In response to:

Schubert's Last Sonatas from the February 2, 1989 issue

To the Editors:

Alfred Brendel’s elegant essay on the late sonatas of Schubert [NYR, February 2] contains many insights, but in one respect he is off the mark. Although he argues generally for scrupulous fidelity to Schubert’s compositional intentions, where they can be known, he seems somewhat disingenuous when it comes to the matter of observing the repeats that Schubert included for the exposition (or opening section) in the first movements of all three late sonatas.

Mr. Brendel argues that “repeat marks must not be taken as orders to be automatically obeyed, as if the repeatable section were written out by the composer.” But in fact the last two sonatas have more than mere “repeat marks”: in each there are several bars of music specifically written out by the composer to lead back to the beginning. This so called first ending (comprising four bars in the A-major sonata and nine in the B-flat) is not played the second time around and would not be heard at all if a performer fails to observe the repeat.

As partial support for the dispensability of repeats Mr. Brendel adduces what seems to me rather shaky evidence: an ignorant remark by Dvorák which can hardly be taken as a “modern” critical insight, and one comment attributed very indirectly to Brahms (reported from memory by a pianist citing an anonymous “young musician”).

Mr. Brendel suggests that in music, as in food consumption, the approach of a “restrained epicure” is preferable to that of a “glutton”: less can be more. Yet it is clear (if the metaphor may be extended) that he wants to have his sonatas and eat them too. He suggests that the last three sonatas form a kind of cycle and thus “illuminate one another” when performed together in one evening. If this is done, he says, the repeats cannot reasonably be observed. But this is circular, even contradictory, reasoning.

Although Mr. Brendel is justified in feeling that the late sonatas “belong together” because they may be said to trace a certain spiritual progression, there is no indication that Schubert considered them a cycle in any practical sense (unlike, say, the song cycle Winterreise). If Mr. Brendel wants to play all three on one program, might he not be accused, by his own criteria, of some degree of gluttony? And if he does choose to perform them together, then why stop short of taking the repeats that Schubert actually wrote? It might add about a half-hour to the program, but I doubt that many listeners would shift restlessly or look at their watches, especially since they have probably come to such a recital out of a love of these remarkable pieces (and perhaps in part out of a desire to hear the first-ending measures so rarely played).

I write not as a performer, but as a musicologist—and as a genuine admirer of Mr. Brendel’s playing and his intellect. I appreciate both the sheer physical endurance necessary to play a long recital and the difficulty of sustaining tension across the vast expanses of Schubert’s music. The repeats certainly compound that difficulty. But they should be accepted as part of the performer’s challenge, not explained away with equivocal historical and aesthetic arguments.

Walter Frisch
Associate Professor of Music
Columbia University
New York City

Alfred Brendel replies:

More than other works, the great sonatas in A major and B flat major have put the enthusiasts of repeats on alert: Did Schubert not write out a few bars of transition that lead back to the beginning? To leave out any original music by Schubert, we are told, is unforgivable; it would be equally unjustified to apply cuts elsewhere.
I beg to differ. In both sonatas, the expositions, or first major sections, end in a way that does not permit a simple return to their beginnings. For this reason Schubert, who in certain external matters of form and notation was much more old-fashioned than Beethoven, and adhered to the convention of indicating a possible repeat, wrote the disputed bars. Beethoven, from his middle-period works on, sometimes ignored this convention. He doubtless would have gone straight into the development (the section following the exposition); in his “Appassionata,” the F minor exposition ends in A flat minor, followed by the start of the development in the unexpected key of E major. It is intriguing to speculate on how Beethoven might have led back to the F minor beginning; what matters is that he did not see fit to do so. He was evidently able to rise above the pressure of convention where the music called for it.

Another argument put forward by the exponents of the B flat sonata repeat emphasizes the amazing newness of the transitional bars; they add something to the piece that would otherwise remain unsaid and, supposedly, change the perception of its character. Even if there were not so many other factors—the generosity of the exposition, the literal recapitulation, the lyrical character of all the themes as well as of the following Andante, the balance of the movements—working against the advisability of this repeat, I would, for once, have to be at odds with Schubert’s judgment. Which elements in the course of the B flat sonata would justify the emergence of the transitional bars in question? Where are they announced? The transition is not to be compared to the irrational explosion in the Andante of the A major sonata: that has a psychological connection to the bleak melancholy of the movement’s beginning, as well as to the chromatic episodes of the preceding Allegro. By contrast, the transitional bars of the B flat sonata upset the magnificent coherence of his movement, whose motivic material seems quite unconnected to the new syncopated, jerky rhythm. Is the material or atmosphere of the transition taken up anywhere in the later movements? Should its irate dynamic outburst rob the development’s grand dramatic climax of its singularity? Most painful to me, however, is that the trill, which is so important to the understanding of the sonata’s main theme, is to be played fortissimo, while elsewhere in the movement remaining remote and mysterious. Schubert’s first draft still presents the trill, after a relatively brief exposition, in pianissimo.


No, I do not play Schubert’s last sonatas in one evening because I am aspiring to be a musical glutton. Or because I want my public to choke. Or starve, deprived of repeats. Or because these works represent “a kind of cycle,” which, I feel, would take matters one degree too far. But they illuminate one another, and they should be called a family of pieces, thanks to their extensive motivic give and take (to be documented in my next book).

Under no circumstancees, and in no program combination, could I agree to play the disputed repeats. Far from wanting to “explain them away,” I have given a list of structural and psychological arguments for the omission of some repeats (and not others). Professor Frisch does not present arguments for the inclusion of the repeats in D.959 and 960 other than that the transitional bars written down by Schubert make them playable, and that he wants to hear them. The only one of my arguments on which Professor Frisch dwells is, to me, the least important. It is not primarily because I like to play these sonatas in succession that I omit their first-movement repeats, although I maintain that, from the angle of the performer, this remains a splendid reason to avoid them.

I am most grateful to Professor Frisch for crediting me with the pursuit of scrupulous fidelity. Yet, there are occasions when to stick to the letter of Schubert’s text would mean to miss the point. Not only in Schubert, but also in Haydn and Mozart, certain repeat marks turn out to be question marks, calling for critical scrutiny. Nobody could admire Mozart’s “Jupiter” Symphony more than I do; a performance that “respects” all the repeats, however, turns sublime pleasure into an endurance test. And the fact, if it is one, that minuet repeats were once executed both the first time around and during the minuet’s recapitulation will not induce me to do so in 1989.

Playing the complete score has hardly been a nineteenth-century concern. Clara Schumann left out, in her performances of “Carnaval,” those pieces that alluded too directly to Robert and the “family,” while Anton Rubinstein, in his celebrated survey of “the history of piano music” in seven recitals (1885–1886), apparently piled up the following Beethoven sonatas, all in one evening: Opus 27/2, 31/2, 53, 57, 90, 101, 109, and 111. To be sure, without repeats. (Nevertheless, I would like to have been present.) What we encounter here is another manifestation of extreme gluttony: where, nowadays, “the whole cake” consists of repeated helpings of the same musical material, the late nineteenth century appears to have crammed as many different dishes as possible into the musical menu, and down the listener’s throat.

Next to such propositions, Dvorák’s suggestion—borne out by his activities as a Schubert conductor—to avoid making Schubert longer than necessary seems to me far from ignorant. I can only find it eminently sane.

This Issue

March 16, 1989