Top of page

Collection The Moldenhauer Archives - The Rosaleen Moldenhauer Memorial

Igor Stravinsky's Threni: Conducting Details

This set of undated printer's proofs for Threni1 was submitted to Stravinsky's attention in three separate missives (pages 1-27, 28-51, and 52-70, as shown on the initial pages of each) by his London publisher, Boosey & Hawkes. Clearly identified on each successive sheet as first proof, it reveals the corresponding first stage of corrections. Subsequent stages include not only second proofs (which the publisher normally supplied for each work, as evidenced by those I have seen for the Fanfare for a New Theater, Introitus, and Requiem Canticles, the last of which I was asked to correct under the composer's supervision, with his express permission for every correction suggested), but a second edition of the score, printed nearly seven years after the first, in May 1965, incorporating a great many changes in barring and eliminating some errors. Although the present set of proofs contains the scribbled, underlined indication "IS's final corrections" on the top left-hand corner of the first page, such are nowhere to be found--at least, not in Stravinsky's handwriting. All corrections on these proof sheets are in one handwriting only, and I assume it to be that of Leopold Spinner, a composer who worked for Boosey & Hawkes during those years. (I had some correspondence with him over the Requiem Canticles proofs in 1966-1967; his small, spare writing can be recognized from the markings on those proofs as well as from his signature.) I do not believe it would have been unusual for Stravinsky to have left the correction of proofs to others; he was understandably impatient with the often thoughtless errors found in all proofs and therefore sought to spare himself the concomitant annoyance. The proofs at hand certainly show an abundance of such errors; they are in this sense not out of the ordinary. Among their more distinctive traits, however, is the frequent instruction for omitting staves, if their only occupants are measures of rests. (See pages 11 and 15 of the score, for instance; all but three of the staves on the lower system of page 11 are eliminated. On the proof sheet of page 15, the proofreader's instruction concerning the numerical indication for measure 124 is to leave it "exactly where it is in glorious isolation"; it is surrounded by a "hole" left by three omitted staves.) When Stravinsky's scores began to exhibit such typographical empty spaces--i.e., all his orchestral scores from Threni on--it was supposed that he had taken his cue from certain graphemically quirky scores of Stockhausen, published by Universal Edition. Perhaps; but Stravinsky's working manuscripts had almost always been written on unlined paper for which he provided whatever number of staves he needed, and only wherever he needed them, by means of a staff-line-drawing implement of his own design.2


Threni stands out among Stravinsky's compositions of the late 1950s by virtue of its elaborate interweaving of small formal schemes, adding up to large, continuous movements, as well as its predominance of unaccompanied choral and solo vocal music, and--in that connection--its frequent canonic writing. There is no way in which the publication, in 1957, of Ernst Krenek's Lamentatio Jeremiae Prophetae3 (written in 1941 for mixed chorus--in up to nine parts--a cappella) could have remained unnoticed by Stravinsky; he was evidently stimulated by Krenek's piece to try his own hand at setting parts of Lamentations in Vulgate Latin, to employ canonic means of continuity, to use unaccompanied voices and--not least--to use an "antiquarian" type of vocal notation without bar lines and in large note values. Yet it must be emphasized that any notion of direct influence by Krenek's work ends at the merest surface. Both pieces may share, above all, certain graphemic features in notation, but their musical identities are unmistakably distinct and utterly different from one another. (Even the device of hexachordal transposition-rotation, employed in characteristic ways by Stravinsky from Threni on, is illustrated by Krenek in his prefatory remarks, but employed for totally dissimilar purposes in his Lamentatio, as though it were an offshoot of Renaissance modality.)

The peculiarity of unbarred notation of large note values in Threni (as found, for example, in the Canon a 4, duplex, from measure 189 on page 25 to the end of page 27) had a certain precedent in Stravinsky's reduction for two pianos of a segment of Agon4, in which he assigns different metrical designs to each instrument: the "Bransle Double" is notated in 3/2 in Piano I, and in 8/4 (3+2+3) in Piano II between measures 336 and 352, and again between measures 365 and 372. Bar lines are shared equidistantly, inasmuch as the polymetric design remains constant; yet in the corresponding "Bransle de Poitou" of the orchestral score, all the music in those passages is written in 3/2, in accordance with the demands of sheer practicability in performance. The difference between the two modes of notation highlights the illusory aspect of the many places that were notated without barring in Threni. Even before the first performance, conducted in Venice by Stravinsky, on September 23, 1958, the practical necessities of rehearsing the chorus and six soloists would have compelled him to realize that beating only the tactus (i.e., successive downbeats) in unbarred passages would make it virtually impossible for singers to count reliably, as well as for the chorus (and orchestra) to make correct entrances. Whether before or during the preparations for the Venetian premiere, or after it, the fact is that by January 1959, in time for the rehearsals, performance, and recording sessions in New York, Stravinsky had written in bar lines for almost all previously unbarred places (the sole exception being the "Monodia" in the first section of the Querimonia: pages 20-21; measures 167, 169, and 172). Robert Craft, who conducted and prepared that performance, took these new barrings into account. I transferred them into my own score while attending rehearsals and the performance at the time; they correspond for the most part to the barrings incorporated into the second edition of 1965--except for several markings made in my score by the composer, and therefore not generally known.


Since the first printing of the score is dated September 1958, it is easy to conjecture that it was, by the time of the premiere, simply too late to undertake the changes required by the practical considerations described above. Ideally, the first edition's plates should then have been scrapped, and an entirely new engraving would consequently have made a more presentably correct, ultimate edition feasible. Even after the seven-odd years it took the publisher to issue the second edition, the now included dotted bar lines and the almost invisibly added parenthetic time signatures are neither very clear nor sufficiently well-spaced, and no provision was ever made for the obvious need to renumber measures. (The amount of time needlessly wasted during rehearsals, in specifying which of the segments defined by dotted bar lines and pertaining to one "mega-measure" might be meant would probably be equal to having to rehearse each "mega-measure" entirely each time a particular note or brief note or brief passage should require some work!) It is useless, however, to speculate as to the decent-looking, practical edition that might have been--or, indeed, on the number of good performances this particularly unnoticed composition might consequently have enjoyed. We shall have to continue to make do with the poor one at hand--and with its never corrected, wrongly aligned attacks in measure 392, which should be located halfway through the spoken sixteenth-note quintolet, after the final syllable of the word "Domine."

Stravinsky's markings in my score are concerned with conducting. Of the two that are not reflected in the dotted-line barrings added in the second edition, the one on page 52 indicates the division of the 10/8 bars: into 4/8 + 3/4 (measures 323-326). But the markings on page 16, over measures 130-131, are especially curious.

For measure 130, the 1 + 4 bespeaks the rhythm of the parlando chorus, although it contradicts the 3 + 2 in the clarinet's and bass clarinet's notation. The next bar's similar indication is in keeping with the instrumental rhythms, ensuring a cleanly accented attack on the second eighth; yet it contradicts the 2 + 3 in the choral rhythm. The square horizontal brackets starting over the 4s in Stravinsky's markings signify the right hand. The preceding 1s are to be beaten with the left (!)--which is bound to remind one of the story about a "famous" conductor who, when confronted by the second movement of Tchaikovsky's Pathétique, decided he would beat it this way. Absurd though this may seem, I recall quite clearly that Stravinsky told me, with a big grin, that it really works very well. (Unfortunately, I did not attend the recording sessions to see him demonstrate that assurance.)

The other conducting marks in my score, which I copied from Stravinsky's, refer to subdivisions of measures and to beating units that are frequently diminutions of the metronomic. The opening measure of the piece is of particular interest in this regard, in view of its rhythmic-metric makeup. If beaten in "straight" 4, the danger seemed to consist in the final eighth-note dyads not to be articulated with the same dynamics as the two previous ones, on account of the metric displacement. Stravinsky's instruction reads that the measure be beaten as 2/4 + 4/8. (My suggestion, in a letter written a few weeks later to a former student and colleague5, was that the measure be divided into two 3s, with the first one beaten as quarter-quarter-eighth and the second simply as 3/8. The advantage--or, as I see it now, the danger--is that the last F-sharp will be given a down-beat. The danger would lie in its being thereby more accented than the previous F-sharp.)

Indications pertaining to conventional divisions of 5s and 7s need not here be detailed--they are notated merely in order to avoid conducting slips. It is of some interest, in that connection, that the metronomics (i.e., counting units) are frequently subdivided: at measures 73 and 80, for instance, the tempo is given as quarter=60, yet the conducting indication is for eighth=120. Moreover, the triplet rhythm at the end of measure 81 is provided with a little triangular reminder marking (although the equivalence of triplet eighths to three eighth notes across the bar line is neither arcane nor difficult); but it is perhaps there because of the increased instrumentation (compared to the similar equivalence between measures 75 and 76).

Stravinsky's metric indications for the "Monodia" (measure 167) are confined to pinpointing built-in units of 3/8. He described these recurring interpolations as "joints."

The unbarred heading for De Elegia Quinta, measure 384, is to be conducted in two large measures: 7/4 and 9/4, respectively, divided into 4 + 3, and 3 + 3 + 3.

The included metric indications for measure 178 et seq. are elucidated by Stravinsky in my score as follows: the measure at 178 (2/4 + 1/8) is to be beaten in 3; the next two measures of 5/4 are both 2 + 3. The telling admonition here was that larger triplets are never to be beaten as triplets, but always in the context (as here) of the prevailing quarter unit (i.e., = circa 56; see measure 166), or of the half-note unit, as in the continuation of the "Canon duplex" on pages 26-27 (measures 191 et seq.). The seeming contradiction between the nonbeaten triplets of these passages and the little triangular instruction in measure 81 is soon resolved, obviously.

The opening of Sensus Spei, measure 194, is preceded by Stravinsky's recommendation of three preparatory quarter beats; measure 194 is then taken as the fourth quarter of an imagined 4/4 measure in the new tempo. The reason for this is found in the necessity both of alerting the performers to oncoming eighth-note motion, and of avoiding a longer pause between sections of De Elegia Tertia. Soon after this, the measure of 20/8 (measure 197) is fanciful in accommodating the presence of double whole notes, but insufficiently clear in its notation for the conductor. Perhaps because of this, Stravinsky beat it as two bars in 4, after the model of measure 196 (which is notated, and beaten as, 3 + 2 + 3 + 2). Here, however, the 4 beats correspond to one bar of 11/8 and another of 9/8. Robert Craft preferred to treat measure 197 as 5/8 + 8/8 + 7/8, while observing the notated groupings. In keeping with the instruction concerning the large triplet in measure 178, the time signature in measure 200 means: beat 3 (i.e., 3/8 + 2/4), so that the triplet will fall neatly into a submeasure.

My score contains a purely ad hoc correction (or emendation) by Stravinsky for the little canon for sopranos and altos in measures 247-248: the composer first wanted the sopranos to double altos on their initial middle D, for needed emphasis. The repeated syllable, as well as the ensuing octave leap (which would likely have caused excessive stress on the upper D) then prompted Hugh Ross, the chorus's conductor, to suggest that the tenors be enlisted to articulate the opening middle D eighth note, and so it was done. As for the canon's ending, with its più f and accents, Stravinsky had the altos iterate their F-F-sharp, and extend the word's third syllable in order to coincide on the sopranos' final notes. (This evidently represents the kind of local touch-up that should not be equated to a change in the score.)

Unlike the 10/8 and 20/8 of measure 196-197, and in view of the metronomic notated by Stravinsky (i.e., eighth-note = 180), measures 323-324 are beaten as 4/8 + 3/4--a design that supports the clarinets' accented attacks and that does not interfere with the vocal phrasing; the syncopations are not lost.

Lastly, the sul ponticello required for the string chords at measures 322, 328, 336, 344, 348, 353, 359, 367-368, and 377 elicited Milton Babbitt's and my suggestion that they be played so that the attacks would be as specified, but that immediately after those attacks the bows would all be moved nearer the fingerboard, so as to allow the simultaneities to resonate properly. Otherwise, these simultaneities would sound indeterminate, squeaky, or like unintentional harmonics. Our suggestion, happily, was taken.

The detailed list compiled above was taken from rehearsals and a performance--all of it now almost forty years ago. A conclusion to be drawn from virtually every item mentioned on that list is that during those intervening four decades the nature of orchestral capability has undergone remarkable changes in this country. (I am referring, naturally, to those still relatively few orchestras that have been given a modicum of opportunity by their conductors to become acquainted, at least,--if not yet really familiar-­with the kind of twentieth-century repertoire of which Threni might be classed as an instance.) It is obvious, therefore, that the precautionary nature of many of Stravinsky's conducting indications and admonishions was called for by the circumstances surrounding short-order performances (i.e., with a very small number of scheduled rehearsals, and with a pickup ensemble of orchestral players) at the time. Yet it is no less evident that, given a marked increase in the level of orchestral capability, such conducting requirements no longer obtain, or no longer apply to the same degree. Stravinsky's notion, for example, of actually beating eighth notes in a rapid 3/8 (as in measure 29) seems, now, quite unnecessary, possibly even confusing, and it harks back to an earlier age in which a change (no matter how temporary) in counting units from quarters to eighth notes was novel. (Stravinsky's conducting habits go back further in this regard than even Jeu de Cartes (1936) whose 3/8 interpolations (at rehearsal number 9, for instance, or at 1 measure before 10) I clearly remember him always to have conducted as a very rapid, tiny 3--at eighth-note=216!) So, if quarters (rather than eighth notes) are beaten in the 4/8 measures preceding measure 29, then that measure is surely to be beaten in 1. Similarly, the carefully marked triplet triangle in the second half of measure 81 would not in the least need to be beaten out; it would simply be the second beat of a 2/4 bar, and the attendant equivalence across the succeeding bar line would be inferred easily enough to enable the conductor to beat the 3/8 measure in 1, by the same token, rather than in Stravinsky's 3. I would think, in fact, that the proper preparation of a performance of Threni would, for the orchestra, nowadays take a third--or certainly half--the time it took in 1959, and it is very likely that the appended performance would be a good deal better into the bargain.

-- Claudio Spies


  1. Igor Stravinsky, Threni, id est Lamentations Jeremiae Prophetae (Boosey & Hawkes, 1958). [Return to text]
  2. See Vera Stravinsky and Robert Craft, Stravinsky in Pictures and Documents (New York: Simon and Schuster, 1978): pl. 12, 22, and 24, following p. 400. [Return to text]
  3. Bärenreiter-Ausgabe 3648. [Return to text]
  4. Igor Stravinsky, Agon: Ballet for Twelve Dancers, score and piano score (Boosey & Hawkes, 1957). [Return to text]
  5. Michael Senturia, conductor of the University Orchestra and member of the Faculty in the Department of Music at the University of California at Berkeley from 1962 to 1992. [Return to text]