Before 1900 in Europe and America, it was at home that music was most often experienced, by family members who played some instrument or sang, and by, willingly or unwillingly, the rest of the family and friends. (In Western society among the lower middle class and upward, most music was made by women, who were generally expected to learn to cook, sew, and play the piano. The majority of professional musicians may have been male, like the majority of professional cooks, but most of the cooking and piano playing was the lot of women. Music, like breakfast and dinner, was part of life at home.) More exceptionally, music could be heard in some public places—concert hall, opera house, or church. The public realm was essentially a complement to the private. It set standards and added glamour.

By the twenty-first century, all this has changed. Both private and public music are being displaced by recordings. Few people make music themselves at home anymore. Because of more cramped living space, it is now inconvenient to house a piano, a once indispensable piece of furniture for any household with even modest pretensions to culture and the instrument that for more than a century was the mainstay of classical music. Outside the big cities, live public music is disappearing as well. Most of the smaller towns that used to have a classical concert series have lost that, and if they are too insignificant to sponsor a popular rock group event, their public music must be confined to clubs. Even live symphony and opera broadcasts have been largely eliminated. At home today we play records. Classical and pop radio stations play records. And often ballet companies and theatrical productions play records in place of hiring musicians.

Robert Philip’s Performing Music in the Age of Recording is a brilliant analysis of how this has affected performance style. It is also incidentally, for much of the time, the best account I know of how musical life in general has changed since the introduction of vinyl and long-playing records in the 1950s, which made it possible for records to invade everyone’s home. But it starts even further back with the end of the nineteenth century, when recording was invented by Thomas Edison, who recited “Mary Had a Little Lamb” into his new machine. The book is full of fascinating detail cogently presented on rehearsal practices and standards, recording on piano rolls, the different instruments used in orchestras, the way records are edited, and the contrasting musical ideals of performers. Philip is large-minded, tolerant, and sympathetic to various positions, and consistently judicious.

His main thesis is that recording has directed performance style into a search for greater precision and perfection, with a consequent loss of spontaneity and warmth. Various expressive devices once common in the early twentieth century have been almost outlawed: “portamento” (sliding from one note to another on a stringed instrument); playing the piano with the hands not quite together (Philip calls this dislocation); arpeggiating chords (not playing all the notes of the chord at the same time but one after another), and flexibility of tempo.

All this is largely true and fully documented by Philip. It needs shading, however. There are other forces at work in this change of approach to musical style that are independent of recording, and the development is more complex than it might appear. What I have to add here should be seen merely as footnotes to Philip’s book, and not as qualifying my high regard for his achievement.

An intimation of the greater complexity of the subject is suggested by Philip’s quotation from a letter of Johannes Brahms to his friend the violinist and conductor Joseph Joachim:

Joachim was to conduct an early performance of Brahms’s Fourth Symphony in Berlin in 1886, and he had written to the composer asking for more guidance about tempi. Brahms then sent Joachim a marked-up score of the symphony with a letter: “I have marked a few tempo modifications in the score with pencil. They may be useful, even necessary, for the first performance…. Such exaggerations are only necessary where a composition is unfamiliar to an orchestra or a soloist. In such a case I often cannot do enough pushing or slowing down to produce even approximately the passionate or serene effect I want. Once a work has become part of flesh and blood, then in my opinion nothing of that sort is justifiable any more.”

This does not mean that Brahms wanted a rigidly metronomical performance of one of his works when it became better known, but he believed that in the end his music would not benefit from exaggerated freedom of tempo and needed a sober, less emphatic, and more unified approach.1

A well-known story tells of Arturo Toscanini’s visit to Giuseppe Verdi to prepare the first performance of the Stabat Mater. The young conductor asked the aged composer for permission to make a slight ritardando (slowing down) at one point. “You have to make a ritardando there,” replied Verdi. “But you didn’t write one,” said Toscanini, and Verdi remarked, “Can you imagine what most conductors would do if I wrote one?”


It is obvious that the freedom of tempo so valued by Philip was both necessary and often disastrously abused. The reaction against it was not just caused by recording, although hearing a particularly unintelligent use of it on a disc that may be played over and over again may have reinforced the prejudice. An impression of a relative unity of tempo remained an ideal even after the advent of the Romantic movement. Beethoven once said to Anton Schindler (who was a pathological liar, but there is no reason to doubt his word on this matter) that the necessary changes of tempo in the largo slow movement of his Sonata Opus 10 no. 3 should be perceptible only to connoisseurs. Of course, variation of tempo was needed for expression (in Beethoven’s last piano sonatas, the direction espressivo is almost always accompanied by a ritenuto, a direction to slow the tempo slightly at once, or the espressivo is followed eight bars later2 by the direction a tempo, indicating a return to the normal tempo).

The string players’ old-fashioned device of sliding from one note to another, or portamento, has, as Philip remarks, a double purpose: “for deliberate effect and as a routine manner of changing position.” The loss of an expressive device, above all one inherent in a musical style, is always regrettable. The reaction against it, however, has a double root: badly used, portamento can have a somewhat soupy effect, making the violin sound as if it were not singing but crooning; and as an easy way of getting from one note to another it often adulterates the purity of pitch. As an expressive effect, however, it should certainly be restored. It is clearly suited not only to much music of the late nineteenth century but also earlier: in certain places in Chopin’s Cello Sonata and in Mendelssohn’s Violin Concerto, for example, it is essential to the phrasing. I doubt there is much to be said for it as a routine way of getting from one note to another now disdained by string players, but Philip has a soft spot in his heart for the practice on old symphonic records:

Just as slight roughnesses in ensemble make one aware of eighty individuals struggling to achieve unanimity, so the old-fashioned habit of portamento makes one aware of the physical process of individuals sliding up and down the fingerboards of stringed instruments…. The orderly, smoothly blended sound of the modern orchestra, in which everyone moves neatly and apparently effortlessly together, has become more machine-like, and less obviously human, than it used to be.

This overlooks the fact that portamento was banished for musical reasons, which may or may not be valid in particular cases. Philip loves recordings and yet, paradoxically, deplores the fact that they are made by machines. I would be willing to settle for a little less routine and more perfection, but Philip’s campaign for the restoration of portamento should be listened to.

A similar ambiguity attaches to the pianistic device of playing the left hand on the beat and the right hand just afterward, which Philip calls dislocation. As Philip knows, Mozart and his contemporaries called this rubato, and it was a Central European expressive form of decoration (when he was traveling in Italy, Mozart wrote to his father that the Italians were astonished when they heard him play rubato). The word “rubato” later acquired the meaning of any rhythmic irregularity, so Philip’s change of terms is a justifiable attempt at clarity. “Dislocation,” however, sounds slightly pejorative, while the word “rubato” has a more favorable effect. In the early twentieth century the practice was widespread, largely among pianists trained in Vienna, but was used in a more discriminating manner by the Russian school.

Dislocation has at least three purposes. It is originally derived, I believe, from an opera singer’s slight hesitation in producing an important and expressive note, as if he or she were momentarily overcome by emotion. It should be more generally recognized that a note can be given expressive quality and importance by making it appear not too easy to produce, for that is the unconscious logic behind the most traditional use of delaying its appearance. (One can see the powerful emotional effect of this kind of rubato when the principal theme is repeated in bars 86 to 87 of Mozart’s great Rondo in A minor.)

The origin of another purpose of dislocation can be traced to the sonority of the piano. Playing the bass note in the left hand before the melody note in the right allows the melody note to enter into an already prepared harmonic frame and also allows the bass string’s overtones or harmonics to be reactivated sympathetically when the right hand enters a split second later. This gives the melody note greater sonority and cantabile (or singing) quality. This was important for performances given in large public halls, an essential part of musical life that came into being only when Liszt invented the piano recital in the 1830s. The unremitting use of vibrato by string players, a modern innovation that Philip regrets with good reason, has a similar cause: the vibrato carries better and with greater intensity in large concert spaces.


The third purpose of dislocation, which comes into play when it is used systematically over a long passage, is to vary the texture by making it more lively: that is the way Mozart uses it in the return of part of the main theme in bar 19 of the slow movement ofhis Sonata in C minor. For the late-nineteenth-century pianist, the extensive use of dislocation throughout long passages or even whole pieces transformed the texture into something more fluid, less hard-edged, blunted the sharp edges of the rhythm, and made the atmosphere beautifully vague.

Since recordings did not exist before the late nineteenth century, we have no idea when the practice began of using dislocation almost systematically all the time. Some pianists early in the twentieth century featured it constantly. My childhood memory of Paderewski’s performance on the radio is that his employment of it was unrelenting. I also once heard, long ago, a beautifully poetic recording by Harold Bauer of Schumann’s Des Abends (“In the Evening”) in which, as I remember, his two hands never coincided even once; it took getting used to, but it gave an aptly soft and misty quality to the music. When the use of it was unremitting, however, in music from Bach to Debussy, it could become self-defeating, imposing a similar expressive style on every passage in every work. Worse, it was an easy and cheap way of sounding expressive, a cuisine of emotion as if one could smear sentiment thoughtlessly over everything like goose fat without regard for the intrinsic differences of individual phrases. That, I think, partly accounts for the reaction against it.

The reaction started very early. After transcribing the judgment of several critics who consider Josef Hofmann the finest pianist of the last century, Philip writes:

Hofmann’s recordings now sound “modern” in one sense, which is that they contain hardly any trace of the old-fashioned separation of bass and treble (as early as 1909 he advised amateur pianists, “This ‘limping,’ as it is called, is the worst habit you can have in piano playing…”).

This is one of the rare places in which Philip, whose accuracy in writing about performance is unequaled, is slightly misleading. Hofmann did use what he called limping not only for an occasional expressive effect for a note or two, but even more extensively. For example, in a recording of Chopin’s Waltz in C-sharp minor, Opus 63, Hofmann plays the beginning and ending sections very soberly with his hands always together, but when he comes to the slower and more lyrical episode in D-flat Major, the bass is always slightly in advance of the right hand, and this invests the section with a sonority that is less hard-edged and more relaxed, more poetic. Hofmann is following here the older tradition, in which the dislocation is not used throughout a piece but is a special effect intended to distinguish and set in relief a particular episode. Chopin uses the word “rubato” very rarely as a direction to the pianist in his works, but when he does, it almost always refers to the reappearance of a theme already played, and I think he intends by this the same dislocation that Mozart called rubato. (When Chopin wanted the kind of rubato that is a gentle lingering over a few of the notes, he indicated this by writing dots over each of the notes under a slur.3 )

The admiration for Hofmann was very great indeed when I was a child, and I went to hear him almost every year until his last recital in 1946. I am told that he and Paderewski were the only pianists paid a large sum by Steinway each time they played a Steinway in public, and that Rachmaninoff was jealous that he was not so remunerated. We learn from Philip that Horowitz revered Hofmann and imitated him (although he never attained the absolute evenness of scale passages that Hofmann commanded). He might have added that Rachmaninoff, who called Hofmann the greatest living pianist, dedicated his Third Piano Concerto to him (but Hofmann never played it—“Too many notes,” he said4 ), and, on hearing Hofmann’s interpretation of Chopin’s Sonata in B minor, remarked, “Another piece I must strike from my repertoire.”

Hofmann recorded extensively (he is said to have made the historic first recording of a piano while sitting in Edison’s lap) but refused to release any records after 1925. Philip speaks only of the recordings of his public concerts, none of which have the satisfactory quality of the Brunswick company’s recordings of 1922 and 1923. Hofmann was interested in engineering, and his pre-electrical recordings sound much better than almost any of the early electrical records made a few years later.

Philip claims that

in public recital, where, according to those who heard him, the “real” Hofmann was to be heard, he took liberties which now sound anything but modern.

It is unclear to me why the interpretations that Hofmann wished to hand down to posterity in his studio recordings are less “real” or even less representative than the few taped live performances, many of which are, in fact, unselective and unrepresentative. The recordings he authorized are extraordinary in their range of tone color and nuance. It is true that Hofmann’s performances in public were capricious on occasion (particularly in his later years he found giving concerts a chore, and I remember that at his last recital in 1946 he came on the stage with the program in his hand, which he placed on the piano so that he could glance at it and see what he was to play next). But most of the performances were hardly wayward. I had austere tastes as a teenager and I would have morally disapproved of an overly eccentric playing of the great Rondo in A minor by Mozart, which I remember as exquisite. I do recall a Beethoven Sonata in D minor, Opus 31 no. 2, that was flippant, and a rendition of the Nocturne in G Major by Chopin in which the difficult arabesques in thirds and sixths were taken at a pace that was hair-raising. (“Why is he playing it at that tempo,” I asked a friend of Hofmann’s with whom I was sitting, and he replied, “He can’t play it any faster.”5 )

On records, to appreciate Hofmann’s art, one should listen to the two sonatas of Scarlatti, the C-sharp minor waltz of Chopin mentioned above, Brahms’s arrangement of the gavotte from Gluck’s Alceste, and the Second Hungarian Rhapsody of Liszt.6 The interpretation of the last is certainly out of style, since Liszt directs the opening movement to be played a capriccio (that is, freely), and Hofmann plays in a strict, almost metronomic, tempo, but that makes the few slight rhythmic freedoms stand out with great effect and endows the rhapsody, which can seem hackneyed, with great nobility. Essentially all these works are encore pieces, but that is pretty much all that was made permanent in those early days of recording. The large works that one heard from Hofmann, Schumann’s Faschingsschwank aus Wien, Beethoven’s Sonata Opus 31 no. 3 in E-flat Major, Liszt’s great fantasy on themes from Don Giovanni—these were never put on discs.

My teacher, Moriz Rosenthal, who had studied with a pupil of Chopin and later with Liszt, was famous above all for his playing of large works like the Brahms Paganini Variations and the Schumann Fantasie, but he recorded only short pieces of Chopin and his own virtuoso transcriptions of Strauss waltzes. Except for one recital, he no longer played in public when I knew him, so I never heard him play these works in concert, never experienced his art at its grandest, although he played passages for me. He frequently used the dislocation Philip writes about, as did his wife, Hedwig Kanner, with whom I also studied—even more so in her case, since she was a pupil of the celebrated Viennese teacher Teodor Leszetycki, who used this procedure—but they did not try to impose it on me. Rosenthal was, in fact, for a famous pianist, absurdly courteous to an eleven-year-old. He never said I was wrong but remarked only, “I have a different idea of this piece,” and went to the piano to demonstrate. I was already used to the style of Serkin, Rubinstein, Hofmann, and Schnabel, where dislocation was little employed.

In more recent years, I have used it on occasion (a record producer tried once to correct a couple of places where my hands were not together); I do not think I was principally influenced by childhood memories, but by further study of nineteenth-century repertory. Musical genealogies are not much use: one reacts against one’s teachers as much as one learns from them. What I received from Rosenthal was a sense of phrasing and, already at my first lesson through watching him play a few phrases, a realization that one could adjust tone quality with magical results for the ear, giving each note of a chord different weight, but without any aspect of the movement of the hands visible to the naked eye that would explain how it was done. I found the elegance of his phrasing of the Chopin mazurkas a marvel that I was afraid to imitate until many years later.


One of the most fascinating sections of Philip’s book is his discussion of how early-twentieth-century composers—Mahler, Stravinsky, Elgar, and others—recorded their works or supervised performances by other musicians: he compares the results with the original directions given by the composers in the text and with recordings by later performers. He understands that there is no single answer to the question whether the composers really approved of performances that were sometimes wildly at variance from the text, or whether they were settling for the best they could get—or whether, indeed, they had changed their minds about the works as so often happens.

The recordings of Bela Bartók are a special case. He was not only a major composer who helped to revolutionize musical style, but a great pianist. Philip, accurate as ever, writes that “his playing sounds surprisingly ‘Romantic’ in its freedoms,” and contrasts his playing of his own music with that of modern pianists who play it “with a more percussive edge, and treat his rhythms and tempos more strictly, than Bartók did himself.” It is certainly true that Bartók’s own performances of his compositions have a relaxed grace that was lost in the work of the younger pianists who took up his music. However, in the composer’s playing there are many details of the score that are either missing or make no effect, not merely details of rhythm and tempo, as Philip points out, but of accent and dynamics as well.

The fact is, Bartók was both a composer who helped to revolutionize the music of the early twentieth century and a traditional pianist in an old-fashioned style. With radical changes of style, it takes more than a decade for performing musicians to catch up and find an adequate way of rendering the new. We can trace this process in recordings of Stravinsky’s music, in which what first sounded awkward and unconvincing was later performed with greater ease and more warmth. The trick, as always, is to find a form of expression in performance that is adequate to the new. Stravinsky’s horror when he heard Koussevitzky’s direction of his Symphonics of Wind Instruments, for example, was caused by the conductor’s imposition of the clichés of Romantic performance. It provoked the composer’s absurd claim for decades that his music was not expressive in his effort to avoid the peril of irrelevant expression.

The discrepancies between Bartók’s text and his playing are not simply explained: first, he expected his music to be performed freely and with the grace and warmth that characterized his own playing; second, his own training as a pianist did not always allow him to cope fully with the originality of his conceptions as a composer. The pianists who followed him were more accurate but lacked the grace he brought to his music. (Among them was Leonid Hambro, who should be famous for having discovered how to play Bartók’s energetic glissandos on the black keys without stripping the skin off his fingers—he used his wallet.) More recent pianists, however, Andras Schiff in particular, have both recovered much of Bartók’s graceful quality and retained the greater fidelity to the text. Do we want to hear what Bartók played or what he wrote? Both, if possible, and at the same time.7

Some years ago, before performing Elliott Carter’s piano concerto, I played a few passages for the composer, and asked him if they were fast enough. “They sound almost too fast,” he replied. “I’ve reduced your metronome marks by 20 percent throughout,” I remarked. In the performances I had heard, the passages most difficult to play or hear had been slowed, while the easier ones were played up to the tempo indicated, and the result was to homogenize the textures and eliminate the contrasts. It is important to keep the relationship of different tempi in Carter’s work, but the general tempo of the entire piece may be varied by the interpreters. Composers may often badly estimate what would be an effective tempo, but the inner proportions of each work are integral to its conception, to its logic. Composers expect their music to be interpreted with the kind of freedom that makes sense of it, not pedantically or with the freedom that comes from the routine imposition of an earlier style.

Philip contrasts the “rhythmic freedom and lightness” and the way “the phrasing breathes” of the Adolf Busch Chamber Players’ recordings of Bach’s Brandenburg Concertos with the irritating performances of the early music movement in its initial dogmatic fervor (which has shifted recently in the direction of sweet reason), and claims that the splendor of Busch’s approach

has little to do with Bach scholarship…. Overall, one has the impression not of a scholar who has researched the way things were done in Bach’s day, but of a player with a natural gift for rhythm and phrasing.

This is not quite true. Busch certainly thought he was being scholarly, and was trying to return to the text exactly as it stood and remove as inauthentic the excrescences of heavy sonority and exaggerated dynamics that had gradually developed in the performance of Bach. In its way it was a successful return to a distant past, even if, like all Golden Ages, the past was partly a constructed illusion. But only partly: as in most revolutions, it was also a real return to the past, as real as Philip’s revaluation of old stylistic habits. This adherence to the text was essential to Busch, and to his brother, Fritz Busch, whose Glyndebourne recordings of Mozart Philip admires; that is why Fritz Busch mistakenly never allowed the necessary ornamentation at the cadences in Mozart’s recitatives because they were not written down, the one aesthetic blot on those records along with the singing of Audrey Mildmay, who was married to the owner of Glyndebourne.

The changes in musical performance registered by Philip were part of a larger movement in all the arts, in the paring away of the clichés of Romantic excess. Listening to a recording of Sarah Bernhardt intoning Racine makes one laugh today. Nevertheless, as in music, the turn to a less mannered style of performance, a more informal way of speaking the verse, has resulted in a loss of poetry, a loss that occurs often enough at the Comédie Française. When Racine’s verse is read in the way that colloquial French is spoken, it no longer scans, and much of the expressive force drains away. The absurd Bernhardt manner had its virtues. Performing Shakespeare so that his language sounds more like informal everyday speech is equally disastrous. Yet we have no reason to believe that a return to the late-nineteenth-century manner—even though it pre-served essential aspects of spoken dramatic verse—would reproduce the way Shakespeare or Racine was spoken in his day.

The purpose of recording changed radically during the twentieth century. At first, on the four and a half minutes available on 78 rpm records, there was time only for short pieces, mostly encore pieces, as we have seen. More ambitious projects were then started involving sets of records: thirty-two Beethoven sonatas played by Artur Schnabel, the first act of Die Walküre with Lotte Lehmann, conducted by Bruno Walter. When all the Beethoven quartets and all the Haydn symphonies were put on records, it was no longer a performance that was being recorded, but a body of music. The emphasis had shifted. Performance is properly ephemeral; musical works endure.

Occasional mistakes in performance do not matter that much in the concert hall, but they are always hard to stomach on a record when one is listening to it for the tenth time. I wince in advance at the very few wrong notes in the scherzo of Rachmaninoff’s magnificent performance of Chopin’s B-flat minor Sonata, or at the place in Horowitz’s first recording of Rachmaninoff’s Third Piano Concerto where he skips two bars and has to keep repeating a figure until the orchestra catches up. (Philip tells us that Rachmaninoff once needed twenty-one takes of a short piece in order to get a performance he would release. Making another take is oddly considered more honest than splicing. When splicing mistakes out on tape became possible, Rudolf Serkin would still begin all over again if he hit a wrong note on the last page, but he changed his policy later in his career.)

When a recording is intended to be a renewable image of the music rather than the capture of an individual performance, then even eccentric details become less desirable. A sudden rhythmic hurrying of the second theme by Schnabel in Mozart’s Concerto in C Major, K. 467, was interesting and effective when I first heard it; now I wait for it come and it is an irritant. Philip is right to maintain that Serkin’s records are less interesting, less spontaneous than his best concerts. The same thing is true of Horowitz: his recording of Prokofiev’s Seventh Sonata is considerably more prudent than his live renditions. But if a record is to be an adequate representation of the music, to be heard not once but many times over, some of this loss is inevitable and even right. The concert, imperfect as it may be, has compensations that no record can bring: a competent live performance of the introduction to the last movement of Mozart’s G minor string quintet brings tears to my eyes. I have never wept at a record.

There are, however, enough recordings where precision and a sober lack of mannerism reach a state of grace. Among them are Schnabel’s recording of Mozart’s Rondo in A minor, Kreisler’s performance of the Brahms Violin Concerto (with Leo Blech, not the one with Barbirolli), Solomon’s Prelude and Fugue in A minor by Bach arranged by Liszt, Frieda Leider’s Immolation scene from Götterdämmerung, Rosa Ponselle and Giovanni Martinelli in the last scene of Aida, Toscanini’s performance of Debussy’s Ibéria. Everyone can make his own list.

Philip’s book is part of an ongoing movement. Young Italian pianists, patriotically inspired by old recordings of Ferrucio Busoni (who was, however, more Germanic than Italian in cultural outlook), are enthusiastically applying the old devices of dislocation and arpeggiation with some benefit to Romantic music (and some comic effect in Bach and Beethoven). Perhaps the best thing about his book is the way it makes us appreciate the heritage that recording has given us and forces us to recognize the limitations of our taste by an understanding of what has gone out of fashion. He allows us to estimate how much we have lost in our rejections.

This Issue

November 3, 2005