Q, W, X. The significant details are italicized in the following text:
B 1. There was twa sisters in a bowr,*
Edinburgh, Edinburgh
There was twa sisters in a bowr,
Stirling for ay
There was twa sisters in a bowr,
There came a knight to be their wooer.
Bonny Saint Johnston stands upon Tay[1]
2 He courted the eldest wi glove an ring,
But he lovd the youngest above a' thing.[2]
3 He courted the eldest wi brotch an knife,
But lovd the youngest as his life.[3]
4 The eldest she was vexed sair,
An much envi'd her sister fair.
5 Into her bowr she could not rest,
Wi grief an spite she almos brast.
6 Upon a morning fair an clear,
She cried upon her sister dear:
7 'O sister, come to yon sea stran,
An see our father's ships come to lan.
8 She's taen her by the milk-white han,
An led her down to yon sea stran.
9 The younges[t] stood upon a stane,*
The eldest came an threw her in.[4]
10. She tooke her by the middle sma,
An dashd her bonny back to the jaw.
11. 'O sister, sister, tak my han,
An Ise mack you heir to a' my lan.
12 'O sister, sister, tak my middle,
An yes get my goud and my gouden girdle.
13 'O sister, sister, save my life,
An I swear Ise never be nae man's wife.'*[5]
14 'Foul fa the han that I should tacke,
It twin'd me an my wardles make.
15 'Your cherry cheeks an yallow hair*
Gars me gae maiden for evermair.'[6]
16. Sometimes she sank, an sometimes she swam,
Till she came down yon bonny mill-dam.
17 O out it came the miller's son,*
An saw the fair maid swimmin in.[7]
18. 'O father, father, draw your dam,
Here's either a mermaid or a swan.'[8]
19 The miller quickly drew the dam,
An there he found a drownd woman.
20 You coudna see her yallow hair
For gold and pearle that were so rare.
21 You coudna see her middle sma
For gouden girdle that was sae braw.
22 You coudna see her fingers white,
For gouden rings that was sae gryte.[9]
23 An by there came a harper fine,
That harped to the king at dine.
24 When he did look that lady upon,
He sighd and made a heavy moan.
25 He's taen three locks* o her yallow hair,
An wi them strung his harp sae fair.[10]
26 The first tune he did play and sing,
Was, 'Farewell to my father the king.'
27 The nextin tune that he playd syne,
Was, 'Farewell to my mother the queen.'
28 The lasten tune that he playd then,
Was, 'Wae to my sister, fair Ellen.'
The traits which have been italicized are characteristic of the Scottish tradition and are not found in the English tradition. More traits might perhaps be found, but the present number will probably suffice for our purposes. A ballad which contains these details may be confidently asserted to be of Scottish origin. Of course there is always the possibility that one may find these peculiarities in a ballad from the south of England, but we need not take this possibility into serious consideration.
Indeed it is a false and misleading caution to question this result without being able to adduce a single reason for doubt. So far as we know, and our knowledge is based on about thirty variants, - these details are not found in England. The situation is comparable to that of a botanist who asserts that palms grow only in the tropics. To be sure, palms may be found in Greenland, but he does not consider that possibility seriously. Nor need he, for his assertion is based on a relatively large number of observations, which are, be it noted, capable of being repeated at any time. Similarly our assertion that these characteristics are definitely Scottish is one which is based on a large number of instances. One might wish that the number were larger, but there is no occasion to feel greater uncertainty regarding the premises than does the botanist in the case of a rare plant. In one regard it might seem that the botanist has an advantage, viz., he can show that the palm is better adapted to warmer than to colder climates. Yet a moment's thought makes it clear that the ballad student has, at least theoretically, the same opportunity. He can prove that these details are in striking accord with Scottish or English life, e. g., they may exhibit phrasal peculiarities explicable only in the light of Scottish or English tradition[11] or they may contain allusions to Scottish manners. Naturally such a proof is often more difficult of accomplishment than the botanist's proof of climatic adaptation, but it is not always beyond attainment.[12] In view of what has been said we accept the *italicized traits as tokens of the Scottish tradition, at least until such an assumption is shown to be false, and in accepting them in this way we do not hesitate concerning the sureness of our footing.
In finding traits characteristic of English tradition we are confronted with serious difficulties. The English ballads are few in number, so few indeed that the absence of the Scottish traits is perhaps a more reliable mark than any other. Yet the geographical and chronological distribution of the English versions is such that we can see something of their development.
The broadside of 1656, "The Miller and the King's Daughter" (A), is followed by a version (Y) from the Wye valley which was sent to Bishop Percy before 1770. The remaining English ballads (or ballads deriving clearly from the English tradition), i. e., J, L, R, S, T [13] have been recorded since 1850. Only version S offers a serious problem. It is found in the Kinloch MSS, but concerning its source nothing is known.[14]
Significant characteristics of the English tradition are pointed out by Professor Cox[15] who did not note the bearing of his observations: "All three [West Virginia] ballads belong to the group represented by Child R, S, U, and Y, as is shown in particular by the refrain, the beaver hat, and the wicked miller."
Precisely these ballads, R, S, and V,Y-U was taken down on Long Island - are English and not Scottish ballads. Quite as significant as the presence of the English traits is the total absence of the Scottish traits. The absence of a trait is just as important as any other mode of identifying a regional variation, although we must bear particularly in mind the possibility that the trait has been lost in the natural processes of oral transmission. Further characteristics of the English tradition are the introduction:
There was a king of the North Countree,
And he had daughters one, two, three[16]
the rhyme swan: woman in the stanza:
'O father, oh father, I see a white swan,
Or else it is a fair woman [17]
and the stanza:
The miller was hanged on his high gate
For drowning our poor sister Kate.[18]
More peculiarities of the English tradition can doubtless be found, but these suffice: the introductory stanza, the refrain, the beaver hat, - which distinguishes the later English tradition from the earlier, - the omission of the adjective "yellow" in describing the girl's hair, the rhyme swan: woman, the wicked miller and his death. It may again be pointed out that these traits are not necessarily old. In fact, we can see clearly enough that several of them are corruptions of the original ballad.
We have thus demonstrated the existence of two distinct traditions in the ballad of "The Twa Sisters." A question which rises in our minds at once is the determination of the relation of American tradition to these groups. The most superficial survey reveals that the American ballads[19] never contain a detail characteristic of Scottish tradition. This complete absence of Scottish peculiarities makes it very unlikely that the American ballads took rise in Scotland. Such an argumentum ex silentio is naturally not conclusive, for one might maintain that all the Scottish traits had been lost in crossing the Atlantic. But the comparison with the English ballads settles the matter. Many striking similarities to English tradition are found, e. g., the introductory stanza,[20] the beaver hat as the lover's gift,[21] and the failure to specify the hair as yellow. The American ballads follow the later English tradition, --which is at least as old as Y, Bishop Percy's version of 1770, - in neglecting the harp and converting the story into a tragedy of jealousy and avarice which ends with hanging the miller on the gate.[22]
We need not be more detailed in our analysis, for the conclusion is obvious. The American tradition is derived from English and not from Scottish sources. Defective, corrupt, and contaminated as the American tradition is, it is nevertheless of some value in giving information about the English ballad. In this particular case we can go farther and point out that two stages of the English tradition have been preserved and that the existence of a third can be demonstrated. The oldest existent stage is that in the first recorded text (A), the broadside of 1656, which evidently maintained itself in oral tradition with some success, for the Welsh-English version (L) recovered about 1850 is clearly its descendant. But the buffoonery of A and L, e. g.,
L 3 And what did he do with her fair bodye ?
Fal the lal the lal laral lody
He made it a case for his melodye.
Fal, etc.
4 And what did he do with her legs so strong?
He made them a stand for his violin.
5 And what did he do with her hair so fine ?
He made of it strings for his violine.
6 And what did he do with her arms so long ?
He made them bows for his violin.
7 And what did he do with her nose so thin ?
He made it a bridge for his violin.
8 And what did he do with her eyes so bright?
He made them spectacles to put to his sight.
9 And what did he do with her pretty toes ?
He made them a nosegay to put to his nose.
cannot be original. Behind this expansion lies a simpler narrative, more or less like the Scottish ballad but possessing no doubt peculiarities which marked it as English.
Into the question of the relation of the English and Scottish ballads we need not enter now, for a full discussion will involve a comparison with the Scandinavian forms.[24] The simpler English ballad which lies behind the broadside of 1656 was, we may conjecture, the source of the later English ballads, which, beginning with Y of 1770, have uniformly lost all trace of the harp and have converted the story into a penny dreadful. From this later English tradition spring the American ballads.
An easy and obvious test of the correctness of our procedure is to see whether these distinguishing marks really do enable us to separate and identify versions of unknown origin, or, in other words, to forecast what sort of ballads we will find. It is no surprise to learn that Professor Belden has in his possession five unprinted American versions with the stanza:
The miller was hung at his mill gate
For drowning my sister Kate.[25]
We can scarcely go wrong in conjecturing that these versions exhibit other peculiarities of the later English tradition and no traits belonging to the Scottish tradition. Or take an unprinted version which has fortunately come into my hands at this juncture:
1. Two little sisters fair and gay,
Sing I down, sing I down,
Two little sisters fair and gay,
The boys are bent on me.
Two little sisters fair and gay -
The younger had the older way.
Refrain: I'll be kind to my true love,
For he is kind to me.
2. Johnny fell in love with the younger one,
The older didn't have one bit of fun.
3 Johnny bought the younger a gay gold ring,...
Johnny bought the other a gay gold ring.
The other did not have one thing.
4. Two little sisters walking by the stream.
The older pushed the younger in.
5 At first she'd sink and then she'd swim,
Till next she came to a miller's dam.
6 The miller took by the hand,
And brought her safely to the land.
7 The miller stripped her of her gold,
And pushed her back into the stream.
8 The miller was hanged[26] on the gallows high,
The sister was burned at a stake close by.
The refrain, the single figure of the miller, and his execution are enough to mark this extremely corrupt text as of English rise. It contains no Scottish traits. We are therefore not surprised to learn that it is an American version which I owe to the kindness of Professor E. M. Albright.[27]
This simple and methodical comparison of texts attains several important results. It defines regional varieties of this Anglo-Scottish ballad. The American texts are derived from one of these regional varieties, the English, and not from the other. This fact gives new importance to the American ballads, for they can be used to supplement the very scantily recorded English tradition. In such an employment of American ballads we must, however, be on our guard lest we overestimate their value, for they represent a late and perhaps corrupt stage of English balladry.
When we have thus identified peculiarities belonging to the English and the Scottish traditions, one question which we have thus far avoided follows as an immediate consequence: Which trait is original? This question involves the relation of English and Scottish traditions and perhaps even the ballad's whole history. To answer it we must call into court Scandinavian tradition and thus go beyond the bounds of our present undertaking.
_________________
Footnotes:
1. B 1, C 1, D 1, E 1, F 1, G 1, II, N 1, O 1, P 1, Q 1; (hall): H 1, M 1; (ball): J 1, W 1. The letter refers to the letter assigned to the ballad text in Child, English and Scottish Popular Ballads, and the number, to the stanza. Texts V-Y are found in Child I 493 ff.
2. C 2, D 3, E 3, H 2, I 2, O 2, O 3, t V 3, W 2. An obelisk (t) is prefixed to defective or corrupt versions.
3. C 3, D 4, H 3, I 3, Q4, VT 4.
4. C 7, E 5, F 4, G 4, H 6, M 7, O 5, Q 6, V 7.
5. This incremental repetition is preserved more or less well in several Scottish texts: C 9-13, D 7-Io, E 6-7, F 6-8, G 6-7, EH 7-10, I 7--9, N 10--11, P 7--I, Q 7-8, W 3-4. It must not be confused with the somewhat similar passage in R 5-6 and Y 6-7.
6. C14, D16, E12, PF18, HI15, I 12, M 14, O 9, P 19, Q 15, V 18, W 10.
7. In the great majority of Scottish versions it is the miller's daughter rather than his son who sees the floating body. This spontaneous fission of the miller as in the English versions into miller's daughter or miller's son as in the Scottish versions is a frequent enough phenomenon. So too is the substitution of a male for a female actor or vice versa. I list those versions which show fission but I make no note of sex: C 16, D 12, E 9, F 11, G 10, H 12, I 10, M ii, O 7, P 12, Q II, V II, W 5, 7.
8. CI6, DI3, E, F12, G2, , H13, I 11, M 12, N 14, O 8, P 13, Q 12, V 12, W 6. Note that the Scottish nature of X 1 appears in its single stanza, so deeply and characteristically dyed is the fabric. The rhyme pointed out here is to be distinguished from swan, woman of R 9, S I, Y 10.
9. C 18-I9, E 11-13, FP 4--5, M 15, N ii, O 10, Q 14-15, V 15, T W 8. Apparently this passage of incremental repetition has exerted some influence on a previous passage; compare, e. g., B 12 and 22, which both mention the girdle.
10. D I6, E 4, P I8, I 12, P 19, V 20, W 10.
11. An example of this in the English group is seen in R, which, as Child observes (I 14o, note on R a), probably borrowed its introduction from Child I A, I B, which are both English ballads. This contamination could not possibly have occurred in a Scottish text.
12. Furthermore, the ballads can be compared to the botanist's plants in another important regard, viz., we can gather more if the number is insufficient for our investigation. This opportunity has been unduly neglected, but it nevertheless exists. Hardly any writer has brought together all the material on a particular theme and if one directs one's efforts to finding new variants from a particular region, success is reasonably certain. And when the printed sources fail us, oral tradition is still at our disposal. Perhaps Kapten Dr. W. Liungman is the only person who has actually done just this thing. He found his materials for the study of the marchen "Prinsessan i jordkulan" (Aarne-Thompson 870) insufficient and deliberately set about collecting new texts from oral sources. He found more than a score of texts in this way and doubled the amount of material available for his study. See Liungman, En traditions studie over sagan om Prinsessan i /ordkulan (Aarnes 870), Gothenburg, 1925.
13. I omit U, the American version, for the moment.
14. Professor G. L. Kittredge kindly writes me as follows under date of April ii, 1928:
The fragment you mention [ i. e., S] is in Kinloch's hand on a single sheet of two pages - once apparently loose but now bound in with the other leaves of the MS. The MS. is a collection, partly in Kinloch's hand and partly in several other hands, but all gathered by Kinloch. This sheet is undoubtedly in Kinloch's handwriting. The fragment is on the recto. On the verso is another song (Scottish, credited to Thomson - i. e. the song-book) also in Kinloch's hand. There is no indication whatever of the source from which Kinloch derived the fragment. The sources of most of his orally derived ballads were Scottish women, but his MS. contains (in his own writing) a number of pieces of the broadside kind. There is no improbability in your conjecture that Kinloch got the fragment from an English source (perhaps oral, since it is a fragment). Indeed, I am inclined to think his source was English. The absence of Scottish dialect in the fragment would favor this conclusion.
15. Folksongs of the South, Cambridge, 1925, p. 20.
16. R, Y. The stanza is perhaps borrowed from Child I A, i B; see Child I I4o, note on Io R a. Note that the contamination is a contamination of English ballads. See above p. 241 n. r.
17. R 9, S I, Y 10.
18 R 13, S 5, Y 12.
19 L. Pound American Ballads and Songs No. 4, J. H. Cox Folksongs of the South No. 3. In the notes to these versions most, if not all, of the American ballads are listed.
20. Pound No. 4 B, Cox No. 3 A, JAFI XVIII (1905) 130, 131, XIX (1906) 233, XXX (1917) 286. A woman is found in Cox Nos. 3 B, 3 C, JAFI XIX (19o6) 234, XXX (1917) 288.
21 Cox Nos. 3 A, 3 C, t Pound No. 4 B, t JAF1 XVIII (1905) 130, 131, XIX (19o6) 233, 234, XXX (1917) 286, 288.
22 This is general in the American ballads and need scarcely be tabulated.
24. See Modern Philology XXIV (1927) 486ff.
25. See Belden JAFL XXX (1917) 287.
26. The informant said he was not sure whether it was hanged or hung.
27. It was sung by the grandmother of one of her former students, Fred H. Sidney.