Bartók ’ s Sonatas for Violin and Piano nos. 1 – 2: The Compositional Process

In the focus of this paper a survey of the draft score will disclose major corrections of the concept and discuss deleted and rewritten sections in both Sonatas for Violin and Piano no. 1 (1921) and no. 2 (1922). A close study of the unusual-type preliminary sketches of the First Sonata in his so-called Black Pocket-Book (facsimile edition: 1987) already gave insight into Bartók ’ s atypical composition when he had to work without a piano at hand for shaping and re ﬁ ning a new major work (see Somfai, “‘ Written between the Desk and the Piano ’ : Dating Béla Bartók ’ s Sketches, ” in A Handbook to Twentieth-Century Musical Sketches , ed. by Patricia Hall and Friedemann Sallis, Cambridge: Cambridge University Press, 2004). The two draft scores (no. 1 5 34 pages, no. 2 5 21 pages, including discarded and rewritten sections) open new vistas in understanding the concept of the individual compositions. The next stage of manuscripts provides a signi ﬁ cant source: the score and violin part used at the ﬁ rst performances, the latter with ﬁ ngering and bowing contributed by the hand of Jelly Arányi and Imre Waldbauer in the First Sonata, Waldbauer, Ede Zathureczky, Zoltán Székely, and Jelly Arányi in the Second. A study of the revision of metronome numbers will conclude the investigation.


THE SOURCES OF THE TWO SONATAS
The source material of the two Sonatas for Violin and Piano (BB 84,85) is almost complete: except the corrected proofs of the Universal Edition (UE) first edition nothing seems to be lost or destroyed. A summary (Fig. 1) 1 displays the complex nature of the manuscript scores and violin parts that were used for the first performances. 2 The First Sonata (1921) has a longer chain of sources, because for the Vienna world premiere played for the initiative by composer-violinist Mary Dickenson-Auner and Eduard Steuermannan extra score and part were needed, while Bartók had his own set prepared for the England-France-Germany tour. The significant first public performances played from manuscript are presented in Table 1. apparently by Jelly Arányi, 6 markings in blue pencil are by Imre Waldbauer, but further markings also appear by a third, unidentified hand. In the violin part D2 no performer's addition can be found; Bartók used this copy for collecting and marking his final revisions and adding the socalled "Stichnoten" (cue notes), 7 before the manuscript went to the engraver.
Apart from a one-page autograph sample copy of the beginning for facsimile reproduction (C) that shows the slightly edited draft version before Bartók finalized it (see below), the Second Sonata has a simpler chain of sources. Interestingly, Bartók himself copied the violin part from the still intermediary version of the draft score. 8 No other manuscript copy of the violin part is known (besides, it is established that Bartók sent the printer's copy to Universal Edition only in late May 1923), thus Imre Waldbauer, Ede Zathureczky, Zoltán Székely, and Jelly Arányi had to play from this copy in February-May 1923 at the early concert performances as shown in Table 2. Both Sonatas appeared in print in 1923, the First in July, the Second in August.

A DIGRESSION: THE MOTIVATION AND THE DATE OF WRITING THE FIRST SONATA
The intensive period of writing the draft score of the First Sonata is well documented: Bartók dedicated and dated the Sonata after the last bar in the printed edition as " A.J.-nek, Budapest, 1921. X.-XII." [To J[elly] A [rányi]. Budapest, October-December 1921]. He even noted down the precise date of the completion of two movements in the draft score, which was not his habit: movement I was finished on (1921. okt. 26.) and movement III on (1921. dec.12.). Dating the "composition"the creation from the notation of the first ideas till the completion of the draft scoreis, however, a delicate matter. No correction of the technically difficult passages in movement I occur, see Bartók's letters to Jelly Arányi: 7 December 1921, "I am afraid that the first movement may rather frighten you. . . . Should the 4-string arpeggios prove to be too difficult I would certainly alter them in some way"; 2 February 1922, "Do not worry about the arpeggios: as I wrote before, I would be willing if absolutely necessary to make them easier . "Stichnoten" 5 cue notes (small-sized notation to help the correct performance in the violin part referring to the piano after a longer rest in the violin part or in rhythmically difficult sections). 8 Therefore, there is a warning in the draft score (p. 6) for Márta Ziegler who copied the score E but had to care about changes realized in D: "lásd a hegedű szólamot" [see the violin part].
Concerning the time of the beginning of the composition the existing documents are contradictory. On her birthday, 19 October 1921, the composer's wife Márta Ziegler wrote to Bartók's mother that, as a birthday present, Béla showed her the Violin Sonata on which he was working: "I'm so grateful to Jelly Arányi whose wonderful playing on the violin has caused this (as he says) long-dormant ["régen szunnyadó"] plan to spring out of Béla." 9 A physical evidence of the long-dormant plan seems to be the six-page sketch complex in the Black Pocket-Book. 10 As far as we know, Bartók used this pocket-sized sketchbook only when he was away from his home and piano; when he could not work quietly in perfect isolation improvising at his instrument before he sat at his desk to put on paper the already crystallized music. The latest of such a possible date was around 1-20 August 1921 when Bartók spent his holiday in Austria. 11 But the sketches, starting on fol. 24v directly after the sketches to The Miraculous Mandarin (BB 82), were probably written earlier, during his 10 July to 6 September 1920 summer holidays in his sister's house in Kertmeg puszta, Hungary. 12 In these sketches, written in several steps, Bartók outlined the opening of movements I, III, and II (in this sequence), altogether the rough form of 78 measures of movement I, 41 measures of movement II, and 77 measures of movement III, with some follow-up passages. 13 Conflicting data are provided in Bartók's 9 November 1921 Hungarian letter to Jelly Arányi, from which only a partial English translation is known: Movements 1 & 2 of your violin sonata are already available, completely finished, even 1/3 of the third. . . . Your violin playing has indeed impressed me so much that I decided on that Tuesday [4 October] when we last played together: I will attempt this, for me, unusual combination only if both instruments always had separate themesthis notion has taken definite form so that already the next day the plan & the main themes for all three movements were ready. I could have mentioned it when we said goodbye on Thursday [6 October]. I wanted very much to do so, but I did not dare, I did not know whether after such a long silence -2 yearswhich had been imposed on me I could still compose. Very soon it appeared that I was still able toand how! [. . .] I compose this sonata for you -& I [will] dedicate it to you (if you accept). 14 The revelation of how "the next day" he outlined the three-movement plan and the main themes, although Márta heard about a "long-dormant plan" from her husband, or that this (i.e. 9 Családi levelei, 325. 10 Bartók Béla fekete zsebkönyve. Vázlatok 1907Vázlatok -1922. Az eredeti kézirat fakszimile kiadása Somfai László utószavával / Black Pocket-Book (Sketches 1907-1922 writing for violin and piano) was an "unusual combination" for him, 15 or the statement that in the last two years he could not compose (although he wrote Improvisations on Hungarian Peasant Songs, op. 20, BB 83 in 1920) are romantic exaggerations of this almost love letter to Jelly.
Thus, it is a fact that extensive sketches to a violin and piano sonata had to go back at least to summer 1921 or more probably to 1920, before the composer became familiar with Jelly Arányi's mature violin playing in Budapest between late September and 4 October 1921. 16 It is worth noting that before this meeting, on 24 September 1921 Bartók was still uncertain whether playing with Jelly Arányi in England would be favorable for him. 17 Only after they had met in Budapest changed his opinion radically: as he wrote to Michael D. Calvocoressi, "she is an excellent violinist," so that he "would very much like to play" the new sonata "with her for the first time" (20 October 1921 letter to Calvocoressi). 18 The motivation of writing a three-movement sonata, as the sketches prove, preceded the appearance of Jelly Arányi in Budapest. The First Sonata was conceived in the wake of Bartók's isolation during the First World War. In the early spring of 1920, he managed to travel from Budapest to Berlin where he gave recitals 19 and gathered fresh impressions of the latest musical trends. He already knew that at a possible recital tour in England or a visit to Paris he would meet an audience familiar with the new music of Debussy, Ravel, Stravinsky, Schoenberg, the young French composers, and the other modernists. What kind of a major work could he offer to such an audience beyond his solo piano pieces? We assume that he could not hope for playing a Bartók concerto with major orchestras (besides, he only had the old Rhapsody, op. 1, BB 36b in this genre), but with his own participation he could offer a large-scale new work in the genre of chamber music, a duo sonata. As regards Bartók's acquaintance with the new output of chamber music, he recently played the piano part of Kodály's Sonata for Cello and Piano and Ravel's Trio in 1920 in Berlin, the latter and Debussy's Violin Sonata on 23 April 1921 also in Budapest (here in addition he played inspiring modern solo piano works as nos. 1-2 from Schoenberg's Drei 15 In fact, Bartók wrote several pieces for this combination: two juvenile works, a C-minor Sonata in 1895 Zongora és hegedű számára [for piano and violin], op. 5, BB 6 (not published); an A-major Sonate für Klavier und Violine, op. 17, in 1895, BB 10 (not published); a three-movement Sonate pour piano et violon in 1903 BB 28 (performed but not published in Bartók's lifetime). NB.: The sequence of the name of the two instruments in these titles ("for piano and violin") was probably influenced by the typical titles in the printed editions of the genre in Bartók's youth, e.g., the Simrock edition of Brahms's every Sonate für Pianoforte und Violine. Occasionally Bartók even mentioned his First Sonata as a sonata "for piano and violin," 29 December 1921, French letter to Cecil Gray, see Bartók Béla levelei Klavierstücke, and Stravinsky's Piano Rag Music), but with a new violin and piano sonata he apparently sought to write something more monumental.
Contrary to the traditions of the genre, as already the sketches demonstrate, Bartók was considering a kind of duet in which the violin, unsurpassable in playing the melody, and the piano, ideal in playing chordal structures with two hands even in contrary motion, were given markedly different themes; they did not play each other's very idiomatic themes. 20 Thus, the decisive components of the concept and the basic thematic material of movement I plus the opening theme area in movements II and III were crystallized before Bartók discovered Jelly Arányi's magnificent violin playing. Undoubtedly in October 1921 he received significant inspiration for rendering the planned composition in final form when in Budapest he met the Arányi sisters, whom he knew from his years at the Academy of Music, 21 who this time returned from London to visit Hungary. The now married Adila (Adila Fachiri) and her younger sister, the 28-year-old Jelly, had become brilliantly accomplished violinists since they last met. In addition, Jelly enchanted Bartók with her aura and femininity, too. As to new repertoire, among others Jelly acquainted him with Szymanowski's Trois Mythes and Notturno and Tarantella that 20 A striking exception is that the most characteristic piano theme of movement I, the Vivo, appassionato theme at fig. 7 (5 m. 66ff), Bartók originally noted down at the top of fol. 25r of the Black Pocket-Book clearly as a violin theme with double stops: Incidentally, this theme is closely related to the ff marcatissimo contrast theme of the Allegro molto first Study of Bartók, though more in gesture and rhythm than in its chordal structure: Since it is documented that Bartók read the proofs of the Studies for piano during his 1920 summer holidays in Kertmeg pusztaaccording to his 1 September 1920 unpublished letter to Universal Edition (preserved in Peter Bartók's collection, deposited in the Paul Sacher Stiftung, Basel, photocopy in the Budapest Bartók Archives) -, on that day the composer sent the corrected proofs back), we may assume that the idea of the violin double-stop theme came to his mind in August 1920. Such "self-inspirations" are known in Bartók's music, e.g., as the Trio theme in movement III of Contrasts (measures 134ff) came to his mind when working on the instrumentation of the last measures of movement II of the (Second) Violin Concerto, see SOMFAI, Béla Bartók, 60-61. 21 In addition to a little Duo for two violins, in 1902 Bartók dedicated a 53-measure long Andante for violin and piano (DD 70, BB 26b) to Adila. featured a range of interesting novelties in violin playing, 22 some of which presumably inspired Bartók in elaborating special effects in the violin part of his new Violin Sonata.

THE APPEARENCE AND FUNCTION OF MANUSCRIPTS: SKETCH
The source material of the two Sonatas is atypical among Bartók's compositions for at least two reasons. First, the sketches, which outline the basic characteristics of the individual movements, were written in an unusual way: instead of working on the movements one after the other, Bartók put down opening ideas for different movements (in case of the First Sonata not even in the final sequence of the movements) on subsequent blank pages and he worked parallel on the continuation of each. Secondly, at the finalization of the violin part the composer let his violinist partners suggest changes in the length of slurs 23 which he incorporated in the printed score; furthermore Bartók let a selection of the suggested fingering be printed in the published violin part, thus the violin in the printed score and in the separately printed violin part are not identical. 24 For understanding the working process on the sketches of his two Violin Sonatas, the first two pages from the First Sonata offer a good example (see Examples 1a-1b, a diplomatic transcription).
Based on a close look at the notationtaking into consideration not only the size and position of notes but also the color and thickness of Bartók's fountain pen, 25 and studying the physical evidences (color, thickness) with the original manuscript in hand, in daylight, repeatedlythe most probable sequence of writing acts is this: -B1-B2: On the top of the still blank fol. 25r Bartók wrote down a violin theme, which he later turned into a piano theme of movement I (see footnote 20). -A1: In two staves, following the end of The Miraculous Mandarin sketches on fol. 24v, he wrote down four bars plus one note, the beginning of movement I. -C1 and C1 cont.: Returning to 25r, Bartók outlined the beginning of movement III (the piano in C sharp just as in the beginning of movement I).
-The sequence of the continuation is uncertain: probably A2 and A2 cont., with significant changes in the violin and piano part alike, followed by the writing act of at least part of C2,

22
Bartók played Szymanowski's Trois Mythes with Jelly privately and with Zoltán Székely on his 12 November 1921 concert in Budapest publicly. As to the characteristic violin techniques that Bartók supposedly borrowed from Szymanowski (the special ondeggiando in movement I from figs. 10, 11, and 12, furthermore special techniques in movement III), see Malcolm GILLIES, "Stylistic Integrity and Influence in Bartók's Works: The Case of Szymanowski," International Journal of Musicology 1 (1992), 139-160), they are not present among the sketches.

23
A legato slur in Bartók's notation for string parts directly indicates the length of one bow stroke. Corrections of slurs in the successive notations of the same string part (including the parts used at rehearsals and corrected proofs of the printed score) indicate that Bartók considered the length of his legato slurs as bow strokes, perfectly fitting to the intended tempo and style. 24 Useful suggestions by Bartók's trusted performers in the string quartets were included in the printed score; in this genre score and parts printed the same text. 25 See details in SOMFAI, "Dating Bartók's Sketches," 123ff.
Examples 1a-1b. The transcription of sketches on fol. 24v and 25r in the Black Pocket-Book Compared to the mosaic-like work in the sketch book, the notation of the draft score is a much more quiet and continuous writing act (see Example 2, the first page of movement I of the First Sonata). After all if sketches existed, as in this case, then writing the draft score was more or less copying, producing a revised version as it were; on the other hand, if there were no sketches, then the draft written at the desk appears in the quiet notation of longer coherent sections which already crystallized during Bartók's improvisation at the piano. 26 Even if there are crossed and rewritten measures or insertions on the margin or between the 1þ2-stave systems, his draft score shows a neat notation. He left an empty stave between the three-stave systems for eventual corrections (except the last two systems at the bottom of the page), gave the clef to each staff, often scratched discarded notes and indicated the full rest in empty measures. In this composition even the slurs are precisely marked for both instruments. Not that Bartók was a pedantic man. Writing the draft in an easy-to-read notation had a purpose: in those years under normal conditions his wife Márta Ziegler made a clean copy of the draft which in a next step Bartók furnished with the missing performing instructions.
With the exception of the last five measures, the sketches to this page of the draft can be found on fol. 24v of the Black Pocket-Book (see Example 1a). Bartók practically preserved all essential features, notably the violin part, up to the minutest details; only slurs had to be added. As to the piano part, while he worked with the same pitcheskeeping the C-sharp arpeggiation against the C-natural central note of the violinhe elaborated it with much more impressive passages. Incidentally, it is conceivable that the A3 corrections in the sketch-book date as late as the time when Bartók began to write the draft score.
The appearance and function of Márta Ziegler's copyin this special case two copies, one for the Vienna premiere and another for Bartók's forthcoming England tourcan also be elucidated by the first page of the First Sonata: on the left page from the score D1 (Example 3), on the right page from the score C1 (Example 4).
In both copies the ground layer, strictly following Bartók's draft score, is written by Márta Ziegler. The additional layers of the two scores were likely written in the supposed sequence: In the score D1 Bartók added the tempo Allegro appassionato with ♩ 5 72 to the violin, furthermore inserted dynamic markings and the pedal instruction; then Márta Ziegler copied his tempo plus MM markings to the piano part (the legato slur in measure 3 of the violin is a later addition).
In the score C1 Márta Ziegler copied Bartók's additional performing markings from D1; the tempo to the piano part is in the composer's hand (Bartók played from this score), and he also signed it with the originally planned opus number: "Béla Bartók (op. 21.)." In the score D1which he got back after the Vienna premiere and designated as the printer's copy -Bartók inserted later decided performing instructions as, e.g., on this page, the addition to the MM: ♩ 5 72-80, or from Jelly Arányi's additions the legato slur in measure 3 (about this see below); on the top of the page Bartók gave instructions about the title, size of printing, etc. In a next step the editors of Universal Edition, primarily the much-feared Josef Venantius von Wöss (see his signature on the right margin) listed the features of notation which would be changed.

DISCARDED AND REWRITTEN PAGES AND PASSAGES
At the time of writing the two Violin Sonatas Bartók's habit of keeping all pages of the draft notation of a new composition, even the discarded pages, already became consolidated. 27 Thanks to this routine, to the preserving of the discarded/rewritten pages, a few critical points of the score are highlighted in both Sonatas: at these points he hesitated, or he was not satisfied, or for some reason he had to stop. Naturally, we can never be sure that we understand the composer's problem if he had a "problem" at all or he was just exhausted in the concentrated work and had to put the draft aside.
In the First Sonata there are three discarded units of the draft score. One is a typical case of interrupted work: there are two measures on the top of a page (p. 33) 28 to movement II, fig. 6 þ2 -3.29 Under it Bartók hastily listed corrections to the one-act opera Duke Bluebeard's Castle (BB 62), so the bottom part of the page had to be cut and sent (Bartók started the continuation of the Sonata on a full page on p. 13). The two other discarded sections are longer and belong to movement III.
From the top to the middle of p. 34 there is a 34-measure-long passage with the preliminary form of the music at fig. 28-30 þ7 plus 6 measures (see Example 5). It is obvious from the notation and the number of corrections that here Bartók gradually became uncertain. On the one hand he revised the original violin part in the first and second braces (figs. 29 -4 -30 -5 ) by replacing the too many quintuplet 16th-note figures with other motives. More importantly, from the middle of the last brace ( fig. 30 þ3 )  The draft of the Second Sonata has only one discarded-rewritten section, but a threepage long one 32 from fig. 18 -3 to fig. 29, the first half of the development section of the sonata-form movement II featuring one of the most impressive parts of the Sonata: the "a bit drunk" episode and a "bagpipe imitation" scene. 33 It is worth noting that Bartók had sketches to all themes from fig. 21 þ7 to fig. 29 on fol. 28r of the Black Pocket-Book, buta rare phenomenon in his compositional routinethe themes in the sketch and the draft appear in different keys or with different central tones: at fig. 22 the sketch is in F-sharp, the draft is in B-flat; at fig. 27 þ6 the sketch is in E, the draft is in A. As to the discarded first version of the draft, a minute description of these pages with analytical notes on why Bartók abandoned it would require a full facsimile reproduction with an extensive study, which is planned to appear in the "Description of Sources" in the critical edition of the two Sonatas. The fact is that around the end of August 1922 Bartók worked on this part of the Sonata very intensively in ideal isolation, 34 yet he hesitated about several points of the draft notation. A physical sign of it is a relatively high ratio of writing in pencil (the violin part between figs. 19 and 20; the piano in figs. 25 þ1 -27 -2 , also the violin in figs. 26 -2 -26), furthermore that Bartók left longer sections unfilled (in the "bit drunk" episode 22 measures of the piano from ca. fig. 22 þ1 ).
From the many dozens of rewritten passages in the draft of both Sonatas a prominent case is the formation of the retransition in the sonata-form movement II of the Second Sonata (see Example 8, from fig. 33 þ2 to fig. 36 þ4 ). A close study of the revisions sheds light on Bartók's inspired work on good and even better constituents of a key scene: as a transcendent reference to the first violin theme of movement I of the Sonata suddenly interrupts the playful development section, then a vibrato, quasi espressivo pizzicato passage stirs up the expectation before the stormy meter Vivo recapitulation. Each correction moves from conventional to extreme, from regular to sophisticate. Before the end of the upper brace the replacement of high-octave violin melody with harmonic is an obvious improvement. Fascinating is Bartók's revision of the pizzicato passage in the first four measures of the middle brace (see Example 7): before crossing and scratching it was a perfect original (O) and inverted (I) form of the same Example 6. First Sonata, correction of the violin part in fig. 47 þ2-4   32 In fact, 2 1/2 pages, the last one is only the upper half of a page (pages 13-15, librarian's pagination). 33 See SOMFAI, Béla Bartók, 71-74. 34 Bartók's wife visited her relatives in Transylvania, he sent his son Béla, Jr., to his sister in Szőllős puszta; on the other hand, he had visitors who interrupted the concentrated work (see Családi levelei, 334-337), but the discarding of three pages of the draft had in all probability inner motivation. seven 8th-note motive, but instead of this mechanical inversion, partly on the margin of the upper brace, he changed it to two hesitant gestures. Also, an improvement is the deletion of three measures before the Vivo, because the gradual widening of the quarter notes, on second thought, seemed to be mechanical.

DO SKETCHES AND THE DRAFT REVEAL ESSENTIAL FEATURES OF A HIDDEN NARRATIVE OF THE SCORE?
Although Bartók never used the term "narrative" in his texts on his instrumental music, occasionally he opened up and revealed his intention 35 as, e.g., in case of the Dance Suite (BB 86). 36 But he did not explain the concept of his two Sonatas for Violin and Piano.
To start with the simpler case, the sketches of the Second Sonata suggest that from the beginning the germ of the concept crystallized. On the top of two opposite pages of the Black Pocket-Book (fol. 27v-28r) Bartók outlined an improvisation-like slow theme vs. a dance-like theme based on the same pitch collection (the so-called "heptatonia secunda," frequent in Romanian folk music, with polymodal coloration). Although the attacca two-movement (slow-fast) concept of the form may point to the Lassú and Friss of a Hungarian rhapsody, Bartók in all probability picked up a Romanian peasant music phenomenon, the only programmatic piece in folk music (his word) that he knew, the two-part "When the shepherd lost his sheep," and "When the shepherd found his [lost] sheep." 37 The opening theme of the first movement is closely related to the hora lung a (the long melody), one of the folk-music phenomena which fascinated Bartók. The narrative of the "evolution" of folk music, as it were, from the hora lung a-like open form (the beginning of movement I) via motivic organization (the end of movement I) to a quasi-stanza form (before the end of movement II), is not revealed in the sketches but is present in the draft score.
The three-movement concept of the First Sonata is simpler; nevertheless, it also has a typical Bartókian message: the emphatically modern-music-style expressive first movement is opposed to a grand-style finale based on inspiration from rural music from Bartók's personal collections. Without actual quotations, here à la folk-dance themes appear in the style of Romanian and Ruthenian virtuoso fiddle music and, furthermore, Arabic dances inspired by his field work in Algeriasignificantly without any Hungarian character at all. 38 In the sketch-book just the opening theme of the finale was outlined, but in the draft Bartók had not much hesitation in fixing the folk-dance inspired sections one after the other; problems occurred mostly with the motivic elaboration in the middle part of the sonata-rondo form.
The eminently Hungarian character in the First Sonata is spared for the Adagio middle movement and here the compositional sources offer a rare insight into Bartók's workshop. The opening 50 measures of the ABA var form at the beginning of the movement present each of the two instruments on its own, the violin with a melody, the piano with a chordal progression. 39 The Black Pocket-Book sketches outline the first A part but contain nothing from the continuation. In the draft score Bartók had a simpler idea about the return of the A section than the one in the printed score (see the facsimile in Example 9 and the reconstructed layers of the changes in Example 10). 40 Instead of exposing the two instruments one after the other, here they Example 9. First Sonata, mov. II, the beginning of the recapitulation in the draft score (PB 51VPS1, p. 14) 41

39
In the slow second movement of his 1903 Sonata (BB 28) Bartók already applied the solo presentation of the theme first by the violin, then the piano, although both instruments played the same theme. 40 The senza sordino belongs to the ground layer: in the score and violin part from which Bartók and Jelly Arányi rehearsed the Sonata (C1 and C2) from fig. 4 the Sostenuto middle part originally was con sordino that Bartók later scratched out. 41 The Hungarian words "két taktust üresen hagyni" [leave space for two measures] instructed Márta Ziegler; in fact, in the copy D1 (printer's copy score) she left a brace empty which Bartók filled with the music from fig. 10 -2 . would play together, the violin performing its original melody in , the piano playing an accompanying chord. But Bartók stopped, hesitated, and arrived at the ideal solution of the problem: themes of the abstract-sublime exposition should return in a richly embellished Hungarian version.

FINALIZATION OF THE VIOLIN PART IN THE SCORE
It is no exaggeration to say that Bartók never let any string player partner inspire the final elaboration of the string part in his new composition as significantly as in the case of the First Sonata. Although Stefi Geyer's characteristic technique and personality considerably influenced Bartók in the composition of the early Violin Concerto (BB 48a, 1907(BB 48a, -1908, she did not play it, thus could not add her personal technical preferences to the score. Imre Waldbauer with his partners invited the composer to rehearsals and excellently realized the scores of the first two string quartets for the 1910 and 1918 premieres, respectively, but would not dare to suggest changes or improvements. Jelly Arányi, however, obviously had Bartók's encouragement to bring out the best from her part during the rehearsals even if only slight changes of bowing were neededmostly by adding slurs or making shorter slurs from long ones. The revision of bowing in the 18-measure-long solo entry of the violin in movement II of the First Sonata is remarkable. Not only were the majority of long legato slurs divided into shorter ones (Example 15), but Bartók even rephrased the end of the first melodic gesture in measure 3: from A 2 the downwards phrase does not end with C 2 but on A 1 .
A typical page of the violin part of the concert copy (C2) includes inscriptions not only by Jelly Arányi, but also by other hands as, e.g., in the reproduced section from a page in movement III (Example 16). Here above the first stave tempo and the sings are in Imre Waldbauer's hand in pencil; the word "hegy" [i.e., head (of bow)] and her typical └┘ down-bow sign in Jelly Arányi's hand; in the last measure the upper grey upbow in her hand, the lower blue one in Example 14. First Sonata, the concert copy of the violin part C2 (BBA, BAN: 1987, p. 4) with Jelly Arányi's additions of bowing and fingering (poco cresc. in Bartók's hand) Example 15. First Sonata, the printer's copy of the violin part D2 (PB 51VPFC2, p. 17) after Bartók scratched and corrected slurs, added separation signs, fingering, and string instruction (sul Re), furthermore added harmonics to the first note, and revised the metronome his hand. Waldbauer added fingering in blue pencil between figs. 9 and 10; an unidentified third hand entered the fingering in indelible pencil in fig. 10 þ2 . Bartók was critical when he selected and finalized the additional performance instructions (fingering, additional accents, and bowing) in the printer's copy (D2) from which Universal Edition produced the printed version (Example 17).
In the Second Sonata Jelly Arányi's contribution to the final elaboration of the violin part also appears but is not as vital as in the First Sonata. In this case Bartók prepared the premiere with Imre Waldbauer, but first he rehearsed the new work with Zoltán Székely because Waldbauer was ill. 42 Besides, while writing this score he made good use of the experiences of the Example 16. First Sonata, concert copy of the violin part C2, excerpt from mov. III (BBA, BAN: 1987, p. 12stm First Sonata. In the simpler source situationonly one score and one extra violin part had to be copiedthere is one surprise, though: It was not Márta Ziegler who copied the violin part from the autograph draft score, as was their routine, but Bartók himself produced an easy-to-read Example 17. First Sonata, Universal Edition print of the violin part with different additions of fingering and bowing copy (D). 43 The obvious explanation is that his wife Márta was in Transylvania when the draft score was more or less outlined and Székely came to Bartók's home to play the just finished work, thus a clean copy of the violin part was quickly needed. At the same time, due to the extremely complex rhythm and the entirely different rhythmic action of the piano and the violin parts, so many "Stichnoten" (cue notes) had to be added for the violinist that Bartók would perhaps have better produced it himself anyway (see Example 18). 44 One of the major changes obviously suggested by a violinist is a special bowing in the cadenza between figs. 55 -2 and 55 þ8 . In the draft score originally there was no cadenza for the violin. From Più vivo at fig. 48 to the triumphant return of the theme at fig. 56 only 59 measures were written down which Bartók gradually enlarged to 88 measures, including the poco a poco più vivace e stretto passage in which the two-note slurs appear (Example 19). In the autograph violin part (Example 20), just as in the copy of the score, following the advice of his violinist partner, Bartók corrected these two-note slurs of the ground notation to four-note or even sixor eight-note slurs. As a matter of fact, the instruction of the three-measure repetition of fig. 56 -5 ("N.B. 3mal stechen" [print 3 times]) was also part of the revision. Since Jelly Arányi's already mentioned └┘down-bow sign at fig. 55 þ9 of the cadenza is present in the autograph copy of the violin part, there is reason to suspect that this virtuoso bowing does not date from the first performances of the Second Sonata but only from the rehearsals with Arányi.
Example 19. Second Sonata, part of the cadenza in the draft score B (PB 52VPS1, p. 20) 43 Later additions from the time of preparing this manuscript as the printer's copy of the Universal Edition first edition, like e.g., tempo markings from a tempo to measure 5, are in Márta Ziegler's hand. 44 Prior to this autograph violin part Bartók produced the one-page sample copy of the first 14 measures of the score for facsimile reproduction (C). Minor but characteristic differences between C and this facsimile violin part: Moderato vs.  or got slower; Bartók basically increased the tempo, sometimes even changed the rhythmic value. In the two other movements the trend was similar; for instance in movement II he corrected the Adagio from ♩ 5 72 to ♩ 5 72-80, the Poco agitato from ♪ 5 66-70 to ♪ 80-88; in movement III practically all tempos became faster: the Allegro molto from ♩ 5 120 to ♩ 5 144-138, the last Vivacissimo from ♩ 5 144 to ♩ 5 168.
In the case of the Second Sonata there is only one set of MM numbers. Bartók added the Italian tempo markings and MM numbers either in the autograph violin part D or the copy of the score E, first in pencil, then in ink. (Márta Ziegler copied them in the other manuscript.) There is no indication that Waldbauer and Székely or Arányi cooperated in the finalization of tempos. But Bartók's own performance demonstrates that later, presumably also depending on the violinist partner, he played several tempos considerably differently. According to the   1920-1945(Budapest: Hungaroton, 1991, HCD 12326-31, CR 5/7.