12. Recorded history

A question to which I will periodically return in this blog is “What sorts of non-textual musical evidence nevertheless bear upon the text?”  Another way of thinking of it is “Beyond the notated sources, what other sources can/should affect a new edition of a work?”  A fairly obvious non-notated source is an recording involving the composer as interpreter, or perhaps involving a performer who had worked directly with the composer.  An editor might introduce, for example, metronome marks to approximate a given recorded performance—although this might very well be misleading.  I have already remarked in this blog of an instance where the composer’s performance tempo slowed down considerably over the years; and in early recordings, where the play duration was short and unavoidably constricted by the dimensions of the playback medium (wax cylinder or shellac disc), performers are known to have opted for faster tempi just to squeeze their rendition into the time available.  As far as new critical editions are concerned, my feeling that the editor should do the due diligence of studying any recording that might have claims to be authoritative in any respect, even if none of the findings make it out of the critical report.

SOURCE:  baerenreiter.com

An interesting example of this is to be found in Jonathan Del Mar’s excellent edition of the Elgar cello concerto (Bärenreiter, 2005).  The critical commentary is a wonder to behold, containing seventeen color facsimiles comprising the whole of the solo cello part in Elgar’s hand prepared for the cellist who gave the premiere (Felix Salmond), four pages of the original short score draft, and the first page of Beatrice Harrison’s copy of the printed solo part.  Later in the commentary Del Mar carefully catalogues the pencilled instructions in this source—a significant document because Harrison recorded the work twice under Elgar’s direction (first in 1919-1920, subsequently in 1928).  These two recordings are among the sources Del Mar uses for his edition, and they feature in one of the most fascinating discussions in the commentary itself:  second movement, bb. 40-48 and the parallel passage at 78-86 (“the most severe dilemma for the interpreter in the entire work,” as Del Mar puts it).

In each of these two passages, a single idea is presented by the soloist and then the orchestra, and then the same exchanged is repeated a third higher:

SOURCE:  my resetting (with Finale) of II mm. 40-48 (using Elgar's piano reduction), reset just to fit it in a smaller space.

Of these last four bars Del Mar asks “Did Elgar intend (but not mark, assuming it as understood) the same ‘largamente   a tempo’ as four bars earlier, or did he, on the contrary (and as some soloists make a point of doing), wish these bars to make a contrast, continuing this time a tempo?”  Here Elgar’s own recordings with Harrison employ the unwritten largamente in these second exchanges, although not a single written source includes it.  (At least not a source in Elgar’s hand; Harrison has added to her copy of the printed part “largamente molto” to b. 43.)  Del Mar concludes tellingly

“Fortunately, there is at least no conflict whatever between individual sources between either group [paper or recorded], so that there is absolutely no doubt as too what we should (a) print (b) play—even if these two groups are in direct opposition with each other.”  [all of these quotations from pp. 36-37 of the critical commentary]

Even more interesting to me is that Elgar apparently took pains to erase some instruction at this point:  what was written above the cello stave here that was subsequently obliterated, distorting even the lines of the blank stave above?  This is bb. 44-46:

SOURCE:  cropped scan of Bärenreiter facsimile (2007) of autograph full score (RCM ms 402), pp. 44-5.

Of this Del Mar remarks, “there is—tantalizingly—distinct evidence of deleted markings, but these are very efficiently scratched out so that almost nothing remains.  Only from the extremities of a few individual letters can we tentatively conjecture that Vers. I might have read (44 ‘largamente’ altered to) 45 ‘largamente a tempo’.”  Interestingly, the obliterations occur in both the autograph full score (above) and in the short score draft.

In any case, acknowledging this distinction between how the music is performed and how it is notated is significant.  Del Mar decided to deal with the whole issue in the critical commentary rather than in the separately published score, but at least a footnote in the score directs the user to the commentary.  A more intrusive editor might impose instructions (bracketed or not) or more extensive footnotes to indicate that the solo in bb. 44-45 should resemble bb. 40-41, etc., citing these recordings as support for that.  (I say intrusive—but is that the right word for this?  Heavy-handed?  Patronizing would be more pejorative; the positive spin might be avuncular.)

Christopher Hogwood has cited an interesting case of this sort of detail:  Aaron Copland’s 1974 Columbia recording of Appalachian Spring in its original scoring (13 players) included a bonus disc with excerpts of Copland rehearsing the Columbia Chamber Orchestra.  At this passage

SOURCE:  cropped scan of Boosey & Hawkes study score (HPS876), p. 5. 

Copland instructs the string players “Would you mark a crescendo on the [first] A—the fermata? [demonstrates]  I’m used to that; I don’t know where it came from.”  (Hear this moment of the recorded rehearsal here.)  Hogwood comments

“That, to me, constitutes something as good as written evidence.  Copland wanted it, asked for it in rehearsal and fixed it in his recording.  That crescendo can then go back into the score, but indicated differently from the crescendos he actually wrote, being one that he dreamed he had written but never had, but asked for, and if you want to explain it in the critical notes, you can.” [pp. 5f]

Hogwood's “as good as written evidence” suggests that if he were editing Appalachian Spring the crescendo would be in the score, modified in some way (brackets, dotted lines, whatever) to indicate an editorial addition, but he felt that an indication of its source is optional:  “If you want to explain it in the critical notes, you can.”  Okay, we have the composer literally on the record in this instance, and the ensuing studio recording backs it up.  The critical notes should say at least “hairpin absent from A[utograph], B[oosey published score], P[arts]....”  I think ideally the notes would be the place to document not only the 1974 rehearsal comment, but also if the crescendo is present in Copland’s other recordings of the work (in its larger scoring).  It could therefore be a task for an editor to seek an answer to Copland's “I don’t know where it came from.”  [ADDENDUM 10 June 2020:  The new critical edition of the original ballet score of Appalachian Spring cites the rehearsal recording among sources, but no mention is made of this crescendo, nor does it appear in the score.  An opportunity missed.]

Patrick Warfield documents a much more complicated situation in his edition of six Sousa marches in the Music of the United States of America (MUSA) edition.  He lays out the case for why the early recordings are not to be trusted—greatly reduced recording forces, truncations made to fit works on to a disk or a cylinder, and uncertainty of the identity of the performing ensembles billed on the record label as Sousa’s band (often conducted by assistant Arthur Pryor).  Add to this Sousa’s jealously guarded authentic sound for his own music in live performance:  the published texts of the marches lacked the details of his own performance practice.  Sousa is quoted as saying “we make some changes now and then to make it a little bit different” (p. xxxii).  At best these “authentic” recordings could document only a moment of that dynamic tradition.

Thus Warfield turns to the recollections of Sousa’s players (each keenly aware, after all, of intentional departures from the face-value reading of the printed parts) to try to establish Sousa’s performance practice as best that he can.  These changes generally involved certain groups of instruments sitting out during a repeat (or a first-time-through), getting the melody brass (cornets/trombones) or the percussion out of the way to let a mellower ensemble sound prevail; or it might be moving players to a higher or lower register—clarinets an octave higher or lower than notated.  Warfield’s edition cautiously refrains from printing these alterations directly into the score (“as no printed source includes these changes” [p. xxxviii]), although they are indicated in bracketed instructions.  For example, this bit of The Washington Post:

SOURCE:  cropped scan from Sousa: Six Marches (A-R Editions, 2010), p. 10; there are further instructions at the bottom of the page as well.

Warfield has done an admirable job presenting the evidence of Sousa’s practice without imposing it.  Conductors may experiment as they like.

A more vexing sort of recorded evidence is conveyed by surviving mechanical instruments like player piano rolls or the eighteenth-century barrel-organs that preserve versions of Handel’s organ concertos.  (For the former, Neal Peres da Costa has done admirable work disentangling performance practice evidence.)  The Handel concertos are shown to be laden with what might otherwise be considered improbable ornamentations.  Of course these cannot be exact transcriptions of Handel’s—or anyone’s—performances, as they have had to be “pinned” by hand (the metal pins hammered one at a time into the wooden cylinder).  Due to the minute distances of linear travel as the cylinder turns, it is hard to believe that these can transmit very much in terms of precise rhythmic relationships, still less is there a pattern to suggest notes inégales:

SOURCE:  cropped scan of David Fuller's transcription of an eighteenth-century barrel organ [p. viii]

For pitches (for example, starting a trill with the principal note) the barrels are much more reliable.  They certainly serve to indicate something of the variety of added ornamentations known (even plausible) at the time, and what sorts of ornaments would happen on repeats while other things might be altered.  When such instruments came to be studied in the 1980s (for example this) there was great hope that they were a sort of Rosetta Stone for understanding Handel’s performance style:  thus David Fuller insists that

“Here, in principle at least, there are no decisions to be made, no opportunities for interpretation.  One may like or dislike what he sees here [in Fuller’s meticulous transcriptions]; one may not dispute it....  The listener may imagine his ear pressed to a speaking tube extending without obstruction nearly 200 years into history.”  [p. v]

This was too good to be true, but that doesn’t make such evidence irrelevant by any means, and a few pages later Fuller backs down a bit to something much more useful:

“That Handel himself played this or that particular ornament on a particular note in a particular measure could not possibly be claimed; this his style of playing was wholly without effect on general English practice of mid-century and thence upon these cylinders is, on the other hand, unlikely.”  [p. x]

Beyond Handel, such barrel organs can offer us a lot about early eighteenth-century ornamentation in general.  Paul Badura-Skoda even opens his Interpreting Bach at the Keyboard with a chapter on the Handel barrel organs. But these tell us more about eighteenth-century musical culture—barrel organs in particular—than they do anything about keyboard playing or ornamentation, and they must be treated with caution.

The barrel organs are recordings of performances rather than notational instructions about music—and I think we must keep that distinction in mind.  In 1958, Charles Seeger articulated concepts of prescriptive and descriptive notation—“a blue-print of how a specific piece of music shall be made to sound” over against “a report of how a specific performance of it actually did sound” (MQ 1958, p. 184).  When we think of music in terms of composers and works, we are (almost invariably) conceptualizing written music as prescriptive:  How did the composer want this to sound?  When, instead, we think of music in terms of performers and performances, we conceptualize notation as a description of that performance:  How did the performer render this?

The notation may well look pretty much the same in either case, as (despite what Seeger argued for in 1958) descriptive notation is still very much bound to the notational elements devised around prescriptive writing, particularly if the descriptive notation is expected to be an adjunct to some sort of recording of the real thing.  Thus the curious, 1100+ page anthology The Beatles: Complete Scores is descriptive of the Beatles recordings, laboriously (although to me not always convincingly) transcribed by Tetsuya Fujita, Yuji Hagino, Hajime Kubo, and Goro Sato.  I presume it is a labor of love, and it’s difficult to know what it is for:  a coffee table curiosity (commercial)?  A handbook for cover bands (prescriptive)?  An ancilliary resource—but a dangerous one—for scholars of the “British Invasion” (descriptive)?  We can see more rigorous approaches in the MUSA volumes devoted to (for example) transcribed recordings of “Fats” Waller (ed. Paul Machlin) or Sam Morgan’s Jazz Band (trans. John J. Joyce)—fascinating volumes to peruse, even when I did not have the recordings immediately to hand.  These volumes do much to emphasize the complexity of this music, and of course the notation acts to “freeze” the improvised music to allow us to scrutinize and dissect it (... to apply, in other words, the autopsy-table analysis that has been the stock-and-trade of music scholarship).

There is much more to be said here, but this post is already overlong.  I should return at some point to some prescriptive transcriptions—that is, of transcriptions from recordings intended to facilitate new live performance of music that was originally improvised—like Maurice Duruflé’s reconstructions of Charles Tournemire’s Cinq Improvisations, or (rather differently) the Jazz Arts Trio’s “note-for-note transcriptions of historic moments in piano jazz.”  In the latter instance, when I sat through a concert in which these transcriptions were realized (and with scores available for purchase), I was left pondering what manner of performance this could be.  Somehow the music seemed to have been violated in an attempt to bring it back to life.

8. The right tools for the job

Although it isn’t unusual to find players disregarding a composer’s instructions about what instrument to use—frequently the specified instrument simply is not available—it is profoundly irritating to find editions that beat the player to it.  An example of this is the supposedly scholarly edition of Bizet’s Carmen edited by Fritz Oeser (Alkor [Bärenreiter], 1964), where the parts Bizet wrote for cornets-à-pistons are labelled trumpet instead:

SOURCE:  scan of Oeser ed., p. 1 (detail, with emphasis added)

I say “supposedly scholarly”:  the use of trumpets is the least of this edition’s faults—and some of these will certainly feature in future posts.  In any case,  Winton Dean made this point fifty years ago, demonstrating that the Oeser edition “goes disastrously off the rails” from sound editorial practice (p. 284).  Dean also remarks on the trumpet/cornet issue, pointing out that this mis-allocation had already happened in the Peters edition edited by Kurt Soldan (on the IMSLP here) although at least Soldan had the courtesy to add “Pistons” parenthentically underneath.

The instruments are regarded as interchangeable, probably because the cornet  (properly with accent on the first syllable, which makes it easily distinguished from the cornett) is virtually an endangered species, particularly in the USA.  A hundred years ago it was the other way around, and the trumpet seems to have taken over in the 1920s (with perhaps the coup de grâce being Louis Armstrong’s conversion to trumpet).  Thus we find Cecil Forsyth, writing in his Orchestration (1914/rev. 1935):

We must not forget that the contempt which is usually bestowed on the Cornet by those who have never heard it properly played is mainly a contempt because it cannot equal or beat the Trumpet in Trumpet passages.  These simple and straightforward phrases were always consciously designed by the old masters to produce their somewhat oppressive effect by the mere weight of the instrument’s tone.  In course of time we have come to associate that type of tone with that type of passage. (p. 107)

Erring on the side of oversimplification, the conical cornet is a melody instrument; the cylindrical trumpet is a rhythmic instrument, with a tone better able to cut through an ensemble.  To best illustrate the timbral difference between conical and cylindrical brass I suggest this performance of the piece we usually call “Jesu, Joy of Man's Desiring” (from BWV 147) by the German Brass.  The arranger has cleverly divided the ensemble into a two distinct consorts.  For most of this arrangement, the 4-part chorale is played by cylindrical instruments (one trumpet and three trombones), while the rest is played by conical instruments (two flugelhorns, two horns, tuba—although at a two points a piccolo trumpet joins in, and in the final measures everyone is playing together).

Here is a handy diagram, from Anthony Baines’s Oxford Companion to Musical Instruments (1992), which compares the profile of several brass instruments, although it warrants a few comments below.

SOURCE:  detail of scan of Baines, p. 43 (entry:  Brass instruments). 
Baines does not clarify this, but this diagram must assume that no valves are depressed, nor is the trombone slide extended; 
V then would indicate the passage through the valve section of the instrument, but not diverted through the crooks the valves would engage.  
Otherwise the proportions don't make sense, as (for example) nos. 4 and 5 at their full length would be much closer to #1.

A mountain out of a molehill?  Profiles 4 and 5 do not appear vastly different from each other.  Indeed, but the modern B-flat trumpet is significantly shorter—and thus proportionally more conical—than its predecessors in the Renaissance and Baroque, and indeed even down to the early twentieth century.  The old valved F trumpet of the late Romantic orchestra (an instrument which will figure in a later post) really would sound noticeably different from today’s ubiquitous B-flat, as the F trumpet has a narrower bore and is about half-again as long.  (Its profile would most resemble #1 in Baines’s diagram, although a little shorter.  I don’t think I’ve ever heard one live, although it can be heard on some of the recordings of the (new) New Queen’s Hall Orchestra; otherwise it is essentially extinct.)  If Bizet had been asking for trumpets in Carmen, he would have been expecting something of that sort—not the instrument we see and hear today.

Note that profile 6 above is only the shorter portion of the standard “double” horn today (F/B-flat); the F side would extend off well to the right, and would thus be proportionally more cylindrical.  For orchestral horns, the principle that the longer the instrument is the more cylindrical its profile has always been true:  even before valves, with the crooks players would use to change the key of the instrument, the low crooks produce a very different sound from the high crooks.  Listen, for example, to these recordings of the opening chorus of Bach's cantata Sie werden aus Saba alle kommen (BWV 65) with either C alto or C basso horns:

• C alto:  Gabrieli Consort (dir. Paul McCreesh)
• C basso:  Amsterdam Baroque Orchestra (dir. Ton Koopman)

It isn’t just the pitch, but the whole timbre of the instrument that is changed.  At the extreme:  the shortest horns of which I am aware are the E-flat alto horns in Mozart’s K. 132 (the higher pair of two pairs of horns), which make a very round, bugle-ish sound.

In the ensembles that put pairs of cornets and trumpets together—as so often in French nineteenth-century literature, or in military band music—one may frequently find composers observing a distinction between the writing appropriate for one or the other.  Thus in his wind band piece Sea Songs, Ralph Vaughan Williams neatly distinguishes between cornet solos and trumpet solos, and (as below) demonstrates how to keep trumpet punctuation from overwhelming cornet lyricism, while still sometime needing either instrument to function as the other:

SOURCE:  scan of Sea Songs full score (Boosey & Co, "corrected edition 1991"), pp. 20-21 (composite details)

Likewise, Emamanuel Chabrier knew when to let the trumpets do the heavy lifting in a thick texture, relegating the horns and cornets to filling in the harmonies:

SOURCE:  scan of pp. 112-13 of 1997 Dover reprint of 1884 first edition of España (composite details; accolade added)

Working on this post got me wondering of any instances where a composer calls for a player to switch between the two instruments.  It seems like an obvious expedient, but it is curiously rare.  The published full score of Bernstein’s Candide reveals that for the Parisian waltz the first trumpet player takes up the cornet (an allusion, perhaps, to the showy cornet part that Berlioz added to the Valse of his Symphonie Fantastique?)  In the last movement (Marcia funebre) of Bartók’s Four Orchestral Pieces, op. 12, the four trumpet players switch back and forth between trumpets and cornets, although it would be hard to demonstrate that he is absolutely consistent with the sort of idiom he gives to one or the other instrument.  

But in other instances where one might expect such a practice, it just doesn’t happen.  In Gilbert & Sullivan’s Iolanthe, for example, the fanfares which begin the chorus “Loudly let the trumpet bray” are played by cornets, Sullivan’s default in the Savoy Theatre orchestra.  (Like Bizet’s Opéra Comique, this may have been a balance issue more than anything else; when Sullivan wrote his grand opera Ivanhoe, he called for trumpets—and even a Wagnerian bass trumpet.)  Seldom do I hear cornets in performances of the Savoy operas.  The rare cornet solo does strike my ear as odd on the trumpet, though.  Compare the solo in the overture of The Pirates of Penzance (by Sullivan's assistant, Alfred Cellier) in these two performances, one with trumpet (at least to my ear, although the player is trying to compensate with a fair bit of vibrato) and the other with cornet.  I note that the Kalmus score (apparently scored from parts) gives no indications that the cornet was what Sullivan had in his ensemble:

SOURCE:  scan of undated Kalmus full score of The Pirates of Penzance, p. 3 (detail)

I started this post remarking on performers disregarding the instructions of a composer, and have given a few examples of editors doing it.  I have been wondering if we could imagine an editor imposing cornet where a composer wrote trumpet?  Has this ever been done—outside of brass band transcriptions?  I can think of a place where I’d like to do it:  the “Tango-Pasodoblé” movement of Walton’s Façade.

SOURCE:  scan of William Walton Edition vol. 7, Façade Entertainments, p. 42 (detail)

This seems to my ear to be begging for a cornet (with its quotation of “I do like to be beside the seaside”), but Walton’s tiny ensemble calls for trumpet.  Changing that is the sort of liberty a conductor is free to do, but not an editor.

2. Risky business

It’s not news that music is a risky business.  Even with the narrower field of musical text, there is an element of risk.  Certainly music publishers are now more than ever in a risky business:  they compete for what is apparently a diminishing number of customers, and this competition may be against free legal downloads even more than against other publishers.  For a composer, notation is itself risky—virtually every element fraught with potential misinterpretation.  If they choose for their music to be performed by live musicians (and that is no longer the no-brainer that it used to be), they must give up some degree of control of “their” music:  the score is the inevitable nexus between the person trying to get the musical material in and those who would get the material out.   And I find the issues that arise from that interface compelling.

Much of what I will write in this blog has been said before—sometimes, indeed, because I want to point out things others before me have said (and done) that deserve renewed attention.  One such example is a small booklet by Walter Emery, published in 1957 by Novello & Co.  He called it Editions and Musicians, but I think his subtitle says it all:

Emery can get preachy at times.  At times that is stimulating, at times perhaps a little tedious.  I appreciate every page of this book, even when I find that I disagree with him.  For example, Emery concludes his chapter “The Need for Editing” thus:

“The fact is that until an editor has done his work, and done it properly, no performer can safely play old music, no analyst can safely analyse it, and no historian or critic can safely assess it.  The editor’s work is the foundation on which all other musical and musicological activities are based.”  (p. 14)

Really?  I can remember a time when I would have endorsed every word of this, but no longer.  It is not that I am against good editions of music—on the contrary, even when they cost a bundle.  It is really the word “safely” that bothers me.  Implicit in it, I think, is the idea of a definitive conception of the piece in the composer’s head, and that we are obliged to try to get as close to that as we can before we do anything else.  Moreover, I think Emery would say that these other activities (performance, analysis, criticism) that occur without such a proper editorial foundation are doomed eventually to crumble.

I’m skeptical that such a definitive conception ever existed in any meaningful and permanent way, and even more skeptical that such an idea could be conveyed to anyone else.  Think of the composers—first-rate composers—who couldn’t leave well enough alone.  (Not just “well-enough”—we might sometimes tend to gush with phrases like “apparent perfection.”)  Chopin might be an extreme example of this:  not only was he praised for “ever different expression” (and “Chopin never played his own compositions alike twice,” etc.), but those remarks are consistent with the documentary record of his own manuscripts and the editions that flow from them (i.e., all sorts of authorized variant readings).  Even given the fascinating editorial collation that has been unfolding through the superb Online Chopin Variorum Edition, isn’t there perhaps just too much here for anyone to take in?  Must the establishment of every authentic Chopin text of a work be complete before anyone can “safely” (to use Emery’s word) perform or analyze or critique it?  Could Emery be satisfied if an authentic text (say, a copy of the first French edition of a work marked up by Chopin for his pupil Jane Stirling) was all that a performer had seen?  This is, of course, extremely improbable, given the glut of Chopin editions on the market, settled in domestic piano benches, littering cabinets and weighing down shelves.  When Emery writes of the need for editing, it is a need to redress a context dominated by editing that was not done “properly” (again Emery’s word).

Or consider J. S. Bach, for whom, as Christoph Wolff has put it, there was apparently “no such thing as ‘untouchable’ text,” evidenced by his changes (not just “corrections”) even to his published works, and his revisions and reworkings of  (for example) the cantatas as he reused them in a subsequent years.  These are often instances where it is difficult to argue one reading is better than the other (indeed, there a few cases, like the two versions of the Canonic Variations, BWV 769 and BWV 769a, when it is seems impossible to say definitively which version came first).

Or Mendelssohn—whose “Revisionskrankheit” was a chronic (fatal?) compulsion to tinker away endlessly at his completed works.  At some point in this blog I will eventually return to the case of Mendelssohn’s celebrated “Italian” Symphony—a work that until recently was known only in its first version; Mendelssohn’s second thoughts prompted him to rewrites of the last three movements; the first movement he ultimately deemed unredeemable.  All this notwithstanding, it is one of his most popular works with audiences, who don’t seem to have noticed how unredeemable it is.  But more on that later.

In my first post, I mentioned the very different manifestations of Mussorgksy’s Pictures at an Exhibition, diverging in many respects from the autograph score, but still holding a secure place in the repertoire of pianists and orchestras.  I’d guess that Emery would regard any performance of the Rimsky-Korsakov version (or the editions derived from it) as a travesty.  But what about when later minds uncover new ways of hearing a piece—and a composer even embraces those new readings?  (Think Bob Dylan’s response to Jimi Hendrix’s version of “All Along the Watchtower.”)

I have been delighted to see that the editorial principles behind the new Works of Gioachino Rossini (Bärenreiter) take into account not just the authorial conception of the works, but also these works in performance, accumulating non-authorial performance traditions.  The score of the recent edition of Il barbiere di Siviglia (ed. Patricia Brauner), for example, includes in appendices not only three substitute numbers Rossini composed for later productions, but also several vocal variants (i.e., ornamented versions) that are extant in Rossini’s hand in as many as three (differing) sources.

Source:  cropped scan of p. 504 of WGR score of Il barbiere... (Bärenreiter, 2009)

Beyond this, however, the critical commentary includes materials about the performance tradition not directly deriving from Rossini:  a) the libretto for a Neapolitan-dialect version with spoken dialogue, and b) an extensive essay (59 pages!) by Will Crutchfield on “Early Vocal Ornamentation.”  I think it is a pity that these gems lie in the comparative obscurity of the very pricey commentary volume.

Would Emery say that as these latter variants are not from Rossini (but rather from the early performance tradition), a “safe” performance would be no performance at all?  Or that analysis or critique would be meaningless because it doesn’t relate directly to the composer?

Then again, just because Chopin was known to interpret his music anew each time, he didn’t take just any pupil; should we read his varying instructions to them with similar caution?  His famous remark to Filtsch (“We each understand this remark differently, but go your own way, do as you feel, it can also be played that way”) could hardly be regarded as interpretive carte blanche for any pianist.  Still, there is a sense in which G. Thomas Tanselle’s conceptual distinction between the text of the work and the text of the document might be useful—another idea I will be returning to in later posts.

I find that I’m more willing to see more risky performances, ponder risky analyses, and read risky critiques.  The safe alternative seems embalmed (fossilized?) in an established text, when the text itself was never sufficient.  That said, inauthentic readings (when they are known) should be acknowledged as such.  The situation can become bizarre, such as an exchange prompted by David B. Levy’s review of a critical edition of Beethoven’s ninth symphony by Jonathan Del Mar (1996).  (Levy's review is “Urtext or Performing Edition?” in Beethoven Forum 9/2 (2002), pp. 225-232).  Del Mar’s edition is a remarkable accomplishment in many respects, but Levy is astonished to see Del Mar depart from an established reading to favor a variant that is admittedly authentic (mvt. I, b. 81; the D in flute and oboe, rather than the familiar reading of B-flat):

Source: detail (woodwinds only) of scan of p. 13 of Del Mar’s edition of Symphony no. 9 (Kassel:  Bärenreiter, 1996)

[Here is a performance of this moment with this text.]

This “new” (original) reading results in a curious discrepancy in the subsequent appearances of this figure, where there is no leap of a sixth.  (This isn’t really an issue of recomposition in the recapitulation, but rather a different conception of motivic unity.)  Yet, as Del Mar remarks in the critical commentary, this D appears in all sources (which for this instance means the autograph (below), five scores made by copyists, and the first edition score and parts (1826)).  Del Mar surmises that the Bb (which first appeared in the 1864 Breitkopf edition)

“… was apparently invented by analogy with 276/80,346 [later recurrences of the motive].  Yet Beethoven wrote d in both instruments in A, so it can hardly be a mistake.”  (p. 25)

Source:  detail of screenshot of http://beethoven.staatsbibliothek-berlin.de/beethoven/pix/sinfonien/9/1/990/00000020.jpg  
(mm. 80-84 as in Beethoven’s autograph f. 8v.)

 Levy puzzles that Del Mar would depart from the reading that would support “many published analytical studies… [which] have drawn attention to the significance of the perfect fourth (and its inversion, the perfect fifth) not only within the first movement, but throughout the entire work” (p. 229).  Del Mar replies:

“… where I as a sensible musician have to judge that the reading in the authentic sources is inconceivable (no less)—and especially where I can show how the error could well have arisen—I will present the more likely text. But if it is conceivable, I have a duty to stick to what Beethoven wrote. So despite all published analytical studies—which inevitably were based on the text they had in front of them—I restore Beethoven’s D in movt. I, m.81. Sorry: if we subsequently find that the analysts’ text was faulty, their studies will have to be rewritten. That is quite simply inevitable, and to argue that we must print a text that accords with previously published analytical studies is obviously putting the cart before the horse.”  (p. 105)

No need, in my view, for these “risky” analyses to be “safely” re-written:  the analyses can account for the work as it came to be known.  More significantly they document an important aspect of reception history:  our(?) sense of what the product of a genius should manifest, with (in this instance) an organic unity of motivic development encompassing the whole work.  The altered B-flat is no less a part of Beethoven’s 9th symphony (i.e., the cultural property it has become) just because Beethoven didn't write it.  If, with Del Mar, we regard the B-flat as bad editorial judgment, we must now correct it; we can't eliminate the effects it has had over the years (and we could, if we chose, continue to play it and analyze it), but we do have to acknowledge that Beethoven didn't write it.  Those who disagree with Del Mar should still acknowledge that Beethoven didn't write it, but they may argue however they may that Beethoven meant it.

As my envoi, take this example from Emery (p. 19), who quotes Donald Francis Tovey’s hedging conclusion when faced with an (unknown) chromatic error in his text of the F# minor prelude from Bach’s WTC II: “a harmonic point of peculiar subtlety.”  But of course.  That’s what we expect from a genius, right?