1 Berger, Karol, ‘Die beiden Arten von Da-Capo-Arie in der Matthäus-Passion’, Bach-Jahrbuch 92 (2006), 127–159Google Scholar .

2 Whaples, Miriam, ‘Bach's Recapitulation Forms’, The Journal of Musicology 14/4 (1996), 475–513CrossRefGoogle Scholar . Whaples alludes in passing to the use of the term ‘modified da capo’ for several of Handel's arias (510, note 48).

3 Daniel E. Freeman, ‘J. S. Bach's “Concerto” Arias: A Study in the Amalgamation of Early Eighteenth-Century Genres’, unpublished paper cited by Whaples in ‘Bach's Recapitulation Forms’, 477, note 5; Freeman's essay subsequently appeared in Studi musicali 27/1 (1998), 123–162.

4 See Schulenberg, David, ‘The Sonate auf Concertenart: A Postmodern Invention?’, in Bach Perspectives 7: J. S. Bach's Concerted Ensemble Music, The Concerto, ed. Butler, Gregory (Urbana: University of Illinois Press, 2008), 57–60Google Scholar .

5 Stephen A. Crist, ‘Aria Forms in the Vocal Works of J. S. Bach, 1714–24’ (PhD dissertation, Brandeis University, 1988).

6 Dürr, Alfred, Studien über die frühen Kantaten Johann Sebastian Bachs (Leipzig: Breitkopf und Härtel, 1951), 140–144Google Scholar , used the term ‘pure’ (rein) in place of ‘strict’. These terms seem not to be used in Dürr, The Cantatas of J. S. Bach: With Their Librettos in German-English Parallel Text, revised and translated by Richard D. P. Jones (New York: Oxford University Press, 2005), in which the forms of individual arias are rarely analysed. On the other hand, the term ‘modified da capo aria’ is used by Boyd, Malcolm, Bach, third edition (New York: Oxford University Press, 2000), 142Google Scholar , and by Crist, Stephen A. in the article ‘Aria’ in Bach, Oxford Composer Companions series, ed. Boyd, Malcolm (New York: Oxford University Press, 1999), 17Google Scholar .

7 Whaples, ‘Bach's Recapitulation Forms’, 490–491, citing Dürr, Studien, 149; Crist, ‘Aria Forms’, 235ff. The concept of ‘parallel’ structure is also central to Joel Lester's view of Bach's instrumental movements as comprising two, three or more parallel sections; see ‘Heightening Levels of Activity and Bach's, J. S. Parallel-Section Constructions’, Journal of the American Musicological Society 54/1 (2001), 49–96Google Scholar .

8 Whaples, ‘Bach's Recapitulation Forms’, 491, finds this analogous to the later cavatina, that is, an aria comprising two statements of a single strophe, without a B section. Like myself, she distinguishes (493) between this type of aria (AA′) and the ‘binary’ (recte bipartite) variety in which a second strophe or other text segment occupies the second half of an aria, producing an AB design.

9 Opitz, Martin, Buch der deutschen Poeterey (Breslau, 1624)Google Scholar ; Gottsched, Johann Christoph, Versuch einer Critischen Dichtkunst für die Deutschen (Leipzig, 1730)Google Scholar . Bach is likely to have been at least aware of the latter work, whose author furnished librettos for at least three vocal works.

10 Einbau, translated variously as embedding and insertion, is, like ‘free’ da capo form, another term in need of critical re-examination.

11 More specifically, with schemes symbolized ‘a.bba’ and ‘aab.ccd’; see Crist, ‘Aria Forms’, 106ff. This conclusion holds only for works composed as part of Bach's first annual cycle (Jahrgang) of Leipzig cantatas (1723–1724).

12 ‘Bach's Recapitulation Forms’, 486–488.

13 This is especially true in early works, as with the arias bwv208/11 and 61/3.

14 As in the late aria bwv30a/3 (parodied in bwv30/3). bwv151/3, from 1725, is superficially similar. Yet although here, too, both A sections are genuinely binary, the initial statement of the A text concludes without a fully fledged cadence (that is, without root motion in the bass; see bars 26–27 and 95–96).

15 As in the arias bwv136/3, 24/1 and 167/1 from the first Leipzig Jahrgang of 1723–1724. The device described here is distinct from a so-called motto (Devise) entry, in which only the opening portion of the A text is sung, then repeated after a restatement of a part of the ritornello.

16 Whaples, ‘Bach's Recapitulation Forms’, 485.

17 Cone, Edward T., ‘On the Structure of Ich folge dir’, College Music Symposium 5 (1965), 78Google Scholar , and Boyd, Malcolm, Bach, first edition (London: Dent, 1983), 132Google Scholar ; both cited by Whaples, ‘Bach's Recapitulation Forms’, 475–476.

18 See Schulenberg, David, The Instrumental Music of Carl Philipp Emanuel Bach (Ann Arbor: UMI, 1984)Google Scholar , 100ff. Examples of sonata form in this sense are especially prominent in the ten preludes from Part 2 of Sebastian's Well-Tempered Clavier that contain a central double bar.

19 I make a similar suggestion in relation to Bach's, keyboard music in ‘Fugues, Form, and Fingering: Sonata Style in Bach's Preludes and Fugues’, in Variations on the Canon: Essays in Musical Interpretation from Bach to Boulez in Honor of Charles Rosen on His Eightieth Birthday, ed. Curry, Robert, Gable, David and Marshall, Robert (Rochester: University of Rochester Press, 2008), 12–21Google Scholar .

20 Such was presumably the case at Weimar, where Bach's principal librettist, Salomo Franck, was chief consistorial secretary, and the ruling duke took an active interest in matters of religion and music. At Leipzig, too, Bach's texts were subject to approval by the church hierarchy and city council for whom he worked.

21 Handel also changed the metre; the two early versions of ‘Rejoice’ represented in Table 2 are in 12/8 time. In its familiar form (version 3) the aria has the same design as version 2 but converts the metre to 4/4, with corresponding melodic variation.

22 See Winton Dean, Handel's Dramatic Oratorios and Masques (London: Oxford University Press, 1959), 452, for a facsimile of the page from the autograph of ‘Lament not thus’ (plate 7) showing Handel's ‘instructions for short-circuiting the da capo by inserting the second part in the middle of the first’. In fact what Handel did was more complex, inserting what are now bars 24–28a; in contrast to the type of modified da capo aria used by Bach, these conclude the first A section in the tonic.

23 Hurley, David Ross, Handel's Muse: Patterns of Creation in His Oratorios and Musical Dramas, 1743–1751 (New York: Oxford University Press, 2001)Google Scholar , discusses the issue in his Introduction (11) and in relation to modifications of forms and texts generally (174–176). He assumes, with Weimer, Eric, Opera Seria and the Evolution of Classical Style, 1755–1772 (Ann Arbor: UMI, 1984), 27–28Google Scholar , quoted by Hurley on page 175, that the reason for modifying the da capo aria in operas of 1720–1780 was to shorten the form.

24 The first is the dance song (gavotte) ‘Endless Pleasure’ with which Semele ends Act 2. This is a genuine, if miniature, through-composed da capo aria; it leads immediately to a chorus that repeats the text and musical material of the A section. Two successive arias sung by Jupiter and Semele in Act 3 likewise form part of a series of movements constituting their climactic scene together (neither character exits after singing these arias). Semele's ‘I ever am granting’ is especially short and simple, and both begin as continuo arias, with strings entering only later.

25 Whaples, ‘Bach's Recapitulation Forms’, 510–511, discusses ‘Rejoice’ and several other arias from late in Handel's career.

26 The quotation is from Smith, Julian, ‘Carlo Pallavicino’, Proceedings of the Royal Musical Association 96 (1969–1970), 67CrossRefGoogle Scholar , cited by Boyd, Bach, 142 (this and all subsequent references are to the previously cited third edition of Bach). Smith points to parallels with sonata form in these arias, but the parallels are limited to the use of similar material in the two outer sections of a single aria, and to the tonal design of the latter, which involves cadences in the dominant and the tonic respectively in the outer sections.

27 Boyd, Bach, 142.

28 D'Accone, Frank, The History of a Baroque Opera: Alessandro Scarlatti's ‘Gli equivoci nel sembiante’ (New York: Pendragon, 1985), 57Google Scholar , apparently unaware of the earlier references to them by Smith and Boyd, describes these as ‘arias in three-part form [that] have a third section that is a variation of the first’. Of close to seventy arias and duets in Scarlatti's Eraclea (Naples, 1700), all but one or two are in a regular da capo form; none is ‘modified’ in the present sense.

29 Number 86 in the edition by Frank D'Accone (Cambridge, MA: Harvard University Press, 1982). The only through-composed da capo form in the first act is a short duet (‘Tu fuggi’, No. 9), which is remarkable for the fact that its first A section ends with a Phrygian cadence on E, the second with one on B. There is likewise only one through-composed da capo aria in Act 2: the final number, ‘La speranza’, No. 74. It is, however, the real thing, complete with framing ritornellos. Less precisely in this form is No. 81, ‘Io son in gabbia’, whose closing section fails to repeat line 2; this is also a strophic aria.

30 Lesbo's aria ‘Allegrezza’ in Act 1 Scene 10.

31 The aria ‘Durch mächtige Kraft’, from the Mühlhausen council installation cantata of 1708 (bwv71), is earlier, and is in a sort of free da capo form, but its A section has a text of just three words. Two other early vocal movements by Bach in da capo form are settings of psalm verses and therefore are not arias, strictly speaking (bwv71/4, 196/3).

32 On the work's origin and text see Maul, Michael, ‘Vorwort’ to J. S. Bach: Aria. Alles mit Gott und nichts ohn' ihn (Kassel: Bärenreiter, 2005)Google Scholar .

33 The closing ritornello for strings is a dense contrapuntal working-out of thematic material related to the earlier portion of the movement; it possesses its own distinctive design, most of the first phrase being repeated at the dominant to yield a sort of through-composed bar form: AA′B.

34 For a contemporary example by Bach see the Corrente of Partita No. 6 (bwv830/3), entered by Bach around 1725 in the second Little Keyboard Book for Anna Magdalena Bach; the movement recurs in an early version of the G major sonata for violin and keyboard bwv1019.

35 As Whaples, ‘Bach's Recapitulation Forms’, 494, observes.

36 A number of additional movements, such as bwv152/2, might also be included in Table 5, but their irregularities are such that it is unclear whether they should be considered da capo forms at all, as opposed to designs that happen to conclude by repeating one or two text lines at the end.

37 Franck's text for this work was printed in his Evangelische Andachts-Opffer (preface dated Weimar, 1715). Whether the libretto as Bach received it appeared in exactly the same layout as the printed one is of course impossible to say; Whaples's suggestion (‘Bach's Recapitulation Forms’, 486) that printed layouts might have been influenced by Bach's musical settings strikes me as unlikely, at least for those texts printed in collections assembled by the poets themselves.

38 The same opening line, ‘Es ist vollbracht’, receives somewhat comparable treatment in bwv159/8, also on a text by Picander.

39 Equally systematic construction is rare in Bach's early choral fugues; for instance, in the fugue of the motet bwv228, two subjects are introduced simultaneously only for the fourth and fifth text segments, and a sixth text segment is set later as a countersubject that is heard only a few times.

40 Another bar form occurs in bwv31/2, composed about a year later, for the same reason.

41 Dürr, The Cantatas of J. S. Bach, 19. The solo version of this technique, Soloeinbau, had been a regular feature of Bach's Weimar arias (it is already present in the early bwv196/3, actually a psalm verse set in aria style).

42 As printed in Franck's Evangelische Sonn- und Fest-Tages Andachten (Weimar and Jena, 1717); although identical in poetic style and form to arias, these initial choral movements are not designated arias, the movements that follow being labelled ‘Aria 1’, ‘Aria 2’ and so on.

43 Recitatives were added when Bach adapted these cantatas for performance at Leipzig.

44 Musically the aria might not be considered a ‘true’ da capo form, as the reprise of lines 1–2, though alluding to the melodic material of the opening section, constitutes a mere tag attached to the preceding ‘B’ music without a substantial articulation. The texts as given here follow Franck's Evangelische Sonn- und Fest-Tages-Andachten; layout, punctuation and even words are altered in, for example, the texts given online by Walter F. Bischof at <http://www.cs.ualberta.ca/∼wfb/bach.html> (30 August 2010). The translations here are by Z. Philip Ambrose at <http://www.uvm.edu/∼classics/faculty/phil.html> (30 August 2010) and are used by permission.

45 But there is as yet no regular recapitulation (compare Table 4 for bwv167/1, above), and relatively little of the vocal part returns from the first section, the effect of recapitulation being created instead by the heavy use of Einbau in bars 30b–33 and 37b–39. Nevertheless, this appears to be the first instance in which Bach uses something like sonata form for a unitary text, that is, a text that does not fall into distinct stanzas; in two previous instances of a similar musical form the text is bipartite (bwv161/5) or through-composed (bwv155/4: eight lines, the last two forming a couplet).

46 For instance, Handel's Rinaldo ends with ‘Vinto è sol della virtù’, a gavotte, and Giulio Cesare ends with the bourrée (or perhaps a rigaudon) ‘Ritorni omai nel nostro core’, whose rhythm, including distinctive three-bar phrases, and rondeau form are precisely reflected in ‘Die Katze lässt das Mausen nicht’, the closing ‘chorus’ of the Coffee Cantata (in which all three singers participate). Handel is already writing dance arias in his Italian cantatas of 1707–1708, as when ‘Quanto più rigida si mostra Irene’ (a sarabande) concludes the cantata Irene, idolo mio (hwv120b) and ‘Altra spene or non alletta’ (a minuet) closes Menzognere speranze (hwv131, copyist's bill dated 22 September 1707). Bach, who would not have known those works, is closer than Handel to the French versions of the dances in terms of melodic and rhythmic style.

47 Boyd, Bach, 144. The same sort of recapitulation occurs also in instrumental works, for instance the final movement of the sonata in D major for keyboard and viola da gamba, bwv1029 (particularly relevant to bwv23/1 as both involve two principal melodic parts), the opening adagios of the violin sonatas bwv1001 and 1003, and a number of preludes from the Well-Tempered Clavier: in E major from Part 1 (bwv854/1), and in C major, A flat major and B flat minor from Part 2 (bwv870/1, 886/1, and 891/1; the recapitulation is absent from the earliest version of the C major prelude, bwv870a). Instrumental dances are usually in rounded binary rather than sonata form, lacking a fully fledged recapitulation section that begins with a return to the opening bar; a rare subdominant return in a dance occurs in the Allemande of the D major partita (bwv828/2).

48 See, for example, the first movement of the Sonata in B major, d575, or the finale of the Sonata in A major d664.

49 Compare bwv91/3, 30/5 (notable also for its gavotte style and ritornello in binary form), 245/24, and 244/49 and 57. In none of these examples is the recapitulation as literal as in bwv23/1. For instance, in bwv245/24 (‘Eilt’) the return is elided, the restatement of material from the A section beginning at bar 133; the latter corresponds to the seventh bar of the initial vocal passage (bar 23, now transposed up a fourth).

50 Two bar-form arias from Weimar (bwv182/6 and 31/2) have been previously noted. In the opening movement of the St Matthew Passion, lines 1–3 of the aria text combine at bar 30 with lines 1–2 of the chorale, constituting the first statement of the Stollen. Line 4 of the aria, containing musical as well as textual parallels to lines 2 and 3, is then combined at bar 44 with lines 3–4 of the chorale, representing the repetition of the Stollen, which is reharmonized to conclude in the relative major. The recapitulation begins in the subdominant at bar 72, setting line 7 of the aria text to music previously used for line 1 (at the opening of the vocal portion of the movement, bar 17). Following the same principle as the duet in bwv23, this subdominant return reverses the soprano and alto parts, which enter in imitation at the fifth. But a textual reprise begins only later, when line 1 is restated in Choreinbau at bar 82 to conclude the movement.

51 ‘Ach, nun ist mein Jesus hin’ (bwv244/30) repeats a version of its opening vocal passage in the subdominant at bar 78. To be sure, the text at this point is not exactly the same as at the opening, and, as both new text and new music follow, this is hardly a recapitulation in the same sense as is the A′ section of a through-composed da capo aria. The movement concludes with a more literal recapitulation of the same music, now in the tonic, at bar 108, set in Einbau – a type of ending that Bach otherwise used only in choral movements, occasioned here by the unique dialogue text, which ends with a question for the soloist.

52 This is the first instance in which strings imitate this motive at the interval of a single bar, hence constituting a stretto that literally embodies the text (stretto = hurry).

53 Wolff, Christoph, ‘The ‘Agnus Dei’ of Bach's B-minor Mass: Parody and New Composition Reconciled’, in A Bach Tribute: Essays in Honor of William H. Scheide, ed. Brainard, Paul and Robinson, Ray (Kassel: Bärenreiter, 1993), 236Google Scholar . The essay is also printed in Wolff, Christoph, Bach: Essays On His Life and Music (Cambridge, MA: Harvard University Press, 1991)Google Scholar ; see 335.

54 This seems to be how Wolff, ‘The “Agnus Dei” of Bach's B-minor Mass’, 237–238, understands the movement, seeing in it a ‘new and asymmetric formal design’ as compared to its parody model.

55 The bass of bars 37b–39a is a variation of that in the last two bars of the ritornello; this implies that the upper parts of the latter were also restated here in some form.

56 This restatement of the ritornello is displaced by half a bar; the downbeat of bar 1 corresponds with the middle of bar 86.

57 The instrumental parts of the early version, bwv210a/6, are lost, but the vocal part is preserved.

58 At this point in the early version (bars 37–41), line 4 is set relatively simply; the vocal part is lightly embellished in the later version. Subsequently, for the restatement of lines 2–3 (bars 41–44), the parody substituted different motivic material, reflecting the more active verb ‘eilt’ in line 3 of the parody text.

59 On the weakening of baroque musical rhetoric in Friedemann Bach's sacred works see chapter 5 in my The Music of Wilhelm Friedemann Bach (Rochester: University of Rochester Press, 2010).