You are on page 1of 396

Off the Record

This page intentionally left blank

Off the Record
Performing Practices in Romantic Piano Playing


Oxford University Press, Inc., publishes works that further
Oxford University’s objective of excellence
in research, scholarship, and education.

Oxford New York

Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto

With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam

Copyright © 2012 by Oxford University Press

Published by Oxford University Press, Inc.

198 Madison Avenue, New York, New York 10016

Oxford is a registered trademark of Oxford University Press

All rights reserved. No part of this publication may be reproduced,

stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise,
without the prior permission of Oxford University Press.

Library of Congress Cataloging-in-Publication Data

Peres Da Costa, Neal.

Off the record : performing practices in romantic piano playing / Neal Peres da Costa.
p. cm.
Includes bibliographical references and index.
ISBN 978-0-19-538691-2
1. Piano—Performance—History—19th century. 2. Sound recordings—History—19th century.
I. Title.
ML706.P37 2012
786.2'14309—dc23 2011023579

9 8 7 6 5 4 3 2 1
Printed in the United States of America
on acid-free paper
To Clive Brown, without whose inspiration and guidance
I would never have undertaken this.
This page intentionally left blank

Foreword ix
Preface xv
Acknowledgments xvii
About the Companion Website xix
Introduction xxi

1. Early Recordings: Their Value as Evidence 3

2. Playing One Hand after the Other: Dislocation 41

3. Unnotated Arpeggiation 101

4. Metrical Rubato and Other Forms of Rhythmic Alteration 189

5. Tempo Modification 251

Conclusion 309
Bibliography 311
Discography 323
Index 327
This page intentionally left blank

This is a book that should be read not only by every pianist who aspires to perform
the music of the late eighteenth and nineteenth centuries on period instruments
in a historically informed manner, but also by every pianist who performs this
repertoire at all. The only pianists who may perhaps be excused from reading it are
those who, for whatever philosophical or pragmatic reason, have made a deliber-
ate and informed choice to take the notation at face value, knowing that by doing
so they ignore much that the composer may have expected it to convey to the
performer. Even such hardened modernists, however, might find it difficult to
resist the persuasive evidence presented in this book. Any open-minded musi-
cians who objectively evaluate the weight of documentary and recorded material
so cogently assembled and examined in the following chapters will be forced to
reconsider their approach to performing nineteenth-century piano repertoire.
The training of musicians during much of the twentieth century was predi-
cated, as it still generally is today, on an assumption that the notation of post-
Baroque music means precisely what it appears to mean. During the first half of
the twentieth century, players and teachers were insistent upon the rejection of
the “bad old habits” of earlier performers, who were condemned for taking unwar-
ranted liberties with the composer’s notation. The first half of the twentieth cen-
tury saw the establishment of a new aesthetic, in which conscientious observation
of the composer’s notation was a fundamental rule of good taste. When I learned
the violin in the late 1950s and 1960s, therefore, we were drilled to observe note
values strictly and to regard the composer’s dynamic and articulation markings,
interpreted according to contemporary understanding, as sacrosanct. Tempo was
now to be maintained steadily, in contrast to the performance style of the older
generation and accelerating in particular, for instance during a crescendo or in
passionate passages, was strongly discouraged. More advanced teachers began to
insist upon the use of urtext editions where these were available, or, where they
were not, to strike out or alter almost all the performance markings and instruc-
tions provided by earlier editors. We were trained always to change position
cleanly so as to avoid the sin of tasteless “inauthentic” portamento. When playing


Peres da Costa-FM.indd ix 3/13/2012 2:16:29 PM

x Foreword

in ensemble, absolutely precise vertical coordination was a sine qua non. On the
other hand, we were encouraged to employ continuous vibrato, conspicuously
absent from the notation, in all repertoires because it was seen not as an improvi-
satory ornamental effect but a basic component of beautiful tone (non vibrato
was to be used only very sparingly as a special effect). We also picked up a habit of
employing rallentando, where nothing was indicated in the score, to emphasize
expressive moments, and this was often followed by the kind of slight articulation
commonly referred to by the German term luft pausa.
These and other practices, which are typical of the orthodox performance style
of the present day but were largely uncharacteristic of earlier periods, were the
product of several factors that radically transformed musical performance in
the first half of the twentieth century. One of the most powerful of these was
recording, which made musicians self-conscious about discrepancies between the
musical notation and what they actually played. In the concert hall, deviations
from the strict letter of the notated score, which differ from performance to per-
formance, dissipate as soon as they have been heard; but in a recording, these
creative nuances and gestures become fossilized, resulting in the loss of their
essential spontaneity. As techniques of mechanical reproduction improved and
recordings assumed an increasingly important role in the dissemination of music,
recording artists, aware that they would be permanently represented by a single
performance, became ever more concerned to avoid the potential charge of
disrespect for the composer’s text, particularly with respect to note values, tempo
flexibility, and vertical precision.
This gradual change of attitude was encouraged and reinforced by what might
be described as the urtext mentality. The concept of the urtext evolved gradually
during the nineteenth century, partly in connection with the work of editors
involved in the revival of earlier music and partly through those who contributed
toward the production of collected editions of the works of more recent compos-
ers whose achievements were seen as justifying their admission into the first rank
of the musical Parnassus. From a practical point of view, such editions were
immensely valuable, for they made many works available for the first time in
print, or replaced early editions, which were often highly inaccurate, with a more
reliable text. By the end of the nineteenth century, a vast corpus of music from
the sixteenth to nineteenth centuries had been enshrined in such monumental
collected editions. The works of the majority of these canonical composers
have subsequently been reedited, or are in the process of being reedited, with
much more systematic scholarly scrutiny and documentation of the sources.
They have thus been invested with even greater authority as the repository of the
composer’s notational intentions.
In practice, of course, the situation is rarely as clear-cut as it seems at first
sight. The editing process has generally been based on a presumption that there is
a single definitive version of the work, which represents the composer’s final
thoughts (the Fassung letzter Hand), and primacy has tended to be given to full
Foreword xi

scores, especially autographs; copies and performance material have tended to be

relegated to a subsidiary position, although these may often represent more
closely what composers and their contemporaries actually played. Notwithstanding
these caveats, critical editions, which provide musicologists and scholars with a
painstakingly researched text, have immense value as a scholarly resource; as
texts for performers, however, they have immense dangers. The performer is very
easily beguiled into thinking that such nice, clean, and apparently authoritative
editions definitively embody the composer’s intentions not only in respect of
notation but also of musical execution.
This misunderstanding has sometimes led to bizarre results. Many conductors
and singers during the twentieth century, for instance, insisted that recitatives, in
which the final syllables of a phrase were generally notated as a pair of notes
(strong–weak) at the same pitch, should be sung exactly as they were written.
They paid no heed to extensive eighteenth-century documentation that this was
not the intended execution and also ignored the practice of nineteenth-century
singers (preserved in early recordings) who had faithfully preserved the old con-
ventions, assuming that their deviation from the notation was just another of the
supposedly corrupt practices they attributed to the older generation. This is an
extreme example of the ways in which the implicit meaning of the notation fell
victim to an unhistorical conviction that fidelity to the composer’s intentions
required the most scrupulous literal observance of the notated text. Another
manifestation of the fallacious notion that composers of the late eighteenth
and nineteenth centuries intended their notation to convey precisely what the
executant was intended to play was the controversy over Mozart’s staccato marks.
Despite the obvious inconsistency of his autographs in this respect, it was felt
by numerous editors and musicologists that the two extremes of dot and stroke
(many of the marks were ambiguous) must represent two different styles of
execution. In the Neue Mozart Ausgabe, editors were obliged to make questionable
decisions about reproducing the very variable staccato marks in Mozart’s auto-
graphs with one of two printed signs (dots or strokes). Inevitably, modern per-
formers, schooled to believe in a simplistic consequential relationship between
fidelity to the notation and to the composer’s intentions, attempted to make
distinctions that were never envisaged. Ironically, this unhistorical attitude seems
to have been most strongly encouraged by musicians who sought to establish
their credibility as historically informed performers.
Over the past couple of decades, it has become increasingly clear to scholars
of performing practice that the notation of Classical and Romantic music is mis-
leading to us in many ways: indeed, that it often meant quite different things to
composers and the performers of their day. This is particularly true of the late
eighteenth and early nineteenth centuries. The accounts of the qualities that char-
acterize an artistic performance, and of the ways in which these can be acquired,
that occur in the final sections of Hummel’s Anweisung zum Piano-Forte-Spiel
of 1828 and Spohr’s Violinschule of 1833 leave no doubt that although a literal
xii Foreword

performance of the notated music can be regarded as “correct” (richtig), it falls far
short of being a “fine” or “beautiful” (schön) one. The idea that with the passage of
time notation became more closely representative of the execution envisaged by
the composer is, however, not incorrect. In general, later nineteenth- and early-
twentieth-century composers took great pains to provide the executant with more
extensive information about dynamics, accentuation, and expression.
Composers also made it clear with greater frequency that they disapproved of
some of the liberties taken by performers. Mendelssohn’s stricture, related by
Joachim, that in performing the music of great composers “it is inartistic, nay
barbaric, to alter anything they have ever written, even by a single note,”1 may
encourage us to believe that by the 1840s composers were already beginning to
see a straightforward correlation between notation and execution. Yet we should
be careful not to jump to conclusions on the basis of such remarks. It is much
more likely that Mendelssohn’s comment was directed at such individualistic vir-
tuosos as Franz Liszt, who, in performances of Beethoven’s piano sonatas for
instance, had the temerity to alter the harmony and to elaborate the text in ways
that undoubtedly exceeded the composer’s expectations. Similar remarks were
made by composers as different as Wagner and Verdi, but in these cases, it is clear
that their statements did not convey the same things in the musical world of the
nineteenth century as they do in ours, for both these composers either sanctioned
or tolerated performing practices in their own works that appear to contradict
their injunctions.2 Nevertheless, by the latter years of the nineteenth century,
with the growing internationalization of musical culture, many younger musi-
cians were coming to rely increasingly on notation rather than on directly trans-
mitted conventions of performance, especially when performing older music that
did not belong to the tradition in which they were trained. Thus Joseph Joachim
remarked about the playing of his Belgian contemporary Henri Vieuxtemps that
he “like the majority of violinists of the Franco-Belgian tendency sticks too much
to the lifeless note-heads when performing the classics, not understanding how to
read between the lines.”3

Andreas Moser, Joseph Joachim: A Biography, trans. Lilla Durham (London: P. Wellby, 1901), 46.
See also Clive Brown A Portrait of Mendelssohn (New Haven, Conn., and London: Yale University Press,
2003), 223.
See Clive Brown, “Performing Practice,” in Wagner in Performance, ed. Barry Millington and
Stewart Spencer (New Haven, Conn., and London: Yale University Press, 1992), 99–119; and Clive
Brown “On Exactly What Is Written,” in Verdi in Performance, ed. Alison Latham and Roger Parker
(Oxford: Oxford University Press, 2001), 81–83.
My translation from Andreas Moser, Joseph Joachim, ein Lebensbild, 2nd ed. (Berlin: 1908–10),
ii, 292: “weil er [Vieuxtemps] sich, wie übrigens in neuerer Zeit die meisten Geiger der französisch-
belgischen Richtung, beim Vortrag der Klassiker zu sehr an die leblosen Notenköpfe hielt, nicht
zwischen den Zeilen zu lessen verstand.”
Foreword xiii

Of course, once we have accepted that notation was intended to convey some-
thing quite different to musicians of the nineteenth century than it does to us
today, we still face the problem of deciding whether or how this knowledge should
inform our performances of particular repertoire. At one extreme of the spec-
trum, we may choose the cautious path of fidelity to the notation combined
with the conventional expressive language of the present day; at the other, we
may run the risk of unwarranted or tasteless license deriving from inadequate
understanding of the historical evidence. The solution to navigating a safe passage
between this Scylla and Charybdis of performing practice lies in books such as
this. The importance of doing so can scarcely be doubted by any musician who
cares about the great music of the past and aspires to enter more effectively into
the composer’s thoughts so that modern audiences may once again enjoy the
spontaneity, vivacity, and immediacy of late eighteenth- and nineteenth-century
music making.
The evidence so carefully collected and cogently presented by Neal Peres Da
Costa clearly demonstrates the significant difference, even in the works of later
nineteenth-century composers, between what was written down on the page and
what cultivated musicians, who in many cases were composers performing their
own music, actually played. In relation to keyboard practice, he has assembled and
analyzed an impressive array of documentary and recorded material, which dem-
onstrates beyond any reasonable doubt that notation was not understood by com-
posers or their contemporaries in the way we understand it today. Much with
respect to nuances of rhythm, the precise vertical placement of notes with rela-
tion to one another and the use of various kinds of tempo rubato was left to the
discretion of the experienced performer who knew how to transform a “correct”
performance into a “beautiful” one. As well as showing that such practices were
characteristic of the performance style of some of the most celebrated composers
and pianists of the nineteenth century, he has demonstrated how many of them
relate back to earlier periods. Furthermore, the persuasive evidence he provides to
elucidate the musical motivation that lay behind the employment of these unno-
tated practices in particular contexts will be particularly helpful to performers.
This book, with its careful and judicious interrogation of aural and documentary
sources, will be an immensely valuable resource for those keyboard players who
want to penetrate beyond the bare notes of the urtext and learn how to read
between the lines.

—Clive Brown
This page intentionally left blank

This book has been several years in the making. It is a fairly complete revision of
my PhD dissertation Performing Practices in Late-Nineteenth-Century Piano Playing:
Implications of the Relationship between Written Texts and Early Recordings
(University of Leeds, 2002). The research toward it really started around 1982
when I commenced my undergraduate studies at the Music Department of the
University of Sydney, though of course at that stage I had no idea that I would end
up writing a book. I really owe my love of detective work in the field of performing
practice to Professor Winsome Evans, who opened my eyes to the wealth and
value of historical written texts and first introduced me to the harpsichord and
fortepiano. My postgraduate study took me to the United Kingdom, where I wit-
nessed the very latest in what has now come to be known as historically informed
performance. For me, it was thrilling to hear not only early-ish repertoire (Purcell,
Bach, and Mozart) but also, for the first time, music of later composers (Schubert,
Berlioz, Schumann, and even Brahms) supposedly in period or authentic style.
I had the great fortune to join forces with some extremely talented musicians in
the field. We formed a period-instrument ensemble, and gave performances of
seventeenth- and eighteenth-century chamber and orchestral repertoire all over
the world and we recorded many award-winning CDs. Our delivery of the music,
to judge from reactions, was very successful. Audiences loved our exuberance and
our style. But what style were we emulating, and how authentic was it? Was it
even possible to be authentic? I was really faced with these questions when, during
the initial stages of my PhD, I listened to some piano recordings and piano rolls
made at the turn of the twentieth century by recognized nineteenth-century vir-
tuosi. I was completely shocked by what I heard. The style bore little resemblance
to anything I was used to. As time went on, I realized that there is a significant
gulf between how we play Classical and Romantic piano music now and how it was
played just over one hundred years ago. But perhaps more surprising to me was
the fact that often what these nineteenth-century virtuosi advised in pedagogical
texts did not match up with how they actually played. I became keenly interested


Peres da Costa-FM.indd xv 3/13/2012 2:16:29 PM

xvi P re fac e

in the implications this had for the interpretation of written texts where no
comparison could be made with audible evidence. This book explores these impli-
cations while discussing many fascinating early recorded examples (available on
the companion website) and providing historical background to some of the
expressive practices that can be heard.

Of course, there are many people to thank for nurturing me through all the stages
of research and writing. First and foremost, I would like to thank Clive Brown,
who encouraged me to undertake a PhD under his supervision and who has con-
tinued to inspire and guide me through every step of the way. Though we now live
on opposite sides of the globe, his invaluable advice and friendship have helped
me get to the finish line. I am extremely grateful for his generosity in providing
the Foreword to this book and humbled by the authority expressed within it.
Equal thanks go to my loving partner and wonderful musician Daniel Yeadon,
without whose unwavering support, advice, cooking, and patience I could not
have done this. His professional editing skills have certainly made my life easier.
Our long-term chamber music partnership has enabled much experimentation in
the area of nineteenth-century performing practice and has helped shape many
of my musical instincts. The hugest thanks go to my dear friend and research
student Megan Lang, who has given me such positive support as my research
assistant. I am extremely grateful to her for her dedication and patience; for the
hours she spent in the acquisition of permissions for the publication of recorded
extracts, in making the majority of the recorded excerpts for the companion web-
site, and in reading drafts and providing me with invaluable criticism. I am also
greatly indebted to my dear friend Robert Dyball, who reformatted many of the
musical examples taken from my PhD and made many new ones as required. His
professionalism and generosity have given me the much-needed space to get on
with other aspects of the book. Heartfelt thanks goes to Robin Wilson for reading
drafts and providing invaluable comments and for opening up the space to exper-
iment in our performances of Brahms. His unbridled enthusiasm for the subject
matter of this book has spurred me on at many crucial stages. Very special thanks
go to Denis Hall for providing me with invaluable advice and comments and for
the use of transfers of many reproducing piano rolls in his collection. Thanks also
to Rex Lawson for his invaluable advice. Special thanks also go to the various
recording companies that have generously allowed the use of recordings. Thanks
also to Nina Platts who gave so generously of her time and editing skills in the
final stages of my PhD.

Peres da Costa-FM.indd xvii 3/13/2012 2:16:29 PM

xviii Acknowledg ments

I would like to thank the Sydney Conservatorium of Music and its previous
Dean Professor Kim Walker for providing financial assistance through a
New Frontiers Grant, as well as Special Study Project leave to undertake necessary
research and writing. Thanks to all my colleagues who have encouraged me along
the way. Thanks also go to my students, who have taken a keen interest in early
recordings. It has been a joy to witness the light go on when they realize that there
are many more choices available in performance. And of course, warm thanks go
to Suzanne Ryan at Oxford University Press for taking such an interest in my
initial proposal and for her patient enthusiasm through the various stages from
contract to publication. Thanks also go to several other people at OUP: Caelyn
Cobb for advice at the final manuscript stage, and for work on the book cover
design and the Companion Website; production editor Erica Woods Tucker; and
copy editor Mary Anne Shahidi. Special thanks also to Colleen Dunham for making
the Index.
Finally, I’d like to thank my parents, George and Cynthia Peres Da Costa, for
their love and for supporting my interest in music in the first place.

The companion website to this book contains audio extracts from noncommercial
and commercial transfers of many sound recordings made in the acoustic era and
later, as well as reproducing piano rolls. In each of chapters 2, 3, 4, and 5, readers
are prompted by this symbol to listen to the extract of the particular perfor-
mance under discussion or analysis. Full reference details for all audio extracts are
available in the Discography.
The companion website also contains a selection of figures that are either too
large to include in the book or are an appendix to the discussion. In each of chap-
ters 2, 3, 4, and 5, readers are prompted by this symbol to view the particular
figure on the companion website.
Readers may access the companion website using username Music3 and
password Book3234. Please note that these are case-sensitive.


Peres da Costa-FM.indd xix 3/13/2012 2:16:29 PM

This page intentionally left blank

Rubinstein . . . was a great stickler for the printed notes and

annotations—but he was so only in his teaching, not in his playing.
When he played, he played “Rubinstein,” whether the piece was by
Bach, Beethoven or Chopin; his intense personality broke through
all barriers of indicative annotations. Though everything sounded
as if it were composed by himself, no one could contain control
over one’s cool, critical faculties because—no matter what he
played—he always delivered a consummate work of art, for there
was so much of impressive beauty in his style of playing as to make
even the most critical auditor forget all about the “composer’s
style,” or the “code of art.” Or the much spoken of but never estab-
lished “eternal laws of aesthetics” and to lose himself in a sea of
beauty both sensuous and emotional.
Constantin von Sternberg (1852–1924)1

The notion of authenticity in performance was first introduced to me in 1982. It

was at the start of my undergraduate music degree, and I quickly became seduced
by the buzz term performing practice. If I did my research, I could find what I
needed to replicate a performance that would be recognizable to the composer if
he or she were alive. I was introduced to the harpsichord and the fortepiano,
sound sources so different from the modern piano. I explored a wider range of
repertoire. I collected the latest recordings from Europe—Bach Cantatas and
Mozart Symphonies played on original or period instruments. The subject of an
original voice was a hot potato indeed. I sat around with fellow music students
discussing with feverish passion what we were hearing. It was all so new and so
different from what we were used to. Rumor was that the period instrument

Constantin von Sternberg, “Keyboard Masters of Other Years,” The Etude (Oct. 1920): 657.


Peres da Costa-FM.indd xxi 3/13/2012 2:16:29 PM

xxii Introduc tion

movement—as it was then called—would soon tackle Beethoven, Berlioz, and

even Wagner. What would this sound like, we wondered? What could be so differ-
ent between how we performed Romantic repertoire now and how it sounded at
the first performance?
Then one day something stopped me in my tracks and made me ponder those
questions more deeply. During a rehearsal of the Sonata No. 3 for violin and piano
in D Minor Op. 108 by Johannes Brahms (1833–97), my violinist colleague asked
me whether I knew of the recordings that Joseph Joachim (1831–1907) made in
the early twentieth century. Though I had never heard of him, I was assured that
Brahms loved Joachim’s style of violin playing. Yet I was astonished to hear that
he used vibrato very sparingly. “How is that possible?” I asked. Nowadays, all
whiz-bang violinists play Romantic music with a huge rich vibrato. It’s the done
thing. Surely there could never have been any other way? I did not then have the
wherewithal to find Joachim’s recordings, and the Internet was as yet unheard of.
But this incredible information planted the first seed for this book. It set me
wondering how different style and expectations might have been, even as recently
as the late nineteenth century.
Some years later I had the very great fortune to meet Clive Brown. He had
already spent many years thinking about this question. He opened my eyes to the
fact that there were quite a few sound recordings of famous nineteenth-century
instrumentalists and singers. He encouraged me to find them and to listen to
them carefully. Not many people were taking them seriously. At first I was shocked
by what I heard: often, the performances bore only an outward resemblance to the
printed edition of the music. Either these supposed virtuosi of the period were
disrespectful charlatans or there were many things missing in the musical nota-
tion. Like most, I had been willing to accept that earlier composers, even Bach and
Mozart, may not always have notated what they expected. But with Romantic
composers, it was a given that everything they wanted was on the page. The score
was the final word. What I had taken for granted was about to be turned on
its head.
I now understand that irrespective of the era, written texts—musical notation
and verbal advice—are imperfect in preserving performing practices of the
past. Several writers have commented about this issue. In 1846, Carl Czerny
(1791–1857) was fearful that Beethoven’s piano sonatas would be misinterpreted
by current and future generations outside the Viennese tradition. For Czerny,
who earwitnessed Beethoven’s piano playing, assiduous study could lead to
“a certain degree of perfection.” But it would take deeper understanding to reveal
the spirit, humor, freedom, and beauty of Beethoven’s works.2 He felt compelled,
therefore, to give “as exact an indication as possible” of the correct tempo

Carl Czerny, The Art of Playing the Ancient and Modern Piano Forte Works . . . Being a Supplement to
the Royal Pianoforte School Op. 500 (London: Cocks, 1846), 30.
Introduc tion xxiii

(with metronome marks) and other verbal advice about character. This, he hoped,
would aid the proper study and performance of these masterpieces to advanced-
level pianists.3
A half-century later, the revered pianist, teacher, and composer Carl Reinecke
(1824–1910) was clear in his opinion that neither musical notation nor verbal
description could preserve subtle details of past styles. Commenting in 1895
on the care with which Beethoven had notated expressive nuances in the
Piano Sonata Op. 111, Reinecke explains that a conscientious reading would, as a
matter of course, transmit all essential details. Yet there remained “much to be
read between the lines which no composer can convey by signs, no editor by
Reading between the lines remains an important and complex issue in the
interpretation of classical music because the score is incomplete and will rarely
reveal what the composer truly intended. The English pianist Harold Bauer
(1873–1951) reflected on this in 1948. Eminent conductors and performers
“deceive themselves strangely” when claiming to adhere exactly to the composer’s
indications. The composer’s actual intentions cannot be realized because the
notation preserves “only relative, and not absolute, directions for performance.”
It is “an approximation which no two people can interpret in precisely the same
way.”5 Others too have concluded that there is an inherent incompleteness in
musical notation. More recently, in The End of Early Music (2007), Bruce Haynes
argues that all attempts to portray musical ideas through it, amount merely
to an approximation. Despite great care for accuracy and detail, “it always seems a
small miracle” if the eye can discern all the subtleties that the ear registers.6
Clearly, musical notation fails to convey all essential aspects of a composer’s
The evolving meaning of inherited notational signs and terminology adds
yet a further layer of complexity. What was once fashionable or accepted may
no longer be so. A good example is seen in the changing interpretation of the
slur over two, three, or four notes. In addition to legato (smooth playing), several
eighteenth- and nineteenth-century instruction manuals prescribe a small or
“imperceptible” accent at the beginning of the slur followed by a diminuendo
through it, with perhaps a shortening of the final note. Florence May (1845–
1923), an English pianist who studied with Brahms in 1871, recalled that “he
made very much of the well-known effect of two notes slurred together, whether

Czerny, Art of Playing, 68.
Carl Reinecke, Die Beethovenschen Clavier-Sonaten: Briefe an eine Freundin (Leipzig, Germany:
1895); trans. E. M. Trevenen Dawson as The Beethoven Pianoforte Sonatas: Letters to a Lady (London:
Augener, 1898), 139.
Harold Bauer, Harold Bauer: His Book (New York: Norton, 1948), 266–67.
Bruce Haynes, The End of Early Music: A Period Performer’s History of Music for the Twenty-First
Century (New York: Oxford University Press, 2007), 105.
xxiv Introduc tion

in loud or a soft tone.” She added that Brahms was insistent about this point:
“the mark has a special significance in his music.”7 In 1879, Brahms himself advo-
cated a shortened final note when two notes under a slur appeared in his own
music. Based on textual evidence, Brown concludes that this would often have
involved a subtle diminuendo.8 For the majority of classical musicians today,
however, such an interpretation is foreign, outdated, overly mannered, and a
threat to the seamless shape of a melody.9

Authentic Performance and Historical Accuracy

During the second half of the twentieth century there was a significant increase in
the practical application of historical performing practices. By the 1970s and ’80s
more and more musicians were aiming to restore to particular repertoire some
of the performing conventions and conditions that apparently prevailed at the
first performance. And many were using historical instruments or replicas and
claiming to be thoroughly “authentic”—to be re-creating the work as the com-
poser originally conceived it. The sounds that issued in concerts and recordings
were unquestionably different in many ways to the mainstream versions of the
same or similar works. They were lighter, less vibrato-laden, more phrased, more
energetic, and more “danced.” The texture was also noticeably clearer. And much
unknown or forgotten repertoire was revived. For many, including myself, this was
inspiring, exciting, and liberating.
We live in an age in which “growing legions of early music enthusiasts now look
for telltale signs that performers have done their research, evident in the use of
period instruments, embellishments (ornamentation), and improvisation,
amongst other things.”10 The historically informed performance (HIP) movement
is certainly on the ascent. Yet the movement’s claims to authenticity gave rise to
intensely heated debate during the last three decades of the twentieth century.
Luminaries such as Richard Taruskin, John Butt, Nicholas Kenyon, and many
others have argued about the extent to which historical authenticity is possible,
much less desirable, and disputed the many reasons that the HIP movement has
taken such a stronghold. My experience through over twenty years of professional
music making, in both the HIP and mainstream scenes coupled with ongoing
performing-practice research, has taught me to be skeptical about the possibility

Florence May, The Life of Johannes Brahms (London: Arnold, 1905), vol. 1, 18.
Clive Brown, Classical and Romantic Performing Practice 1750–1900 (Oxford: Oxford University
Press, 1999), 233–34.
For further discussion, see Haynes, The End of Early Music, 107.
Wendy Weisman, “Christie to Launch 2-Part Residency at Julliard: Workshops Herald
New Degree Programme in Early Music,” The Julliard Journal vol. 23, no. 3 (Nov. 2007): 1.
Introduc tion xxv

of historical accuracy. For me Haynes hits the nail on the head in surmising that
“totally accurate historical performance is probably impossible to achieve” and in
any case it is impossible “to know it has been achieved.”11 A big part of this is that
we do not, indeed we cannot, hear music now as audiences in previous eras did. In
Evenings in the Palace of Reason (2005), James R. Gaines makes this point lucidly:

Just as Bach influenced all the music and history that came after him, all
that music and history changed him, or changed at least how his music
could be heard. For this reason and others, no matter how “original” the
instruments or groupings of choristers, however “authentic” a perfor-
mance strives to be, Bach can never be heard as his contemporary
audiences heard him.12

But is historical accuracy or authenticity the raison d’être for the study of
performing practices? Whether or not historical accuracy is possible, I—like many
others—see great value in arming oneself with as much information as possible
about the original performance ideals for any musical work. Through this process
the work can be viewed from new or different perspectives, amplifying the choices
available in its realization. Having more choice makes for a more varied and
flexible musical intuition. This is a laudable goal for something that, as Taruskin
postulates, has been formed through “long years of unconscious conditioning.”
For trained musicians “intuition will be the unexamined mainstream, your most
ingrained responses, treacherously masquerading as imagination.”13 With a newly
stimulated musical intuition, reading between the lines can produce original and
interesting results that are a product of our own time, the style of now.14 Brahms’s
symphonic works on period instruments, for example, are making people sit up
and listen. As Bernard D. Sherman reports “HIP Brahms is thriving more than
I expected, because it continues to rekindle musician’s passion for Brahms.”15

Historical Performing Practices: Their Value

The study and practical application of historical performing practices earlier than
the mid-nineteenth century have inevitably focused on data preserved in a range

Haynes, The End of Early Music, 10.
James R. Gaines, Evenings in the Palace of Reason: Bach Meets Fredrick the Great in the Age of
Enlightenment (New York: HarperCollins, 2006), 271.
Richard Taruskin, Text and Act (New York: Oxford University Press, 1995), 78.
Taruskin, Text and Act, 102.
Bernard D. Sherman, “Orchestral Brahms and ‘Historically Informed Performance’: A Progress
Report,” http://www.bsherman.net_Brahms_Diapason_Sherman_English.htm (accessed December
15, 2010).
xxvi Introduc tion

of written texts. These include pedagogical or teaching texts, earwitness accounts

of performers and performances, and analyses of composers’ notational practices.
This so-called authentic, period, or historically informed approach has provided
invaluable information about evolving tastes and musical vocabulary. In his arti-
cle “The Present Position of Authenticity” (1989), Robert Donington vehemently
argues in favor of the use of historical texts, extolling their virtues and claiming
that we can come close to understanding a composer’s intentions with sufficient
knowledge of his “notational symbols and unnotated conventions.” He is also
confident that “a reasonably reliable and consistent view of large and important
areas of factual information” can be pieced together. Taking into account differ-
ences in taste and style, this approach would give rise to an “educated flexibility”
in performance.16
This educated flexibility is certainly a desirable quality about which other writ-
ers have commented. In 1915, Arnold Dolmetsch (1858–1940)—a pioneer of the
HIP movement—emphasized the importance of this. He advised students to
investigate thoroughly “what the Old Masters felt about their own music, what
impressions they wished to convey, and, generally, what was the Spirit of their
Art.”17 If nothing else, this means can certainly bring a deeper connection with
the composer and the music. And as Howard Mayer Brown speculated in 1988,
it might also inspire “an imaginative re-creation in performance of the notes on
the page.”18 In the end it is really a question of the possibilities available in perfor-
mance. Mayer Brown observed that “intelligent performers” will want to know
“about the possibilities open to them.” And nowadays, the most intelligent per-
formers will almost certainly be historically informed.19 There is certainly no harm
in being informed. Clearly, the more we are informed, the more equipped we are
to make informed choices and to fill in the gaps in our knowledge. This is certainly
Taruskin’s stance. For him, performing practice research attempts to create a
bridge between the information in old musical texts (musical editions, scores, and
parts) and “what was actually heard in typical contemporary performances.”20
But even after these considerations, the question remains: is it really possible
to form a reliable and consistent view of information about historical performing
practices? Do we have access to enough pieces of what is after all a rather complex
jigsaw puzzle? The answer is uncertain. However much it may be possible to glean
the meaning of particular notational symbols or some of the many unnotated

Robert Donington, “The Present Position of Authenticity,” Performance Practice Review vol. 2,
no. 2 (1989): 119–20.
Arnold Dolmetsch, The Interpretation of the Music of the Seventeenth and EighteenthCenturies
(1915; 2nd ed., London: Novello, 1946), vii.
Howard Mayer Brown, “Pedantry or Liberation?” Authenticity and Early Music: A Symposium,
ed. N. Kenyon (Oxford: Oxford University Press, 1988), 41.
Mayer Brown, “Pedantry or Liberation?” 55.
Taruskin, Text and Act, 18.
Introduc tion xxvii

conventions that existed for a specific time and place, the precise aural effect of
these remains largely unclear. We simply cannot be certain how these actually
sounded or were perceived.
To clarify this, imagine trying to restore an ancient fresco that is cracked and
faded, and has many little pieces missing. Information from written sources of the
era might provide some clues about fresco style and methods of production. But
without having seen the fresco in its original state, we can never be sure of the
intricate details and patterns, nor the intensity of color, that have been perma-
nently erased.21 Similarly, the seemingly plentiful information preserved in
written sources about musical performance conveys, in the majority of cases, only
an approximation to actual practice. Without audible evidence, it is impossible to
appreciate many of the features of past performing styles that were transmitted
aurally and taken for granted.

Urtext Performance
Undoubtedly, such features were of both a general and idiosyncratic nature and
simply too complex to notate or describe. About this matter, Will Crutchfield
concludes that the conviction, passion, grace, confidence, and stylish freedom of
great and cultivated performers of the past cannot be “reclaimed from the trea-
tises nor extrapolated from critical editions.”22 Written texts and urtext editions
do not preserve the myriad of unnotated conventions. For Taruskin, these include
the expressive dynamic and tempo changes that would no doubt have graced per-
formances of the past but were impossible to describe or notate, as well as the
idiosyncrasies that all creative musicians must have had. For this reason he was
critical of the HIP movement’s performances, which generally failed to acknowl-
edge such factors and produced merely “the aural equivalent of an Urtext score.”
Notes and rests are meticulously accurate, but the performance is completely
neutral: “nothing is allowed to intrude into the performance that cannot be
‘authenticated.’”23 Ironically, this neutrality has been strongly fostered by many
in the HIP movement, who—perhaps fearful of criticism—selectively apply
practices preserved in pedagogical written texts. Those that are at odds with
current aesthetics of good taste are ignored. But, it is the very features (some
of them alien) missing from musical notation that, as Brown hypothesizes,

Interestingly, while we are used to and even cherish the clean white marble exteriors of
classical Greek statues, these were apparently originally painted in vibrant, gaudy colors. See Matthew
Gurewitsch, “True Colors,” Smithsonian Magazine, July 2008. Available: http://www.smithsonianmag.
com/arts-culture/true-colors.html, 2.
Will Crutchfield, “Fashion, Conviction, and Performance Style,” Authenticity and Early Music:
A Symposium, ed. Nicholas Kenyon (Oxford: Oxford University Press, 1988), 25.
Taruskin, Text and Act, 72.
xxviii Introduc tion

“created the essential sound-world within which the composers of the age
conceived their music.” While a lack of attention to such details would not detract
from the hallmarks that make the best works immortal and relevant, taking them
into account might help the performer in particular, to understand the music at a
deeper level.24

Sound Recording: A Window into the Past

The development of sound recording in the late nineteenth century is of great
significance. It provides, for the first time, direct evidence of the features of indi-
vidual musicians’ performance styles that would otherwise have been irretriev-
ably lost. Early acoustic recordings and player-piano roll recordings shed light on
the performance traditions of approximately the last 150 years. To our modern
ears these recordings often sound shockingly foreign: their style bears little
resemblance to anything heard today. The performances are sometimes haunting,
sometimes declamatory. They exhibit expressive qualities that are heartfelt and
leave an indelible impression. Above all, they preserve a tradition that is, as
Taruskin states, “instantly recognizable as premodern.”25 These recordings show
that late-twentieth-century traditions and styles are quite different from those of
even seventy or eighty years earlier. Robert Philip’s analysis and comparison of
recordings from 1900 to 1950 clearly illustrates that the following aspects of per-
formance style changed considerably: accentuation, articulation, ideals of tone
color and timbre, tempo fluctuation, tempo rubato; and in the case of vocalists,
string, and wind players, portamento and the variation and shading of tone
production through the use of vibrato and nonvibrato.26
In 1962, the British music critic Edward Sackville-West (1901–65) made refer-
ence to a significant change in the style of piano playing that had by then taken
place. Comparing the recordings that the pianist Moriz Rosenthal (1862–1946)
made during the 1920s with those of a later generation, he remarks that
Rosenthal’s playing is “stamped with a musical personality strikingly different”
from the best contemporary pianists: his “poetry and distinction” are perfumed
with the scent of a bygone era. Describing Rosenthal’s practice of nonsynchrony
of the hands (playing one hand after the other), as well as an apparently careless
attitude to wrong notes, Sackville-West observes that the pianist did not seem to
care whether anyone was listening. But he is quick to add that such practices would
no longer be acceptable and “nothing but a cast-iron technique” would ensure the

Brown, Classical and Romantic Performing Practice, 631.
Taruskin, Text and Act, 168.
Robert Philip, Early Recordings and Musical Style: Changing Tastes in Instrumental Performance,
1900–1950 (Cambridge: Cambridge University Press, 1992).
Introduc tion xxix

success of a public performance career.27 A cleaner, more precise and more

text-faithful style, with less flexibility of rhythm and tempo, was increasingly
considered the benchmark in piano playing.
Undoubtedly, such changes in attitude and playing style went hand in hand
with developments in sound recording techniques during the second half of
the twentieth century. The invention of sound recording itself provided a strong
catalyst for change. As Taruskin posits, it led to the eroding of personality and
emotion from the performer “since it offers the possibility of permanence to a
medium that had formerly existed only in the moment.”28 This also allowed the
performer objectively to compare and model their own playing on that of other
artists. Nevertheless, recordings from around the turn of the twentieth century
remain a veritable treasure trove. They provide a window into the era before the
change. These recordings reveal an ad hoc or seemingly careless approach to the
literal, notated notes and rhythm, tuning, and basic pulse. To “modern” sensibili-
ties, the effects seem primitive, old-fashioned, and curious-sounding. In fact,
these are intrinsic performing practice elements. In this respect John Butt con-
cludes that those features that might sound “casual, senselessly erratic and only
accidentally expressive” now, were once the keystone of artful interpretation.29
Early recordings are characterized by originality and freedom of expression that
produces a feeling of improvisation. Comparing the evidence in recordings,
Haynes is very clear about this. He provides useful distinctions between Romantic
and modern pianism:

If Romantic protocol was heavy, personal, organic, free, spontaneous,

impulsive, irregular, disorganized and inexact, Modern style is the
reverse: light, impersonal, mechanical, literal, correct, deliberate, consis-
tent, metronomic, and regular. Modernists look for discipline and line,
while they disparage Romantic performance for its excessive rubato, its
bluster, its self-indulgent posturing, and its sentimentality.30

Late Romantic pianism, as evidenced in early recordings, made use of many

wonderfully expressive practices now considered by many to be capricious, pecu-
liar, or disturbingly eccentric. But it was these that contributed to what Kenneth
Hamilton describes as “variety and individuality” in performances. And although
these and other practices are not to be heard in concerts and recordings today,

Edward Sackville-West, “Rosenthal,” Recorded Sound: The Journal of the British Institute of Recorded
Sound vol. 1, no. 7 (1962): 214.
Taruskin, Text and Act, Footnote 16, 61.
John Butt, Playing with History (Cambridge: Cambridge University Press, 2002), 35.
Haynes, The End of Early Music, 49.
xxx Introduc tion

for him there has undoubtedly been “a sacrifice of interest, spontaneity and sheer

The Quest for Perfection

Sound recordings, particularly from an era when few or no artificial editing
techniques were available, is the most important primary source for appreciat-
ing performing traditions of that period. They are, as Taruskin boldly puts it,
“the hardest evidence of performance practice imaginable.”32 In stark contrast,
much in present-day commercial recordings does not always give a true impres-
sion of actual performance style. My own experience of recording has taught me
that a producer’s power of intervention through modern editing techniques can,
and often does, alter several significant elements of the initial performance.
Consequently, in some cases the recording’s worth as preserved evidence may
have been devalued. There was already a growing anxiety about this in the first
half of the twentieth century. In 1931, pianist Mark Hambourg (1879–1960)
expressed concern about the growing power of the producer, the so-called engi-
neer-musician associated with both radio and recording companies who could
modify the sound “to suit their own ideas.” He found this abhorrent and ques-
tioned whether one could “reconcile the artist-musician with his fantasy and the
engineer with his mechanical expediency in the realms of art.”33
The power of the editor, coupled with the recording industry’s drive toward
artificial perfection and the resulting consumer expectation, has shaped late-
twentieth-century taste. In this context, and as Philip has shown, perfection has
come to mean stricter rhythmic precision of ensemble, absolute respect for nota-
tion, and an eradication of the various types of tempo rubato that were integral
aspects of expression and phrasing around the turn of the twentieth century. This
quest for perfection can be taken to extremes. Frequently we, the recording art-
ists, strive for and insist upon an almost unattainable perfection. We expect the
producer to do whatever it takes in the editing process to achieve the best results.
I have certainly felt what pianist Charles Rosen observes as “an immense pressure
to make everything correct” in recording sessions. No one wants to be accused
of not being able to “get it right,” despite the ease with which editing can take
place.34 Evidently, this pervasive pressure was felt quite early on in the history

Kenneth Hamilton, “The Virtuoso Tradition,” The Cambridge Companion to the Piano, ed. David
Rowland (Cambridge: Cambridge University Press, 1998), 73–74.
Taruskin, Text and Act, 168.
Mark Hambourg, From Piano to Forte: A Thousand and One Notes (London: Cassell, 1931), 291.
Charles Rosen, Piano Notes: The Hidden World of the Pianist (London: Penguin Books, 2002),
Introduc tion xxxi

of recording. Hambourg acknowledged the difficulty of making “an interesting”

recording because the soloist’s imagination has become stultified by an obsession
with accuracy. The result is a “lifeless” performance, “a mere succession of notes.”
He advised performers to retain their humanness and “to impart the warmth of
their personality to what is reproduced.”35
We can appreciate now that early recordings provide a window into the recent
past. But for many decades it was almost exclusively museum curators who valued
the capacity of the recording process “for preserving the voices of the living and as
a means of immortalizing historic performances of great musicians.” Discouraged
by the poor sound quality and “lack of fidelity” characteristic of recordings of
the era, musicians and music educators showed far less interest. It eventually
dawned that there was much important information preserved in the recordings,
which might one day be revealed “to a remarkable extent.”36 Yet, by the end of
the twentieth century, few apart from those in the antiquarian trade showed
much interest in early recordings. Taruskin surmises that this is because early
recordings are too far removed from modern taste: they sound too much like
caricatures.37 Nevertheless, these preserve the performing practices of recog-
nized virtuosi of the era. For this reason, I think Taruskin is absolutely on the
money in his suggestion that if we really wanted to understand what it would
take to perform in a historical way, “we would begin by imitating early-twentieth-
century recordings of late-nineteenth-century music and extrapolate back from
Scholars have only recently begun to engage significantly in the academic study
of early recordings. The transfers and reissues of many of the earliest acoustic
recordings (as early as 1889 in the case of Brahms) and piano rolls to long-playing
records and compact discs provide the most important primary source of evidence
for late nineteenth century and, in some cases, even earlier traditions. We can
all now readily access examples of some of the most famous and revered artists
of the late Romantic era including the soprano Adelina Patti (1843–1919),
the violinist Joseph Joachim, the pianists Carl Reinecke, Theodor Leschetizky
(1830–1915), Camille Saint-Saëns (1838–1921), Edvard Grieg (1843–1907), and
Vlademir de Pachmann (1848–1933), as well as a younger generation of pia-
nists such as Jan Paderewski (1860–1941), Fanny Davies (1861–1934), Moriz
Rosenthal (1862–1946), Carl Friedberg (1872–1955), Adelina de Lara (1872–
1961), Ilona Eibenschütz (1873–1967), Alfred Cortot (1877–1962), Etelka Freund

Hambourg, From Piano to Forte, 289–90.
Oliver Read and Walter L. Welch, From Tinfoil to Stereo: Evolution of the Phonograph (Indianapolis:
Sams, 1959), 163–64. Unless otherwise stated, all future references refer to this first edition.
Taruskin, Text and Act, 168–69.
Taruskin, Text and Act, 168.
xxxii Introduc tion

(1879–1977), and many others.39 The surviving audible evidence preserves vital
information about general performing practices of the era as well as documents
the idiosyncrasies of individual performers.

The Evolution of Tradition

Although early recordings tell us much about a particular era, it is also well worth
considering the extent to which they capture an ongoing tradition. No doubt the
nineteenth-century musicians noted earlier modified elements of their style
during the course of their careers, but any change was likely to be more gradual
than the rate of change in the late twentieth century. Modern communication
systems and multimedia have hastened both the transfer and absorption of infor-
mation, increasing the possibility of rapid change. A fascinating example illustrat-
ing a fairly slow rate of change is seen in the evolution of expressive devices used
in string playing from the latter part of the nineteenth century to the middle of
the twentieth century. The earliest recordings show that, around 1900, many
string players of the German school still played without continuous vibrato,
employing instead a varied portamento (sliding between notes), as the main
expressive device. Vibrato was treated as an ornament, reserved for certain musi-
cal effects like accent or emphasis. But during the first decades of the twentieth
century, vibrato can be heard more continuously alongside portamento. It is
not until the mid-twentieth century or later that portamento falls almost com-
pletely out of use, leaving vibrato predominant as a means of coloring tone. Yet
even this type of vibrato is narrower, less continuous, and less intrusive than the
vibrato generally employed at present.40 Clearly, the transition from the judicious
use of vibrato with frequent portamento to one employing a wide, continuous,
and largely unvarying vibrato, with little or no portamento, has taken almost
100 years.41
An analogous development in piano playing is noticeable in the practice of
playing one hand after the other (manual asynchrony), creating separation
between melody and accompaniment. Employed particularly in expressive music,
this technique can be heard in the playing of many of the earliest generation of
pianists to record—those whose careers spanned the second half of the nine-
teenth century, and to varying degrees those of later generations. Remnants of

YouTube is also an invaluable resource in this regard.
This is not true of those adopting period or authentic approaches, who, in general, seem to have
rejected portamento as a general expressive device and use varying speeds of vibrato ornamentally,
combined with vibratoless tone.
Presently, there is a move in some circles toward more subtle use of vibrato. Clearly, the evolu-
tion process continues.
Introduc tion xxxiii

this technique are preserved on recordings made as late as the 1940s and 1950s.
Noticeably, however, some late-nineteenth-century pianists and the majority of
pianists trained during the twentieth century employ these devices to a much
lesser extent or not at all. Undoubtedly, changing tastes and technical develop-
ments account for the move away from a performing practice once considered
indispensable, though the change took place over a considerable period of time.
A century of sound recording has shown clearly that musical traditions are
constantly evolving. It is dangerous to assume, therefore, that one musician’s
playing style, no matter how venerated, has been adopted and transmitted by
following generations. The fact that your teacher’s teacher studied with Franz
Liszt (1811–86) does not automatically mean that you have inherited Liszt’s own
playing style. Hamilton comments that “pianists are often exaggeratedly proud of
their pedagogical pedigrees—as if musical talent were passed on by apostolic
succession.”42 In this respect, Crutchfield has concluded that musical style would
never change if students simply mimicked their teachers. Students do not play
like their teachers, but they react to their teachers in myriad ways that are
imitative and rebellious. And their playing reflects the progressive style of the
young “twenty-year-olds”—the new virtuosi on the block—rather than “the
style that was current.”43 The recordings of Liszt’s pupils will not tell us how Liszt
himself played, but, as Hamilton suggests, it will inform us of “the broad stylistic
range within which he expected most piano music, including his own, to be inter-
preted.” For him, this is of particular importance because the playing of Liszt’s
pupils exhibits so many practices that are entirely different from performances

The Effect of Recording

Indeed, the recording process itself was largely responsible for changes in taste
and performance style. Before recording became possible, musicians employed
certain techniques considered expressive without being able to appreciate the
aural effect purely from a listener’s point of view. The legendary accompanist
Gerald Moore (1899–1987) remarked in 1962 that with the advent of the micro-
phone, “it is still beyond our capacity to see ourselves as others see us.” But by
then at least, the more sophisticated recording apparatus allowed a faithful
hearing of oneself. Dismayed by what they heard, newcomers to recording
would ask, “Is this really what I sound like?”45 Once it became possible to hear

Hamilton, “The Virtuoso Tradition,” 72.
Will Crutchfield, “Brahms, by Those Who Knew Him,” Opus vol. 2, no. 5 (1986): 14.
Hamilton, “The Virtuoso Tradition,” 72–73.
Gerald Moore, Am I Too Loud? (Harmondsworth, U.K.: Penguin, 1968), 56.
xxxiv Introduc tion

one’s own performance, critical listening, intermingled with anxiety about the
fact that the recorded result would survive for posterity, influenced certain
aspects of performing style. Musical and technical features, be they intentional or
not, “are fleeting and lost” in live performances, but recordings remain available
for “repetition, comparison with others and critical analysis.”46 Moore, whose
experience of recording was wide-ranging, noted the effect it had on his own

The microphone exposed—and continues to expose—so many short-

comings in my playing that I sometimes wonder why I am ever re-engaged.
I can only assume it is because I have never been found out. It is a
humiliation to record a piece of music one has performed in public for
years and then to discover how poorly one has played it.
But at least the mike has taught me to listen to myself mighty
critically. I owe much to it though still hating and fearing it.47

Most of the recordings examined in this book were made in the first decade of
the twentieth century, and some later recordings have also been considered. These
capture the styles of key pianists and a few other important musicians at the end
of their careers, in some cases trained in the mid-nineteenth century. The research
to date has largely overlooked some of the most significant evidence of late
Romantic pianism as heard in the playing of Reinecke, Leschetizky, Saint-Saëns,
and others. In the chapters that follow I look at what is happening in these musi-
cal “snapshots” to distinguish what is old-fashioned or modern, idiosyncratic or
a general trend. Given the limitations of the early recording process (explored
in chapter 1), expressive practices discussed here include only those that are
clearly distinguishable: playing one hand after the other (dislocation of melody
from accompaniment), unnotated chordal arpeggiation, rhythmic alteration, and
tempo modification. I compare these with contemporaneous written texts on
performance (some highly detailed; others more general) to evaluate the corre-
spondence between actual practice and its written description. Looking at the
recordings of the oldest generation of pianists, as well as of those who followed, in
specific conjunction with written texts, reveals far more than is possible from a
study of recordings or written texts alone.

Read and Welsh, From Tinfoil to Stereo, 164.
Moore, Am I Too Loud? 58.
Off the Record
This page intentionally left blank

Early Recordings
Their Value as Evidence
I consider that to play properly for recording purposes needs
superhuman nerve, as the machine itself is superhuman. Nothing
can be altered on the gramophone once it is recorded. There stands
the music as the artist played it, and if mistakes are made there
they remain to convict him. It also gives the artist an uncanny
feeling to hear his own playing emerging from the sound box.
It seems at once curiously strange and familiar.
—Mark Hambourg1

By the turn of the twentieth century, two methods for capturing the artistry of
famous pianists became possible: acoustic recordings and reproducing piano rolls.
The developments that followed weave fascinating and complex tales, theaters
of personalities, inventions, and intrigue, all of which could fill many books. In
providing an overview of what took place, I am piecing together the most impor-
tant details as I have come to understand them and to set parameters within
which, I believe, the evidence may safely be used. Of necessity, I am drawing on
the insights of published literature, the reminiscences of key figures, and expert

Acoustic Recording
Acoustic or preelectrical recordings relied on the transmission of sound vibrations
via a conically shaped funnel called a trumpet or horn to a sensitive membrane
attached to a needle (a stylus). Sympathetic movements of the membrane
caused the needle to make an impression (a sound line) into a suitable medium.
It continues to astonish me that this sound line could preserve data from the
transmitted sound vibrations.

Hambourg, Piano to Forte, 293.

Peres da Costa-Ch01.indd 3 3/9/2012 2:07:35 PM

4 off the record

Talking Machines: The Phonograph

The overture sounds with Frenchman Édouard-Léon Scott de Martinville (1817–
79). As early as 1857, he patented the Phonautograph, a machine that recorded
visual sound tracings (a phonautogram) but was unable to play them back.2
Twenty years later, in 1877, French scientist Charles Cros (1842–88) submitted a
proposal for a sound recording and reproduction apparatus called the Paleophone.
His idea was to trace sound waves onto glass covered in lampblack (soot). These
tracings, he theorized, could be photoengraved to produce indentations that
could then be played back. But, in the same year, an invention by Thomas Edison
(1847–1931)—now known as the father of the sound recording industry—
eclipsed Cros’s idea. He devised a “simple” cylinder machine called the Phonograph
(patented in 1878). This was the real curtain raiser, a monumental breakthrough.
For the first time, it became possible both to record sounds and reproduce them
at will.3
In Edison’s first phonograph, sound vibrations were recorded by indenting tin-
foil stretched over a cylindrical drum. This created a vertical (“hill-and-dale”)
sound line in a groove of fixed depth.4 Inevitably, there were some serious issues
to resolve. No external amplification was possible in either recording or playback.
All the power (and therefore, the volume) had to come from the sound source
itself. What’s more, the tinfoil produced an unacceptable level of background
noise.5 And its shape became distorted after just a few replays.6 At this stage, the
phonograph, also known as a talking machine, was little more than a form of

The Graphophone
The novelty of tinfoil recording quickly wore thin. Some other more successful
recording medium would have to be found if sound recording were to be taken
more seriously. By 1885, Alexander Graham Bell (1847–1922), together with his
cousin Chichester Bell (1848–1924) and associate Charles Sumner Tainter (1854–
1940), hit on an interesting solution while experimenting with Edison’s cylinder.
Instead of using tinfoil, they covered the circumference of a cardboard cylinder

Jody Rosen, “Researchers Play Tune Recorded Before Edison,” New York Times, March 27, 2008.
Oliver Read and Walter L. Welch, From Tinfoil to Stereo: Evolution of the Phonograph, 2nd ed.
(Indianapolis: Sams, 1976), 10.
For detailed descriptions of Edison’s phonograph, see Peter Ford, “History of Sound Recording,”
Recorded Sound vol. 1, no. 7 (1962): 222; Read and Welch, From Tinfoil to Stereo, 21–23; Roland Gelatt,
The Fabulous Phonograph 1877–1977, 2nd rev. ed. (London: Cassell, 1977), 20–21.
Peter Copeland, Sound Recording (London: British Library Board, 1991), 7.
Gelatt, The Fabulous Phonograph, 26.
Early R ecording s 5

with a wax-like substance called ozokerite. And they substituted Edison’s indent-
ing stylus with an incising stylus.7 Now sound vibrations were cut (rather than
indented) cleanly into wax. The resulting hill-and-dale undulations were greater in
magnitude and preserved more information than before. Advantageously, the
wax-coated cardboard cylinders could be mounted and removed without too much
distortion to the wax.8 On playback, there was noticeably less background noise.
With this, the process of wax cylinder recording on the Graphophone (patented in
1886) was launched.
So successful was the Graphophone that Edison eventually responded to it by
adopting wax and the incising method into his “Improved Phonograph.” But
instead of wax-coated cardboard, he introduced solid wax cylinders that could be
reshaved for continuous reuse.9 By 1888, electric motors were starting to be installed
into both the Graphophone and Phonograph, ensuring a constant speed for record-
ing and playback.10 With such innovations, wax cylinder recording machines had
moved on from being mere toys. Initially they were employed as dictating machines
in offices and subsequently as sources of popular entertainment—the nickel-
in-the-slot machine, introduced by 1889 and reincarnated as the juke machine.11
Such improvements also paved the way for recordings of serious music. In
1888, the twelve-year-old piano prodigy Josef Hofmann (1876–1957) was the
first famous artist to record on an Edison cylinder.12 And this was soon followed
by the German pianist/conductor Hans von Bülow (1830–94). This was probably
the only recording that Bülow made and sadly it appears not to have survived.
Fortunately, Brahms’s 1889 cylinder recording does survive, and despite the dis-
torted sound quality, it provides some important clues about Brahms’s playing
style, more of which will be discussed later.
Although wax cylinder recording was revolutionary, the technology for mass
production did not exist. By setting up a bank of machines—up to twenty at a
time—one take could yield twenty master cylinders.13 Eventually, duplication
became possible through a pantographic process.14 About 25 copies of each master
could be produced in this way. A successful take might therefore yield up to

Read and Welch, From Tinfoil to Stereo, 31.
Copeland, Sound Recording, 7.
Copeland, Sound Recording, 8.
Gelatt, The Fabulous Phonograph, 38.
Joe Batten, Joe Batten’s Book: The Story of Sound Recording (London: Rockliff, 1956), 44.
Hofmann’s recordings from 1895 and 1896 can be heard on Marston 53011-2 (2008). These are
transfers of wax cylinder recordings made in Russia by Julius Block. Having secured a phonograph
from Edison, Block made recordings from 1889 onward of many famous artists who lived in or visited
Moscow and St. Petersburg.
Frederick W. Gaisberg, The Music Goes Round (New York: Macmillan, 1942), 6.
In this process the hill-and-dale grooves were mechanically copied from a wax master to a wax
blank by linking together a playback stylus and a cutting stylus.
6 off the record

500 cylinders. To produce more of the same, the artist had to record the work
repeatedly during the recording session—a somewhat Herculean feat. In real
terms, supply simply could not meet demand. The process was commercially

The Gramophone
Enter Emile Berliner (1851–1929), a rising star of the scene. He was dissatisfied
with the level of noise and distortion produced by the vertical cut groove of wax
cylinder recording. Revisiting the phonautographic process of Scott de Martinville
and the photoengraving process of Cros, he carried out experiments to find
solutions to the problems.15 By 1888, Berliner had patented a new, intriguing
process—recording on a flat disc with “lateral vibrations”—a basic process that
survived right up to the end of the vinyl long-playing record. Instead of hill-and-
dale incisions of varying depth, the stylus made lateral (sideways or horizontal)
etchings of even depth onto flat discs rather than cylinders. Initially glass discs
coated in lampblack were used and subsequently zinc discs coated in beeswax.
Fred Gaisberg (1873–1951)—the accompanist and talent scout for Berliner at the
time—was present at the creation of the first gramophone record. He recalled
how, after the recording, Berliner plunged the bright zinc disc into a bath of acid
for a few minutes, which he then washed and cleaned. When the acid-etched flat
gramophone disc was replayed, Gaisberg was simply astonished: the sound was
more natural than the “tinny, unnatural reproduction” of wax cylinder phono-
graphs. He was spellbound by “the beautiful round tone.”16 With the lateral-etched
method of flat disc recording, the stylus met with far less resistance than with the
hill-and-dale cut method in cylinder recordings.17 The result was less distortion
and more volume.18 Nevertheless, distortion still remained a problem: listeners
had to endure “crude and noisy records.”19
Unlike the wax cylinder, the flat disc gramophone could be used only to play-
back prerecorded sound. But this fulfilled its purpose, which was not for office use
but for the reproduction of speech and music solely for home entertainment.20
The great commercial advantage was that discs could be mass-produced easily and
cheaply. To that end, Berliner made electroplated molds (matrices) from the later-
ally etched groove created in the initial recording. These were inside-out, or nega-
tives, of the recording, with outward jutting ridges rather than inward grooves.21

Read and Welch, From Tinfoil to Stereo, 121.
Gaisberg, The Music Goes Round, 9–10.
Read and Welch, From Tinfoil to Stereo, 123.
Ford, “History of Sound Recording,” 225.
Gaisberg, The Music Goes Round, 18.
Read and Welch, From Tinfoil to Stereo, 122.
Copeland, Sound Recording, 8
Early R ecording s 7

From them, “stampers” were made and positive records were pressed. By this
method literally “hundreds of thousands of records” could be produced from a
single performance.22
Berliner experimented with celluloid, and then vulcanized rubber, for his discs.
Around 1897 he settled on Durinoid—a compound of powdered shellac23 and
byritis (clay), bound together with cotton flock and colored with lampblack.
Durinoid was easy to mold under pressure and preserved the groove shape when
cooled. Gaisberg recalled the “sounds of undreamed quality” that issued. The new
shellac compound “poured into every crevice of the sound-track,” revealing tones
that were not possible to hear before.24 The use of shellac compound for 78 rpm
disc records continued well into the mid-twentieth century.
By 1896, further significant improvements were made to the gramophone by
Berliner’s associate Eldridge R. Johnson (1867–1945). These included incorporating
a spring or clock-motor to ensure an even speed for the rotation of the turntable,
and the use of solid wax instead of zinc discs for recording. According to Gaisberg,
making the master recording onto wax ensured “a cleaner cut, less surface noise
and the music more faithfully registered.”25
Shellac disc records, with a time limitation of about four and a half minutes,
dominated the market until the late 1940s, at which point the long-playing record
took over.

Cylinder versus Disc

During the early years of the twentieth century, there was an explosion of activity
in the sound recording industry. Several disc-producing companies vied for top
position. To judge from the number of legal battles, deceit and dishonesty poi-
soned the waters. Disc recording and production certainly cornered the market,
but the cylinder industry continued producing some wonderful recordings until
1929, the year Edison finally called it quits.
Before 1900, wax cylinders were superior to zinc discs in terms of sound qual-
ity. This gulf was somewhat reduced once wax discs replaced zinc discs. But after
1900, the sound quality of cylinders became even better, leading eventually to
Edison’s superior sounding large diameter cylinder on which “the highest over-
tones could be recorded with greater facility and fidelity.”26 But somewhat sadly,

Gaisberg, The Music Goes Round, 44.
A refined form of a natural resinous substance, modified from tree sap, secreted onto the
branches of forest tress in India and Thailand by the tiny female lac insect (Kerria Lacca); see http:// and
Gaisberg, The Music Goes Round, 12.
Gaisberg, The Music Goes Round, 44.
Read and Welch, From Tinfoil to Stereo, 151.
8 off the record

commercial pressures militated against the success of these and the smaller
diameter cylinders prevailed.
Cylinders had one significant advantage over discs. A cylinder’s surface speed
remained constant throughout the recording. A disc’s surface speed, on the other
hand, slowed down progressively from outer to inner grooves. In the case of a
standard ten-inch disc for example, the groove speed beneath the stylus became
diminished by more than 50 percent.27 This resulted in some deterioration of
sound quality toward the end.28
Despite this and other factors, however, the shellac disc with laterally cut
grooves triumphed. Almost from its introduction around 1900 it was more popu-
lar than the cylinder, and was the industry standard until after World War II.29
Discs were easily manufactured and stored. And, above all, they were affordable to
the general public.

Electrical Recording
Acoustic recordings continued until the development, in the early 1920s, of high-
quality microphones to satisfy the growing needs of the radio broadcasting indus-
try. Microphones—devices that convert acoustical energy (sound waves) into
electrical energy—had already been successfully developed for use in telephones
during the latter part of the nineteenth century. These were a specific type of
carbon microphone in which “particles of carbon are alternately compressed and
relaxed by the diaphragm under the influence of sound pressure.” The resulting
alteration of resistance produces “a signal current proportional to the change in
resistance.”30 But due to the limited frequency response and dynamic range, as
well as high levels of distortion, carbon microphones were suitable really only for
speech. For this reason, the Bell Telephone Company, via its manufacturing branch
Western Electric, developed electrostatic (capacitor) and electrodynamic (moving
conductor) microphones. Both types “used a fixed electrical charge on the plates
of a capacitor, one of which was a moving diaphragm and the other a fixed back
plate. Sound waves caused a slight variation in capacitance, which in turn was
translated into a variation in the voltage across the plates.”31
These microphones were able to collect much more sonic information than
the acoustic horn.32 Frequency range was increased by two and a half octaves,

Read and Welch, From Tinfoil to Stereo, 153.
Gelatt, The Fabulous Phonograph, 166.
Jerry McWilliams, The Preservation and Restoration of Sound Recordings (Nashville, Tenn.:
American Association for State and Local History, 1979), 10–11.
John Eargle, Sound Recording (New York: Van Nostrand Rheingold, 1976), 108.
John Eargle, The Microphone Book (Oxford: Focal Press, 2004), 4.
McWilliams, The Preservation and Restoration of Sound Recordings, 9–10.
Early R ecording s 9

encompassing 100–5,000Hz. Extended bass and treble frequencies added body,

definition, and details previously missing. Now recordings had a larger dynamic
range and were louder. At last, musicians could record in more normal, spacious,
and reverberant surroundings.33 And although the original signal generated was
too weak to be used directly in playback, it could be amplified to almost any
required strength “with all its moment-to-moment fluctuations faithfully
reproduced.”34 Importantly, it was also possible “to amplify certain frequencies
and to diminish or even reject those that were undesirable.”35
Early experiments revealed that for electrical recording to be successful, three
components—a matching high-quality microphone, amplifier, and electromag-
netic cutter (to cut the groove in the wax master disc from which the commercial
record would eventually be made)—must be combined to make a complete system.
The microphone converts sound waves to electrical signals that are amplified,
and the boosted signals drive the electromagnetic cutter to produce the record
master.36 The Western Electric system fulfilled these requirements and also
provided a scientifically designed acoustic gramophone on which to play the elec-
trically recorded disc. Western Electric also incorporated the principle of matched
impedance, a way of connecting two pieces of equipment to avoid any power
waste at the connection.37 By the standards of the day, the resulting sound quality
was truly magnificent. Electrical reproduction followed swiftly with the develop-
ment of electrical gramophones and radios.38 The use of microphones dramatically
changed the face of sound recording and sound recording techniques through the
course of the twentieth century.

Reproducing Piano Rolls

Very early in the twentieth century, pianists were presented with another means
of preserving their art. A remarkable system enabled pitch, rhythm, and tempo,
as well as dynamic nuance and pedaling to be registered by making perforations
onto a paper roll. The roll could be played back on a specially adapted reproducing
piano that used suction created by a pneumatic pump to operate the mechanism.
The ability, at last, to hear playback on a normal sounding piano instead of through
the crackle and hiss of a wax cylinder or shellac disc was very much welcomed. The
reproducing piano system was extremely popular and prized for its advanced

Gelatt, The Fabulous Phonograph, 223.
D. E. L. Shorter and John Borwick, “Recorded Sound” II/1, The New Grove Dictionary of Music and
Musicians, 2nd ed., ed. S. Sadie (London: Macmillan, 2001), vol. 21, 13.
Batten, Joe Batten’s Book, 67.
McWilliams, The Preservation and Restoration of Sound Recordings, 9–10.
Copeland, Sound Recording, 18.
John Borwick, “Electrical Recording 70 Years Old This Year,” ICRC (Nov. 1995): 90.
10 off the record

design and technology.39 It was hailed for its expressivity and ability to capture
successfully the playing of the famous pianists commissioned to record on it.

The Welte-Mignon
One famed system—and the first of its kind—was the Welte-Mignon, patented
by Edwin Welte (1876–1958) and his brother-in-law Karl Bockisch (1874–1952)
in Freiburg im Breisgau, Germany, in 1904. The Welte-Mignon came along just
in time to capture the artistry of some significant pianists who did not make
acoustic recordings for one reason or another. Its recording method was a closely
guarded secret and will probably remain so. The Welte factory in Freiburg, which
housed vital information about the recording process and the recording equip-
ment itself, was severely bombed in 1944.
Much debate continues about the actual process that led to such seemingly
miraculous results. In 1950, Richard C. Simonton, who had personally interviewed
Welte and Bockisch shortly after World War II, gave his impression of the

There was a standard Steinway grand piano, equipped with a trough run-
ning the length of the keyboard and immediately under it. In this trough
there was a pool of mercury, and when the key was depressed, a carbon
rod attached to the bottom of the key engaged this mercury and caused
an electrical contact to be made. The resistance of this contact varied
with the pressure exerted on the carbon rod so that actually, depending
upon the blow with which the key was struck, there was a corresponding
change in the electrical resistance of the contact made. All of the keys
were connected by wires to the recording machine, which was usually
some feet away from the controlling piano. This machine had within it
the conventional rolls of paper which were entirely blank and without
perforation, but were ruled their entire length with over one hundred
fine lines, each corresponding to the center line of its control mecha-
nism. Above the point at which the impression actually took place on the
paper was a series of small rubber rollers of a composition similar to
the type used in a printing press, and these rollers were linked with an
ink similar to that used by the printing industry. The result was that
as the keys of the piano were depressed, these rollers engaged and trans-
ferred their inking to the paper in such a way that, depending upon
the blow or touch exerted upon the keys of the piano, there was a
corresponding difference of the inking of the paper on the master roll.

Larry Sitsky, “Introduction,” The Classical Reproducing Piano Roll: A Catalogue-Index (New York:
Greenwood Press, 1990), vol. 1, xiii.
Early R ecording s 11

Other functions of playing were also transferred, such as pedaling. After

the recording was completed, it was sent to the laboratory and very care-
fully prepared for being used in the reproducing machine, or used in
reverse in order to give a performance and re-create once again the actual
playing of the artist as the roll had recorded it.40

Playback could be achieved using two methods. In the first, the perforated roll was
placed into a piano with an inbuilt Welte-Mignon pneumatic mechanism. In the
second, the perforated roll was placed into a freestanding machine that resembled
a sideboard, specially constructed by Welte and Bockisch.41 This cabinet player, or
Vorsetzer, also had a built-in pneumatic mechanism. As the German name sug-
gests, it sat in front of the piano. According to Simonton, the Vorsetzer was “the
exact opposite of the recording piano,” with felt covered levers (fingers) that
played the keys. In both methods, the perforated roll actuated the pneumatic
mechanism to “re-create once again the actual playing of the artist as the roll had
recorded it.”42
It is most likely that Simonton’s description is an oversimplification of the
process.43 This raises more questions than are answered. According to concert
pianolist Rex Lawson, it is unlikely that Simonton could have known for certain
whether the mercury trough method was actually used, though it probably was.44
And clearly, Simonton does not address some other important issues. These
include the development and improvement of, among other things, the method
of recording dynamics and the role that an editor might have had in such matters.
About this, there is polarization of opinion. Some believe that there was simply
no way of recording dynamics automatically—suggesting the probable existence
of an editor. Others believe that the process for recording dynamics was fully

Other Systems
During the first thirty or so years of the twentieth century, several other major
reproducing piano companies made their mark developing individual approaches
to recording and production. These included the Hupfeld Dea from 1907, Aeolian

Richard Simonton, [Sleeve Notes], Great Masters of the Keyboard, Columbia Masterworks
ML4291–5 (1950): unpaginated.
Mark Reinhart, “The Welte-Mignon Recording Process in Germany,” The Pianola Journal: The
Journal of the Pianola Institute vol. 16 (2005): 3.
Simonton, [Sleeve Notes], unpaginated.
Denis Hall, “Recording Welte-Mignon Piano Rolls in Germany,” The Pianola Journal: The Journal
of the Pianola Institute vol. 16 (2005): 50.
Rex Lawson, pers. comm., in conversation.
Reinhart, “The Welte-Mignon Recording Process,” 18.
12 off the record

with its Duo-Art from 1914, and Ampico from 1912, as well as others such as
Artrio-Angelus and Philipps Duca. In all systems, the finished roll preserved a
combination of very precise information registered automatically during the
recording and other information edited in afterward.

The Duo-Art
The Aeolian Company officially launched its Duo-Art system in New York in
1914. Eventually, both the American and the British Aeolian companies manu-
factured Duo-Art pianos and rolls.46 In 1924, Reginald Reynolds—Duo-Art’s
British recording producer—gave a succinct description of the note recording

In a secluded room stands a Weber grand piano, in tone and in outward

appearance not different from the usual model, nor does the touch betray
the magic power beneath the keys. Upon closer inspection the secret is
partially revealed by the electric cable which can be seen coming from
beneath the instrument; and if it were possible to trace this back into
the piano, there would be found 160 wires, half of them leading to
specially devised contacts under the keys, the remainder running to posi-
tions near the point where the hammers strike the strings, while the
cable itself passed through the wall of the room, coming out into a sound-
proof chamber, in which is installed the amazing mechanism that consti-
tutes the Duo-Art recording apparatus. Here the other ends of the
wire are attached to electro-magnets, which operate the punches in the
powerful perforating machine, each punch corresponding with each key
of the piano. The pianist plays—the punches perforate—the record is

Dynamics (expression details) were registered by the recording producer and cut
into the rolls while the recording took place. At a later stage, the dynamics were
edited under the supervision of both producer and recording pianist.48

The Ampico
The American Piano Company, established in 1908, was to become Aeolian’s main
competitor. Recognizing the commercial advantage of the reproducing piano,

Denis Hall, “Duo-Art Rolls: A Description of Their Production and an Assessment of Their
Performance,” The Pianola Journal: The Journal of the Pianola Institute vol. 10 (1998): 5.
“Player Piano Supplement,” The Gramophone (Feb. 1924), in Hall, “Duo-Art Rolls,” 11.
Hall, “Duo-Art Rolls,” 11.
Early R ecording s 13

the company invested in the talent and invention of Charles Stoddard, a young
engineer. Stoddard set about designing and patenting a reenacting piano that
“included not only the player mechanism, but also the recording system, and the
design and production of the accompanying music rolls.”49 Stoddard’s system—
eventually called the Ampico system—was remarkable in its sophistication and
developed out of two patents. As explained by Lawson, in the first patent designed
for an upright piano, two thick contact wires attached at the base of a piano
hammer shank are able to make electrical connections with mercury baths. As
long as the hammer is at rest or striking a string, an electrical circuit is incom-
plete. But “during the brief forward movement of the hammer as the key is
depressed, and also after the hammer has rebounded and is in its check position,
both wires and both mercury baths are in contact, thus allowing an electrical cir-
cuit to be completed.” This causes a stylus to press down on the recording roll
“during the travel of the hammer to the string, and also after the note has been
sounded, as long as the key is held down.” The result is a series of short lines—the
lengths of which are in inverse proportion to the dynamic level of particular notes,
followed by a series of longer lines—more or less equivalent to the duration of the
notes. The second patent is an adaptation of the first. It still employs mercury
baths but was designed for a grand piano. Here, however, note durations and
dynamics are recorded on two separate rolls, “the dynamic roll running at a higher
speed, thus allowing greater accuracy to be obtained.”50 During the life span of the
Ampico system, dynamics were recorded by different means and edited after the
initial recording.
The reproducing piano system had peaked in popularity by 1925 and to all
intents and purposes was finished by 1930. Its demise was hastened by the grow-
ing popularity of the radio and electric gramophone, and finally the great stock
market crash of 1929, after which few could afford to invest in luxury items such
as a reproducing piano.

Limitations of Acoustic Recordings

Recording Singers
Written accounts illustrate the somewhat extraordinary conditions that surrounded
early acoustic recording sessions. Fred Gaisberg, the first recording engineer and
talent scout for the Gramophone Company in 1898, had the honor of recording
on disc many great artists of the era including the celebrated soprano Adelina

Rex Lawson, “Cleaning the Windows of Time,” The Pianola Journal: The Journal of the Pianola
Institute vol. 12 (1999): 15.
Lawson, “Cleaning the Windows of Time,” 15–16.
14 off the record

Patti (1843–1919). During her first recording sessions in 1905 she underwent the
ordeal of singing into a small funnel while remaining stationary. According to
Gaisberg, her Italian nature and temperament produced “flashing movements” as
she acted out the role. He had “to pull her back” when she sang high notes with
“beautiful attacks.” This made her most indignant at first.51 Magnificent as they
were, Patti’s heartfelt characterizations were simply too much for the recording
Gaisberg was no doubt safeguarding against cataclysmic damage to the wax.
This was probably the type that a young chemist once humorously related to
Edison. “When an extra-loud sound occurs in a song—you know, when an
Eyetalian has suddenly fallen in love or somep’n—the recorder needle gives a
jump, and then a tiny bit of wax is chipped out.”52 If this occurred, the recording
had to start over again. The effort of being careful had somewhat of a restraining
effect on the artist. While collecting folk songs on a gramophone in 1907, Percy
Grainger (1882–1961) was told by one of the folk singers that singing for the
gramophone was like singing “with a muzzle on.” Nevertheless, he sang to the
best of his ability.53 Patti, too, gave her very best and was pleased with the results.
She wrote enthusiastically to her nephew the pianist Alfredo Barili (1854–1935),
“My voice and phrasing come easy and simply perfect out of the instrument and
I think the company will make a fortune.”54
Recording methods in these early sessions could be rather crude. Artists
became agitated by the unusual conditions of the recording room. Capturing
dynamic shading and accentuation was most problematic. Louise Barili—Alfredo
Barili’s daughter—witnessed one of the sessions in 1906 at which her father
accompanied Patti.55 She describes how Patti became understandably nervous
as she balanced on a movable platform, which was shunted backward and for-
ward from the funnel to create dynamics.56 But although this procedure might
have ensured some dynamic variation, it could also lead to artificial results.
Percy Grainger noted that in order to preserve the dynamics truthfully, the sing-
er’s mouth must be positioned at “a practically unvarying distance from the

Gaisberg, The Music Goes Round, 91.
M. A. Rosanoff, “Edison in His Laboratory,” Harper’s Magazine (Sept. 1932), in Ronald W. Clark,
Edison: The Man Who Made the Future (London: Macdonald and Jane’s 1977), 167.
Percy Grainger, “Collecting with the Phonograph,” Journal of the Folk-Song Society, vol. 3, no. 12
(1908): 147.
John F. Cone, Adelina Patti: Queen of Hearts (Portland, Ore.: Amadeus, 1993), 244; Alfredo Barili
was born in Florence into the famous Patti-Barili family of musicians. He was one of the first profes-
sional musicians to move to Atlanta, where he supported a wide range of music making and activities,
particularly through setting up a successful school of music.
Gaisberg, The Music Goes Round, 92.
Cone, Adelina Patti, 246.
Early R ecording s 15

recording trumpet.” Being moved alternatively toward and from it could lead to
arbitrary “louds and softs.”57

Recording Pianists
Pianists, too, found themselves in unusual predicaments. Barili became agitated
because he was placed with his piano on boxes towering above Patti. From this
position, he was instructed to play without dynamic shading because such subtle-
ties would not be picked up.58 The pianist Joe Batten (1885–1955)—recording
manager of the Musiphone Company—confirms such practices for recordings
made around 1900:

Most of the space of the room . . . was occupied by an improvised ros-

trum, five feet in height, upon which an upright grand piano had been
hoisted to enable its soundboard to be on a level with the recording horn.
The back and front of the piano had been removed, so that the maximum
of sound could be obtained, thus leaving only the action and soundboard.
There was no music stand, the music being held up by hand by anybody
who had nothing else to do at the time. . . . I hammered out the accompa-
niments. Dan Smoot [the recording engineer] had demanded of me to
make the tone “double forte” and double forte it was. From time to time
the singers whispered appeals to “keep it down.”59

Although playing loudly was deemed necessary, pianists had to be careful of

making extra loud sounds or violent accents. These could result in crashes, which
Hambourg described as “the worst pitfalls, the ruin of many otherwise fine
gramophone performances, nullifying hours of hard work.”60
In 1962, the English pianist and accompanist Gerald Moore (1899–1987)
recounted his first recording experiences. Apparently, even as late as 1921 he was
instructed to play with unvarying dynamics and especially not to play softly. The
recording engineer insisted on his playing forte consistently throughout even
though the work to be recorded was gentle in character—a berceuse. His initial
protests resulted in him not being heard in the test recording. But the piano could
not be brought closer to the recording horn. In the end he did what he was told,
clattering his part of the lullaby “like a charge of cavalry” to everyone’s approval.61
It stands to reason that, on that particular occasion, Moore might have been

Grainger, “Collecting with the Phonograph,” 147.
Cone, Adelina Patti, 246.
Batten, Joe Batten’s Book, 33
Hambourg, From Piano to Forte, 291.
Moore, Am I Too Loud? 53.
16 off the record

forced to play more consistently loudly because the piano was necessarily placed
further from the recording horn than the violinist he was accompanying.
Later in 1983, Moore expressed his shock at what he considered an excess of
“passionate fervour” in some of Paderewski’s acoustic recordings. He concludes
that Paderewski must have been forced to play with a “consistently penetrating
forte” in order to record successfully onto the soft wax. Moore explains that this is
what he himself was forced to do in the early 1920s “when to make a diminuendo
or attempt to play softly reduced the recording engineer to despair.”62 Be that as it
may, Moore’s opinion of Paderewski’s recordings should not be taken at face value.
Paderewski’s recordings of Chopin’s Mazurka Op. 17 No. 4 made in 1911 and 1912
(2008 Appian transfer), for example, reveal that the dynamic levels are anything
but consistently penetrating. Indeed, the dynamics sound both varied and clear,
though obviously not to the extent that we expect to hear in a recording now.
Perhaps Moore was simply shocked by the lack of dynamic range (particularly at
the soft end) between Paderewski’s disc recordings and his live performances.
Or perhaps he simply did not appreciate Paderewski’s interpretation.
The difficulty of recording dynamic shading, and therefore other nuances such
as accentuation and articulation, was a matter that naturally concerned many
pianists. But in the early years, there was also concern about the state of the
recording piano. Speaking of the late 1890s, Gaisberg reminisced that a halfway
decent result was more likely “if the instrument was old and had a tinny, shallow
tone.”63 In 1909, Hambourg recalled playing on an “old tin kettle of a piano” during
his first recording for the Gramophone Company. It was “quite the worst-toned
one” that he had ever played. And although it was a great trial to play, the piano in
question produced by far the best results.64
To help capture the sound of the recording piano, the studio itself sometimes
had particularly severe characteristics. Moore complained about the sterile sur-
roundings and the overresonant acoustic necessary for maximum impact on the
vibrating diaphragm. But he was particularly shocked by the metallic stridency of
the piano tone, which he claimed “had been rendered as percussive as possible by
the filing down of the felts on the hammers.” To him the piano sounded like a
“brass spittoon.”65 It is possible that Moore was exaggerating or playing for laughs
in describing such seemingly drastic modifications to the piano hammers.
Certainly such changes, if in fact they were made, must have been confined to
particular studio recordings. Barili would almost certainly have used Patti’s piano
for accompanying her at her Welsh home during the 1906 sessions. With her

Gerald Moore, Furthermoore: Interludes in an Accompanist’s Life (1983, reprinted in Collected
Memoires, Middlesex: Penguin, 1986), 396.
Gaisberg, The Music Goes Round, 8.
Hambourg, From Piano to Forte, 288.
Moore, Am I Too Loud? 52.
Early R ecording s 17

wealth, the piano is unlikely to have been “a clunker.” Even through the crackle
and hiss, it sounds like a nice instrument (1988 GEMM transfer). And in 1911,
Paderewski played his own Erard concert grand piano during recording sessions at
his villa in Morges, Switzerland. This must surely have been a piano in fine condi-
tion, and the recording itself supports this (1998 Philips Classics transfer). It is
also well known that Paderewski was extremely particular about the instruments
on which he played. By 1920, in any case, pianists could record on normal grand
pianos relatively free from modification. This was due to improvements in record
manufacturing (which led to a reduction in surface noise) as well as better, more
efficient gramophones.66
In the very early years up until about 1900, playback also caused a fair amount
of consternation. It was well nigh impossible for wax recordings to reproduce a
good, natural piano tone. For Batten, the impression was that the hammers were
striking “tin cans instead of strings.”67 Hambourg felt that the reproduced tone
was “thin and tinny like the plucked string of a banjo or guitar.”68 And this was an
opinion shared by others.69 Sergei Rachmaninoff (1873–1943) described a thin,
tinkling piano tone, which conjured up “the Russian balalaika.”70 In general, louder
works were more successfully reproduced than soft works. And slow movements
requiring legato touch “were unobtainable.”71
Pedaling was yet another aspect that was difficult to record via the early acous-
tic process. Given the need to play at a high dynamic level, the una-corda, or
soft pedal, was either prohibited or used sparingly. The sustaining pedal contrib-
uted to a blurring of the already disadvantaged piano sound in the overresonant
acoustic of the recording studio. Upright pianos were generally used in studio
recording sessions, during which it was seen as an imperative to wedge or jam the
pedals to stop them from being operated.72 Even as late as 1919, Feruccio Busoni
(1866–1924) was told by Columbia to watch how he used the pedal because it
sounded bad.73 Nevertheless, there are many examples in which pianists appear
to have pedaled fairly normally. Raoul Pugno (1852–1914) uses the sustaining
pedal in his 1903 Paris recordings for the Gramophone and Typewriter Company
(2008 Marston transfer). And Louis Diémer (1843–1919) uses the sustaining
pedal quite normally for his 1904 G & T sessions in Paris. But for his 1906 session

Copeland, Sound Recording, 15.
Batten, Joe Batten’s Book, 37
Hambourg, From Piano to Forte, 288–89.
Frank Swinnerton, “A Defence of the Gramophone,” The Gramophone (Aug. 1923): 10.
Sergei Rachmaninoff, “The Artist and the Gramophone,” The Gramophone vol. 8, no. 95 (April
1931): 525.
Hambourg, From Piano to Forte, 288–89.
Ford, “History of Sound Recording,” 228.
Ferrucio Busoni, [Letter to his Wife November 20, 1919], Letters to His Wife, trans. Rosamund
Ley (London: Arnold, 1938), 287.
18 off the record

(2008 Marston transfer), he appears to have been told not to use the pedal—and
to have obliged. Alfred Grünfeld (1852–1924) uses a fair bit of pedal on his 1899
seven-inch Berliner discs (1993 OPAL transfer).74 In the end, it is probably safest
to judge the pedaling on a case-by-case basis.

Recording Orchestral Instruments

Although the main focus of this book is practices in piano playing, some discus-
sion about the recording of orchestral instruments and orchestras provides a
broader view of the early recording process. The full tonal spectrum of string
instruments was difficult to capture in the early acoustic process because their
sound was among the least powerful. Batten described how orchestral accompani-
ments sounded like “muted strings against blatant brass.” Violins, violas, cellos,
and basses often came across as “a pathetic and ghostly murmur.”75 The effect of
muted strings was not, therefore, a viable consideration.76 To give the violin a
more penetrating sound, Augustus Stroh (1828–1914) invented his special “Stroh”
violin in 1904. He substituted the sound box with a diaphragm (that was attached
to the bridge) and metal horn similar to that of an acoustic gramophone. The
sound of the violin could now be projected through the metal horn, which itself
was raised or lowered to focus the sound in the direction of the recording horn.77
A single Stroh violin would sometimes represent the entire string section. In this
context (and for fairly obvious reasons), the pizzicato, glissando, and vibrato char-
acteristics of the instrument had to be heavily exaggerated in order for the instru-
ment to be heard.78 Another means used occasionally to create impact was, as
German composer and scholar Max Chop (1862–1929) noted in 1909, to replace
the higher strings with high-pitched winds such as flutes, clarinets, and trumpets.
However, he considered this to be wholly unmusical and truly objectionable.79 It is
important to remember that these recording practices were typical of the early
acoustic recording era, not the later one. Things changed as time went on.
The recording of larger ensembles also posed problems. Orchestras often had
to be modified in size, layout, and instrumentation in order to fit into the record-
ing studio and to produce enough sound. Scores were arranged to suit and most
often the scoring was for tutti orchestra throughout. This was simply overwhelm-
ing in the cramped studio space. In spite of this, not much of the din registered

Denis Hall, pers. comm., letter, Nov. 7, 2008.
Batten, Joe Batten’s Book, 35
Herbert C. Ridout, “Behind the Needle—V: Looking over Forty Years of the Gramophone,”
The Gramophone (Nov. 1940): 131.
Batten, Joe Batten’s Book, 35.
Gaisberg, The Music Goes Round, 41.
Gelatt, The Fabulous Phonograph, 175.
Early R ecording s 19

onto the wax.80 The volume, timbre, and sound direction of different instruments
meant that both a successful balance and dynamic contrasts were difficult to
record. Ridout described a typical recording session with a singer accompanied by
an orchestra, the sort he is likely to have witnessed from 1906 onward. It was
indeed a comic scene:

The French horns, having to direct the bells of their instruments towards
the recording horn, would turn their backs on it and were provided with
mirrors in which they could watch the conductor. The tuba was posi-
tioned right back from the [recording] horn and his bell turned away
from it; he also watched in a mirror. The big drum never entered a record-
ing room. For a flute obbligato the flautist would leave his seat, dash
round and take his place alongside the singer, and then rush back to his

To maximize sonority, lower woodwinds and brass instruments (contrabassoons

and/or tubas) sometimes augmented or replaced cellos and double basses. Chop
viewed this as a necessary evil that did not alter the total sound pattern too

Limitations of Frequency Range

The acoustic recording process was beset by another serious problem. Only a fairly
limited range of frequency (100–4000 Hz)83 compared with the range audible by
the human ear in a concert hall (20–20,000 Hz) could be recorded and reproduced.
On the most successful acoustic recordings, therefore, the range might well be
limited to notes from the “E below middle C to three octaves above middle C.”
Additionally, because many of the low frequencies and harmonics are missing,
the character and timbres of many notes sound distorted. Although the note
itself remains recognizable and its pitch is apparently unaltered, the result is a
somewhat metallic quality to the sound.84
Clearly, the early acoustic recording process came with its problems and incon-
veniences. Solo voices came off best. Piano recordings yielded mixed results, and
orchestras were the least successful. Very quiet sounds were simply impossible to
capture.85 But as time went on, the situation improved dramatically. Many early

Batten, Joe Batten’s Book, 36.
Ridout, “Behind the Needle,” 131.
Gelatt, The Fabulous Phonograph, 175.
Opinion is varied about these figures.
Gelatt, The Fabulous Phonograph, 204.
Timothy Day, A Century of Recorded Music: Listening to Musical History (New Haven, Conn.:
Yale University Press, 2000), 9.
20 off the record

recordings—even orchestral ones—are decent enough to be enlightening. Listen,

for example, to the 1913 recording for HMV of Beethoven’s Symphony No. 5
by Arthur Nickisch (1855–1922) and the Berlin Philharmonic Orchestra (1991
Symposium Records transfer). Nickisch’s interpretation, with its rhetorical phras-
ing, graded dynamic shadings, and tempo modifications, comes through clearly.

Limitations of Recording Length

Nowadays, we rarely have to worry about the length of a work during a recording
session. But this was a factor needing special consideration in recording sessions
right up to 1950, particularly when the length exceeded available recording space.
Before 1900, four-inch wax cylinders could record approximately 2 minutes of
music. This was doubled to approximately 4 minutes when Edison introduced his
“Amberol” (celluloid) cylinder in 1908. Before 1900, single-sided five-inch and
seven-inch wax discs could record between 1.5 and 2 minutes of music, approxi-
mately. By 1901 this was increased to between 2.5 and just over 3 minutes for a
ten-inch disc side and, by 1903 up to 4.5 minutes for a twelve-inch disc side. In
extreme cases a little more time could be gained by narrowing the grooves on the
disc, but this was less than satisfactory: “the quality of tone deteriorated when the
needle approached too near to the centre.”86 By 1904 “long-playing” cylinders
were developed, albeit very scratchy-sounding ones that recorded up to 8, 10, and
12 minutes. And there were many other attempts up until the 1930s at increasing
the length of discs. This yielded anything between 12 and 20 minutes of recording
time. In most, however, the sound quality remained problematic.87
Many musicians found such limitations very frustrating. Hambourg recounted
telling Gaisberg that his desire was to record the masterworks of piano repertoire,
not merely “salon” pieces. He acknowledged the great difficulty of overcoming
“the problems of timing, speed management, and tone quality.”88 A variety of
methods were developed to enable accessibility to standard repertoire via the
acoustic recording process. In this respect, the reminiscences of the English
pianist Harold Bauer are revealing. In preparation for recording Beethoven’s
Moonlight Sonata Op. 27 No. 2 in 1926 for the Victor Company in New York, he
was understandably worried about the recording length:

Was the first movement not too long and too slow to be recorded on a
twelve-inch disc (the largest size)? I played it through and it took just
over five minutes. The limit was four minutes and forty seconds. I tried it

Moore, Am I Too Loud? 54–55.
Robert Dearling, Celia Dearling, and Brian Rust, The Guinness Book of Recorded Sound (London:
Guinness, 1984), 190.
Hambourg, From Piano to Forte, 290–91.
Early R ecording s 21

faster and did not like it. I thought of making a cut . . . horrible! I thought
of playing it in two sections . . . equally horrible!89

In the end, Bauer recorded the piece in just four minutes and thirty-seven seconds.
But I must explain here that his reason for playing it faster had nothing to do with
recording time limitations. In his search for a solution, Bauer discovered that
Beethoven had marked the time signature as alla breve rather than common time
(the signature given by many editors). Instead of four pulsations to the bar, there-
fore, he experimented with two. This obliged him to play faster, but resulted in the
rhythmical effect being slower.90
For longer repertoire, musicians sometimes opted to make cuts. Hofmann’s
1923 recording for Brunswick of an abridged version of Chopin’s Scherzo No. 1 in
B Minor Op. 20 lasts four minutes and thirty seconds. His unabridged version
of the same work on a 1926 Welte-Mignon piano roll lasts eight minutes and
six seconds. The tempo in both versions is very similar. So, although a good deal
of the music was missing from his acoustic recording, the cut allowed Hofmann
to play at his normal tempo.91 Some considered having to make cuts to be the
“gravest defect of all.” Far too often the music was “ruthlessly cut.”92
An alternative to cutting was to spread longer works (sonata or symphony
movements) over multiple record sides. This entailed stopping periodically
through the movement, occasionally “on an unresolved discord.” Although this
option saved having to make cuts, it was considered by some to be simply unbear-
able.93 Naturally, dividing movements in this way disturbed the continuity. In
Hambourg’s opinion, the public’s preference was for a complete work to be
recorded on “one side of a disc only,” a fact that on occasion forced performers to
play faster than desired or musically sensible to fit into the time constraint.94 This
is an important point to consider before we take the tempo of some recordings
too literally.
Yet despite being issued on several discs, extended works were sometimes
severely truncated. Examples of these include the following: Beethoven’s Eroica
Symphony conducted by Sir Henry Wood (1922)—half the music missing; Elgar’s
Violin Concerto conducted by Elgar (1916)—two-thirds of the music missing; and
Elgar’s Cockaigne Overture conducted by Elgar (1917)—over two-thirds of the
music missing. Undoubtedly an extremely cruel task, cuts were often made in

Bauer, Harold Bauer, 268.
Bauer, Harold Bauer, 269.
Day, A Century of Recorded Music, 7.
Swinnerton, “A Defence of the Gramophone,” 10.
Moore, Am I Too Loud? 54.
Hambourg, From Forte to Piano, 290.
22 off the record

order to ensure that the record was commercially viable.95 In these cases, profit
margins were seen as more important than preserving the musical sense.
But with the release in 1913 of two complete sets of Beethoven’s symphonies,
gramophone companies both in England and Germany started to record large-
scale works in complete form. And some of these were highly praised for the qual-
ity of the sound reproduction as well as the preservation of original orchestration.96
Before the introduction of electrical recording in 1926, there was an explosion of
recordings of full orchestral works of the great masters.97 For example, Elgar
recorded for HMV his Enigma Variations Op. 36 during 1920 and 1921, and his
Second Symphony in 1924, both complete. Polydor recorded Mahler’s Second
Symphony complete in 1924.
Although longer works or movements might sometimes have been played more
swiftly, shorter works—songs, arias, and instrumental movements—did not have
to undergo such treatment. These were more suitable both artistically and eco-
nomically for all 78 rpm records, both acoustic and electric.98 The proof that these
works were not accelerated is seen in the fact that many shorter compositions
recorded both on disc and on piano roll take practically the same time. For example,
in 1919 Busoni made an acoustic recording of Chopin’s Prelude Op. 28 No. 7 for
Columbia (1989 Pearl transfer), which takes approximately fifty-nine seconds.
Compare this with his 1923 Duo-Art piano roll (1997 Nimbus Records transfer),
which takes one minute and five seconds. Edvard Grieg’s (1843–1907) acoustic
recording (1903) of his Bridal Procession Op. 19 No. 2 for G & T in Paris takes three
minutes (2008 Marston transfer), whereas his 1906 Welte-Mignon piano roll
takes three minutes and five seconds when the roll is played at the right speed.99
Saint-Saëns’s 1919 acoustic recording of his Valse mignonne Op. 104 (2008
Marston transfer) takes two minutes and twenty-six seconds, compared with his
1905 Welte-Mignon piano roll, which takes two minutes and nineteen seconds
(2008 TACET transfer). Clearly, the variations here are negligible. In the case of
shorter piano works, 78 rpm disc recordings generally preserve the normal tempo
intentions of the artist.
Recording time limitations added a significant level of stress to a process that
was (and still is) a somewhat unnatural activity. In 1919, Busoni complained to
his wife about having to record the Gounod-Liszt Faust waltz, which normally
took about ten minutes, in four minutes. This entailed “quickly cutting, patching
and improvising” to preserve the musical sense, as well as being careful with the

Day, A Century of Recorded Sound, 7–8.
Gelatt, The Fabulous Phonograph, 181–82.
Gelatt, The Fabulous Phonograph, 204–5.
Day, A Century of Recorded Sound, 8.
Hall, pers. comm., letter, Nov 7, 2008; note that the 1934 Odeon transfer appears to be faulty
in terms of speed.
Early R ecording s 23

pedal and with the accentuation of particular notes so that they would register
onto the wax. The process of cutting was undoubtedly troubling, though presum-
ably it saved Busoni from having to modify the overall tempo. But his suffering
was far greater because he felt unable to let himself go “for fear of inaccuracies.”
And he was conscious throughout that every note would be there forever.
Exasperatedly, he questioned whether under these conditions his playing
expressed “inspiration, freedom, swing or poetry.”100 No doubt, his playing did
still have these qualities—they were after all his hallmark. But clearly, the
worry about both perfection and future accountability was felt by him to be too
pervasive. On some level, recording was hampering his artistry. The Faust waltz
recording was never issued commercially.
But not all performers disliked the recording process. For some, the gramo-
phone was a major factor in the success of their careers. This was certainly true of
bass-baritone Peter Dawson (1882–1961) and tenor John McCormack (1884–
1945), both of whom became known worldwide for their best-selling recordings.
Indeed, they had much bigger gramophone reputations than on the concert
platform. Elgar, too, is known to have loved conducting for recordings.
Whatever the perceptions of and the difficulties experienced by musicians
during the acoustic recording era, things changed dramatically once electrical
recording became possible. Having adapted to the requirements of the acoustic
process, musicians had to reassess their recording philosophy in front of the
microphone. Now they were required to perform as highly sophisticated musi-
cians, “with refinement, with light and shade, with delicacy of nuance” and
even “very softly when necessary.”101 The microphone picked up everything, and
recordings could now be edited. Such factors were to have a profound effect on
performing practices, accelerating the change in style and attitude during the
course of the twentieth century.

Early Recordings: Summary

Given some of the limitations discussed earlier, certain allowances have to be
made in using early recordings (particularly preelectric ones) as yardsticks for
judging the performances of revered concert artists. As Philip concludes, they are
“a partial representation of what the musicians would have achieved in concert
performance, adapted to suit the limitations of the recording machinery of the
day.”102 It must be acknowledged, however, that some musicians fared much better
than others. In any case, much important information about particular playing

Busoni, Letters to His Wife, 287.
Moore, Am I Too Loud? 56–57.
Philip, Performing Music in the Age of Recording, 28.
24 off the record

styles can be gleaned from these recordings. In this respect, music critic Harold C.
Schonberg felt that an experienced collector would know “how to compensate,
intellectually and aurally, for the low-fidelity sound of acoustic piano discs.”
Indeed, his assessment is that “as much can be derived from them as from the
high fidelity of modern recording.”103 After much listening, I have adopted a simi-
larly enthusiastic outlook. Despite the difficulties associated with the acoustic
recording process, many performing practices both general and idiosyncratic are
well preserved in these representations. For me, early recordings are invaluable.

Reproducing Piano Roll Recordings:

Endorsements and Praise
During the first thirty years of the twentieth century, piano rolls were considered
by many to preserve the playing of revered pianists more successfully than acous-
tic recordings. Welte-Mignon often received glowing testimonials from its record-
ing pianists. For Reinecke, it was difficult to believe that the artist himself was not
performing. Claude Debussy (1862–1918) vouched for its ability to attain the
greatest perfection of reproduction. Hambourg was impressed with how it could
reproduce the finest shadings of touch. And Hofmann enthused that with it, the
pianists’ art could now live forever.104 Similar statements were to be found in the
catalogues and advertisements of other companies. Indeed, the companies them-
selves often made great claims about the results that could be achieved. In 1924,
the Aeolian Company described the conscientious and painstaking collaboration
between its artists and technicians. This, it assured, successfully captured the
essence of the recording pianists’ genius and skill with results that “are beautiful
and characteristic beyond the highest hopes of all concerned.”105 The Welte
American Licensee stated in 1927 that the notes and expression—in short, every-
thing characterizing the pianists’ individuality—were faithfully reproduced.106
The value of such endorsements is questionable. The recording pianists were
often requested or contractually obliged to provide testimonials. Some may have
felt pressure to be excessively positive. Others “may have signed on the dotted
line immediately after the recording without ever hearing their roll played back.”107
Be that as it may, there is certainly truth in a proportion of the claims. Grieg
was very excited by what he heard on the Phonoliszt—an “expression” piano made

Harold C. Schonberg, “The Ampico-Argo Piano Rolls,” The Gramophone (Dec. 1966): 308.
Library of Welte-Mignon (Licensee) Music Records [Catalogue] (New York: De Luxe Reproducing
Roll Corporation, 1927), 181, 53, 84, 93.
Catalogue of Music Rolls for the Duo-Art Reproducing Piano (New York: Aeolian Company,
1924), 5.
“Foreword,” Library of Welte-Mignon Music Records, unpaginated.
Holliday, Reproducing Pianos, 102.
Early R ecording s 25

by Hupfeld.108 He commented in his diary in 1906 that what it achieves “is incred-
ible.” Liszt’s “Rhapsody” played by Alfred Reisenauer (1863–1907) “was really like
listening to Reisenauer himself.” Grieg cherished the prospect of hearing his own
works replayed by this means.109 It is of course possible that some of Grieg’s
unbridled enthusiasm for the contraption was in part due to the novelty of the
new technology.
Other accounts suggest that the reproducing piano method truly preserved the
pianists’ playing. In 1929, a reviewer from the Musical Times was thoroughly
impressed that Ampico rolls seemed able “to produce every possible degree of
tone-colour, phrasing etc.” This is done in such a manner that “the artist’s playing
is almost uncannily faithful.”110 Indeed Ampico’s seemingly remarkable results
were demonstrated in test or comparison concerts held in Europe and America
that received rave reviews.111 In 1920 one such concert took place in New York’s
Carnegie Hall during which, according to the critics, the results were quite
extraordinary. This took the form of a joint recital that boasted five of the
world’s greatest living pianists—Leopold Godowsky (1870–1938), Mischa Levitzki
(1898–1941), Benno Moiseiwitsch (1890–1963), Leo Ornstein (1893–2002) and
Arthur Rubinstein (1887–1982)—“in which their playing was heard in direct
comparison with its repetition by the Ampico.”112 Reportedly, the critics found it
hard to distinguish man from machine. According to Philip Hale of the Boston

The Ampico piano made remarkable reproductions of the characteristic

interpretations of these pianists. At times the pianist would stop play-
ing; the Ampico would go on as if he were not idle. Then the pianist would
again take up the wondrous tale. In some instances the performance by
the Ampico of the whole composition was identical with the original; in
other instances the Ampico surpassed what had gone before, and thus
did justice to the pianist when he had fallen below his own standard.
Especially noteworthy were many charming nuances. Not for a moment
was there any suggestion of rigid, inflexible, purely mechanical mimicry.

This was “an expression piano powered by an electric suction pump, with three levels of
automatic dynamics, and variable speed crescendos between the levels.” See
reproducing/reproducing_dea.cfm; see also Holliday, Reproducing Pianos, 72.
Edvard Grieg, Diaries, Articles, Speeches, ed. and trans. Finn Benestad and William H. Halverson
(Columbus, Ohio: Peer Gynt Press, 2001), 116; Alfred Reisenauer (1863–1907) was a German pianist,
composer, and music educator who had studied with Liszt and led an active concert career. He also
recorded for the Welte-Mignon in 1906.
Anon., “Player-Piano Notes,” The Musical Times vol. 70 (1929): 905.
Holliday, Reproducing Pianos, 103–5.
Purchaser’s Guide to the Music Industries (1926) in Sitsky, “Introduction,” xxiv.
26 off the record

The playing was as free, elastic, spontaneous as though a gifted mortal

were the performer.113

The Duo-Art system was at least as positively received during its life span as the
Ampico system. Duo-Art’s recording expert in London, Reginald Reynolds, speci-
fied that the system was “primarily intended for the reproduction of records by
celebrated pianists.” Its accuracy resulted in “perfect reproduction of the most
subtle rhythm.”114 Percy A. Scholes, a musician and scholar of great repute, was
very positive that Bauer’s 1918 Duo-Art roll of the first movement of Beethoven’s
Appassionata Sonata Op. 57 came very close to Bauer’s playing. He suggested humor-
ously, “If you want something a shade nearer still, you can pay Bauer 10,000
[pounds] a year to come and live in your house. If you cannot afford this, the thing
is to have this new ‘Duo-Art’ Roll. It is a very, very near thing to Bauer himself.”
Scholes also explains that many listeners were unable to tell when Alfred Cortot
was playing and when his role was being played by the Duo-Art.115
Busoni recorded for the Duo-Art in 1915 and 1921. Of the 1921 recordings,
one of his students expressed amazement with the results: “These rolls are not
merely reproductions—they are Busoni himself.”116 However, Bauer and others
were critical of these particular rolls. This may have been, as Hall explains, because
they were not recorded in America, and therefore not edited properly. They would
otherwise have been a good representation of Busoni’s playing.117 We will have to
make up our own minds about their success and will return to the important issue
of roll editing later.
Roll recording provided an attractive alternative to acoustic recording because
the sound production in playback was beautiful and realistic. The Musical Times
reviewer who raved about Ampico rolls (cited earlier) was also very enthusiastic
about the quality of their playback. Robert Schmitz’s (1889–1949) Ampico roll of
Debussy’s “Jardins sous la pluie” is singled out for its beautiful reproduction.
According to the reviewer “this leaves even the best gramophone records of piano-
playing [sic] far behind, if only on the score of tone quality—the weak point of the
gramophone where the pianoforte is concerned.”118 Others too admired the better
tone quality in roll reproduction. “How much more successful a Pachmann roll is
than a Pachmann gramophone record!” exclaimed another reviewer from the
Musical Times. “In the records the tone is usually bad, whereas, presuming the

Sitsky, “Introduction,” xxv.
Reginald Reynolds, “On Playing the ‘Pianola,’” in Percy A. Scholes, The Appreciation of Music by
Means of the Pianola and Duo-Art (Oxford: Oxford University Press, 1925), 149.
Scholes, The Appreciation of Music, 27.
Anon., “Busoni: [by] One of His Pupils,” The Pianola Journal: The Journal of the Pianola Institute
no. 10 (1998): 61–62.
Hall, pers. comm., letter, Nov. 7, 2008.
Anon., “Player-Piano Notes,” 905.
Early R ecording s 27

player piano to be a good instrument, it is impossible to get a bad tone from a

roll.”119 Rather humorously, the reviewer was also pleased that the roll had not
recorded Pachmann’s audible mutterings, which were often rather distracting in
live concerts.
With such positive reinforcement, it would be easy to sweep away any qualms
about the ability of a reproducing piano system to do what its companies and art-
ists claimed. But there are other factors that have bearing on the truth. For exam-
ple, what or how the pianist actually played during the recording, and what was
changed or added through editing afterward must be considered. But perhaps an
even more important consideration is the extent to which the integrity of the
aural effect is compromised when the conditions for playback of the rolls are
at variance with those that were originally intended. These two factors alone
will have some impact on the value of piano rolls for the study of performing

Limitations of Piano Roll Recordings

The technology for producing piano rolls developed in a remarkable way during
the course of the nineteenth century. But it was during the very early years of the
twentieth century that the recording of live piano playing was perfected. All roll
masters (those from which others were copied) contained several layers of infor-
mation. Much of the essentials—the note pitches, positions, and durations as
well as subtle speed variations and the use of the sustaining and soft pedals—
were recorded automatically.120 This is also true of the agogics—the momentary
note lengthenings and expressive breathing points essential for any artful inter-
pretation.121 Experts agree that all of this information is extremely accurate.
But dynamic shading was not recorded automatically. Information about pedaling
and dynamics appeared as extra perforations down both sides of the roll. The
reliability of this information, however, is open to question.
Modern playback of rolls as heard on commercial transfers has sometimes met
with harsh criticism. Typically, dynamics and accentuation are found to be faulty,
dysfunctional, or unnatural, and pedaling is labeled as suspicious and inexact.
In 1996, Gregor Benko—cofounder of the International Piano Library—was
scathing of those who made the transfers of specific Duo-Art rolls on the Nimbus
label. To him the results are “insensitive distortions of the noble playing of the

Anon., “Player-Piano Notes,” 135.
Denis Hall, “The Reproducing Piano—What Can It Really Do?” The Pianola Journal: The Journal
of the Pianola Institute vol. 14 (2001): 7; see also Holliday, Reproducing Pianos, 1–3.
Werner König, “The Welte-Mignon Reproducing Piano and Its Place in the History of Music,”
The Pianola Journal: The Journal of the Pianola Institute vol. 18 (2007): 52.
28 off the record

individual pianists presented, replete with a catalogue of bizarre, inhuman effects

including jerky phrasing and inappropriate pedaling.”122 Note here that it is the
playback that Benko found unsuccessful, not the rolls themselves.
Others are concerned about the fact that the rolls have been edited. In the
main reproducing piano systems, a fair amount of expression was edited in by
technicians postperformance, rather than being captured accurately at the time.123
For this reason some people are suspicious of the results. For example, Schonberg
singles out Ampico as having inserted expression marks, evened out scales and
altered dynamics on their rolls. Ampico, more than other companies “insisted on
technical perfection,” which entailed a significant amount of editing.124
But is this quite as abhorrent as is it seems? For starters, in all companies the
dynamic data were taken down by people who were both musically and technically
highly skilled. And only a select few were employed to edit the recordings of
revered pianists throughout the life span of the reproducing piano.125 At least for
the Duo-Art and Ampico systems, the recording pianists themselves had quite a
hand in the postrecording production process. There was ample opportunity to
make comments and requests, which the skilled editors would accommodate as
much as possible. Benno Moiseiwitsch (1890–1963) remembered attending up to
three postrecording sessions for each of his Ampico rolls. This could take several
weeks, and it was only after he had given his approval that the master roll was
ready for mass production and issue. Moiseiwitsch could request all sorts of
changes: “I want more crescendo or there’s one note sticking out which shouldn’t.”126
Paderewski is reputed to have written onto a Duo-Art master roll “I do not play
this evenly. Can you make it even for me?”127
What a relief it must have been to be able occasionally to make such requests.
Would the resulting changes have significantly affected the overall conception or
the individual style and performing practices of the recording pianist? I do not
believe so. Striving for more successful musical and technical results does not
automatically negate or invalidate everything that is preserved on a recording.
Those of us who make recordings now are accustomed to the high level of editing
that inevitably takes place. That we strive to give our best in the initial recording
is a given. But even after a good day, it is standard for the producer to tidy things
up. Often this involves editing out wrong notes and technical blemishes and per-
haps editing in alternative versions that are deemed more successful from an

Gregor Benko, “More on Piano Rolls,” International Classical Record Collector vol. 2, no. 7
(1996): 81.
Philip, Performing Music in the Age of Recording, 31.
Schonberg, “The Ampico-Argo Piano Rolls,” 308.
Hall, “The Reproducing Piano,” 13.
Moiseiwitsch: Opinions on Player Piano Rolls, interviewed by Denys Guerolt (Oct. 26, 1962), BBC
Archives LP 28654 (12” 33 1/3 rpm) mono disc.
Schonberg, “The Ampico-Argo Piano Rolls,” 308.
Early R ecording s 29

interpretative point of view. On the other hand, there are days when the producer
might literally have to sew some parts of the performance together, sometimes
bar by bar or, in extreme cases, note by note. In contrast, a reproducing roll is one
complete performance, but with some editing—not several “takes” joined up.
Nowadays just as then, editing is seen as a necessary evil. I can say with confidence,
however, that my recordings made on a good day are representative of what I can
achieve in a successful live performance. At the very least, my performance style
and idiosyncrasies are well preserved.
By its nature, piano roll production was fairly complex because the process was
not fully automatic. To evaluate the evidence preserved in rolls, it is important to
understand which features of the initial performance were recorded accurately
and how much and what type of editing took place afterward.

Roll Editing
For each system, it was quite possible to change the position of perforations and
even to add extra ones. So the initial information could be modified in all sorts of
ways. Hamilton asserts that rolls “were often edited to produce results just as
deceptively faultless” as more recent recordings that are an amalgam of multiple
takes.128 Although there is certainly some truth in this, the reality is that each
company had differing views about the type and amount of roll editing that should
be undertaken.129

Note Positions
Correcting mistakes on rolls was not difficult. The method was to cover over the
hole that reproduced the incorrect note and punch a new hole for the correct
one.130 It is easy to assume that editing in this way was something undertaken by
all companies as a matter of course throughout their life spans. But this is not the
case. Welte apparently did very little editing in its early years (before World War I),
even though all that was needed to modify the pitch or rhythm was “a sharp
blade and some sticky tape.” According to Lawson, proof of this lies in the fact
that the company did not always edit out wrong notes: corrections were not
made to any great extent.131 Hall shares Lawson’s opinion, explaining that fea-
tures such as rhythms and the breaking or dislocation of hands could not easily
(if at all) be changed. He believes that even in later years Welte made no changes

Kenneth Hamilton, After the Golden Age: Romantic Pianism and Modern Performance (Oxford:
Oxford University Press, 2008), 143.
Lawson, pers. comm., e-mail, Feb 5, 2008.
Rosen, Piano Notes, 147.
Lawson, pers. comm., e-mail, Feb. 5, 2008.
30 off the record

to note positions. There were slight changes to the dynamics, but nothing that
altered the performance to any significant degree.132 From the start, Welte appears
to have had a no-editing policy. After the recording there was no further consulta-
tion with the pianist and “no editing of the basic recorded data was allowed even
by the Welte-Mignon engineers.”133 In this respect, the contract between English
pianist Fanny Davies and Welte is telling. She was invited to Freiburg to listen to
and approve the Welte-Mignon recordings she made in London. If she did not like
any of them, she could rerecord them. Notably, there is no mention of “correcting”
or “editing” the rolls.134
Other companies adopted rather different approaches. In both the Ampico
and Duo-Art systems, all sorts of editing were undertaken. Wrong notes were an
inevitable factor in most recordings. Reynolds—Duo-Art’s English producer and
technician during the 1920s—noted that one of the finest pianists “recorded
no less than three-hundred and sixty false notes in a single composition” in one
session. In fact, most of these were probably caused by the pianists’ normal brush-
ing of adjacent notes, which was registered by the sensitive recording apparatus.
But these would not ordinarily have been heard in live performance. Fortunately,
the rolls could be edited under the supervision of the pianist, removing every
blemish and adding missing notes, as well as improving touch and rhythm in
whichever way was desired. According to Reynolds, the result was “a most finished
interpretation.” It was such sophistication that led Grainger to enthuse that his
Duo-Art rolls represented not simply how he did play, but how he “would like to
play.” For Paderewski, Duo-Art rolls gave “the same feeling in his heart” as when
he played the music himself.135
Many pianists preferred roll recording to disc recording. The latter caused anx-
iety because of the pressure of having to get things right. Moiseiwitsch explained
that it was very seldom that he could record a disc and think of it as perfect. But
the experience of making an Ampico roll was so much more pleasant for him.
At the editing stage he could “sit back in a chair and smoke and have a drink
and say—no, no I want more crescendo there, or this is too much accelerando.”
And the recording engineer punched the holes according to what he was told.136
Apparently, extreme measures were sometimes necessary at the roll editing
stage in drastic situations. Hambourg retells a tale that sounds like one of those
good after-dinner stories. The famous pianist engaged to record doubted that he
had “a sufficiently reliable finger technique” to perform successfully the passages
of thirds in Chopin’s Etude Op. 25 No. 6. Under these circumstances, he was

Hall, pers. comm., letter, Feb. 5, 2008.
Roy Howat, “Review,” The Pianola Journal: The Journal of the Pianola Institute vol. 13 (2000): 39.
Hall, pers. comm., letter, Nov 7, 2008.
Reynolds, “On Playing the ‘Pianola,’” 152–3.
Moiseiwitsch: Opinions on Player Piano Rolls.
Early R ecording s 31

persuaded to play only the upper notes; the lower ones could be cut into the roll
afterward. In this way, “all the laboriously difficult progressions of double notes
sounded as smooth and easy as one could wish.”137 Extreme cases aside, however,
the purpose of editing was not to alter drastically the overall concept of the initial
performance. It was used as a means of tidying up what was already there and
to make the result as musically effective as possible—to emulate the particular
pianist on his best day.

All companies boasted that their instruments could play with a great variety of
dynamic levels. Special perforations in the roll provided the necessary data to
create expressive effects like those heard in actual piano playing. These include
instantaneous changes from one level of dynamic to another, crescendo and
diminuendo effects, accentuation of individual notes as well as groups of notes,
and differentiation between melody line and accompaniment.138 Yet how authen-
tically or successfully dynamics have been registered on the rolls is certainly
a matter for consideration.
The process for recording and editing dynamics was quite different for the
three main companies. From the start, Ampico and Duo-Art systems used meth-
ods to record dynamics that were not automatic. For Ampico rolls produced before
1926, an editor—a trained technician who was also a fully professional pianist—
added dynamics postrecording. This editor either guessed the dynamics or created
an aide-memoire—a score annotated during the recording itself—with the record-
ing pianist’s dynamic shapes and shadings as well as accented notes and counter-
melodies.139 Although generally satisfactory, the results could certainly be at
variance with the intentions of the pianist. Frequent consultations between pia-
nist and editor (at least three in the case of Moiseiwitsch) were necessary in order
to get the dynamics right. From about 1926 onward, however, Ampico developed
a machine that could help confirm the dynamic shading in the initial recording.
This used the principle of a spark chronograph, in which “the velocity of the
piano hammers during their last half inch of travel toward the string” could be
measured extremely accurately.140
In the Duo-Art system, the approach was rather different. A recording pro-
ducer or studio master manipulated two control knobs on a special recording
console. The overall dynamic level, together with levels for accented notes and

Hambourg, From Piano to Forte, 298.
Holliday, Reproducing Pianos, 5–6.
Hall, “The Reproducing Piano,” 8.
Larry Givens, Re-Enacting the Artist . . . a Story of the Ampico Reproducing Piano (New York: Vestal
Press, 1970), 28.
32 off the record

melody lines, was registered by “dialing in the appropriate power levels.” In this
way, a “rough dynamic coding” was registered as the pianist played. This appeared
as extra perforations next to the actual note perforations.141 Afterward, the
editing of the roll took place, taking special account of these dynamic codes.
It has been suggested that a gramophone recording was made simultaneously
with the cutting of the roll, which could be used later as an aide-memoire to check
dynamics and accentuation.142 This seems unlikely however. First, the four-and-a-
half-minute time constraint would have limited which works could be recorded.
And second, the production of a 78 rpm record was a lengthy process.143
The Welte-Mignon system used a completely different method to record
dynamics. The system’s patent, dating from 1904, describes in great detail “an
arrangement in mechanical piano playing apparatus, by which the striking of
the keys can be graduated as desired, in the most perfect manner.” This is achieved
“by the different arrangement of bellows, which are operated dependently on
each other by relays.” The various relays actuate and disengage the forte-piano,
crescendo, and mezzoforte—action to produce various dynamic levels.144 Welte
claimed, and current opinion affirms, that dynamics were recorded automatically
in real time. This may well have been very simple at the outset.145 In 1904, Welte’s
resources were apparently rather limited with regard to electrical technology;
therefore the process of recording dynamics was probably fairly “primitive at
best.”146 The process itself was a closely guarded secret, so no matter what theories
have been expounded, the method by which data for the dynamics was obtained
remains unknown. It is highly likely, however, that the company was indeed able
to collect much important data during the actual recording. This is supported by
Welte’s claim that no further input from the artist at later stages of production
was necessary.147 The extent to which the Welte-Mignon system was able properly
to register key speeds may never be clear. But has Hall concludes, it must have
been able “to differentiate between loud and soft playing, and between accented
notes and melodies, and accompaniments.”148 It is also prudent to acknowledge
that all the systems were considerably more versatile than a description from a
patent implies. This is most certainly the case with the Welte-Mignon. As Hall
explains, “The more experience I have with my own Welte reproducing piano,

Hall, “The Reproducing Piano,” 8.
Holliday, Reproducing Pianos, 58.
Hall, pers. comm., letter, Oct. 7, 2010.
Welte-Mignon, British patent no. 10219 (1905): 1–4.
Lawson, pers. comm., e-mail, Feb. 5, 2008.
Reinhardt, “The Welte-Mignon Reproducing Piano,” 19.
Hall, pers. comm., letter, Feb 5, 2008.
Hall, “The Reproducing Piano,” 7–8.
Early R ecording s 33

the more amazed I am at just how good even the earliest 1905/6 rolls are
But no matter how well the Welte-Mignon system worked, other complexities
surrounded the recording of dynamics. König explains:

Its one big defect lay in the fact that dynamics as recorded could not be
reproduced directly (as could the notes, note-lengths, etc.) but had to be
adapted to the capabilities of the pneumatic reproducing-mechanism.
For example, a trill came out as a trill. To reproduce it [with] exactly the
same shading as the player had given it was a sheer impossibility. Further
limitations arose out of the fact that the pneumatic mechanism was
divided into bass and treble departments. . . . If both hands were playing
in the same department, the possibility of distinguishing between them
dynamically was very much reduced. A further distortion was due to the
constant use of the left pedal [soft pedal] in piano and pianissimo pas-
sages, not by the pianist himself but so to speak by the recording firm;
where the performing pianist actually worked the left pedal is impossible
to tell from these rolls.150

König concludes that for Welte-Mignon rolls all information pertaining to dynam-
ics “must be regarded as an approximation only.” On this matter, Hall differs with
König. But even if the rolls do not convey a true picture of the dynamics, they
nevertheless provide important information about all the other facets that go
into the interpretation of a musical work.151 There is general consensus about this.
Dynamics are the least important factor in recognizing a performance. Remove
the dynamics and a lot can still be told from note placement alone. The same
would not be true in reverse.152 For me, note placements on piano rolls are often
staggeringly clear, sometimes bluntly so. As Hamilton astutely observes, it is the
roll’s “robotic failure to reproduce tone colorings and dynamics adequately” that
makes important features such as playing one hand after the other and chordal
arpeggiation “much easier to hear on rolls than in early recordings.”153
Although the Welte-Mignon, Duo-Art, and Ampico systems gathered infor-
mation about dynamics by different means, the next stage of production was
similar. Editors analyzed and interpreted the information and then created extra

Hall, pers. comm., letter, Nov 7, 2008.
König, “The Welte-Mignon Reproducing Piano,” 53; Hall and Lawson disagree with König about
the use of the soft pedal in German Welte-Mignon production. They believe that the soft pedal is as
the pianist used it. This is certainly different for Ampico and Duo-Art rolls, in which the use of the soft
pedal (half blow) was an additional means of dynamic expression in the editing process.
König, “The Welte-Mignon Reproducing Piano,” 53.
Lawson, pers. comm., e-mail, Feb. 5, 2008.
Hamilton, After the Golden Age, 143.
34 off the record

perforations to effect dynamic contrasts in the playback. There must have been a
degree of subjectivity here: the personality of the editor would inevitably have
intermingled with that of the pianist.154 In terms of Welte-Mignon rolls, it is dif-
ficult to assess the authenticity of the dynamic shading. As Reinhardt concludes,
it will never be known “how much of the Welte-Mignon magic was in the record-
ing process and how much was in the skill of its editors.”155 For other systems,
such as Ampico and Duo-Art, on the other hand, the editing process certainly
allowed greater latitude in accepting, rejecting, or molding particular expressive
effects like dynamics in consultation with the pianist. Moiseiwitsch confirmed
that in his Ampico rolls the dynamic shading and the expression of each individual
note was edited in—“according to my instructions.” For him, it was possible, with
patience and care, to reproduce faithfully the pianist’s intentions in the Ampico
system. Sometimes, he was really astonished by the results.156 Indeed, Bauer went
so far as to describe the Duo-Art as “the instrument of a new musical art.” He was
completely enthralled by the possibility through editing of being the sculptor:
“I can change what I wish. I can remodel and refine.” He considered the result to
be a “carefully considered artistic conception of the music.”157

For all systems, the use of the soft and “normal” sustaining pedal was registered
in the same way as the notes by creating extra perforations in the rolls. The infor-
mation for this was gathered automatically during the initial performance.158
However, there appears to be reasonable doubt about the faithfulness of pedaling
For early Welte-Mignon rolls played on a well-regulated reproducing piano,
there is little reason to doubt that the “normal” sustaining pedal was registered,
by and large, just as the artist had played. Nevertheless, as Hall points out, some
sustaining-pedal effects on published rolls sound somewhat eccentric, if not dis-
tinctly suspect. This was almost certainly due to the setting of the sustaining
pedal contact very early in the travel of the pedal. He believes that if the pianist
did not allow the pedal to be fully released, the canceling of the pedal may have
been missed.159 In all systems, subtle half-pedal effects were difficult or impossible
to record and reproduce. Ampico employed a process that produced something

Hall, “The Reproducing Piano,” 13.
Reinhardt, “The Welte-Mignon Reproducing Piano,” 29.
Moiseiwitsch: Opinions on Player Piano Rolls.
Anon., “The Medium of a New Musical Art: An Interview with Harold Bauer and the Duo-Art
Pianola,” [Advertisement], The New York Sun, March 26, 1916.
Hall, “The Reproducing Piano,” 7.
Hall, pers. comm., letter, Oct. 7, 2010.
Early R ecording s 35

approximating to a half-pedal effect, but neither Welte nor Duo-Art took such
Regardless of whether sustaining-pedal effects were registered correctly, it
appears their exact reproduction was problematic and less than efficient in all
systems. Reproducing pianos use a pneumatic motor to operate the sustaining
pedal, which, if adjusted correctly, will function efficiently. But to duplicate accu-
rately the operating speed of the sustaining pedal from one instrument to another
“is well-nigh impossible.”161
The use of soft pedal is almost certainly as the pianist played in the original
Welte-Mignon rolls, as well as those Duo-Art rolls produced in London.162 But the
soft pedal on Ampico and Duo-Art pianos is always a half-blow, even in grand
pianos, and therefore its use cannot always be taken as when the pianist used it.
With some possible exceptions, in playback the soft pedal is quite often employed
“more frequently than in the original performance,”163 as an aid to reproducing
the pianist’s dynamics.

Other Considerations
There remain a few other matters for consideration. All reproducing piano actions
are divided into treble and bass sections. This allows the treble to play at one
dynamic level while the bass plays at another.164 But when both hands play in the
same section, the ability to distinguish between the dynamic level of the melody
and that of the accompaniment is extremely reduced.165 This difficulty was over-
come in the editing stage by a faking process, in which the multiple levels were
separated by a time interval short enough to deceive the ear. In effect, a melody
note very slightly precedes its accompaniment, or is delayed. This deception
apparently takes place very frequently in all reproducing piano roll systems.166
And when handled properly, it works amazingly well.
The inability to play at more than one level at a time also means that reproduc-
ing piano actions are unable to re-create properly all the subtle variations of
touch present in the original performance.167 In the end, there is enough informa-
tion to give a very good impression of the pianist’s playing, but obviously this

Holliday, Reproducing Pianos, 108.
Hall, “The Reproducing Piano,” 11–12.
Hall, pers. comm. letter, Dec 11, 2010.
Hall, “The Reproducing Piano,” 16.
Hall, “The Reproducing Piano,” 14.
König, “The Welte-Mignon Reproducing Piano,” 53; see also Holliday, Reproducing Pianos,
Hall, “The Reproducing Piano,” 14.
Hall, “The Reproducing Piano,” 14.
36 off the record

is not exactly as the pianist played. The result may be considered a portrait
rather than a photograph. And a portrait can sometimes be more telling than a

The playback of piano rolls was always best suited to the instrument on which it
was recorded, but this was not always possible. We know that many pianists
approved and signed the master roll after having heard the recorded playback. But
the playback was, as Sitsky points out, “on the same piano and in the same acous-
tic ambience” as the initial performance—in other words, the ideal conditions.168
The burning question is this: what happens when these conditions for playback
are modified? What if the room ambience is significantly different? And as Philip
asks, what if the piano used in playback has “different acoustical properties”
and/or “hammers in a different condition” to the original? In order to make an
effective transfer of the information on rolls, delicate adjustments to the playback
mechanism need to be made. This will certainly have an impact on the result.
For this reason, Philip questions “how close the reproduction is to the original
performance on the original instrument.”169
Bauer acknowledged this very problem in 1948. He claimed to be “always some-
what discouraged” by the final result of his Duo-Art rolls in spite of the great
efforts that had been made. The dynamics on the editing piano were “set to
produce certain effects,” but these varied when the roll was played with a different
piano. This Bauer attributed “to minute differences in quality of tone and in
resistance within the action.” Unfortunately, there was no known method of
overcoming the difficulty.170 Nevertheless, Bauer made many fine Duo-Art rolls.
This issue is potentially even more serious now. Fashions in sound, as well as
playing, have certainly changed. The pianos in use during approximately the first
thirty years of the twentieth century had “lighter and softer hammers” than
pianos built or refurbished more recently.171 Also significant is the fact that roll
recordings were generally made using pianos that were smaller than concert size.
According to Lawson, the toning on these pianos was much softer and warmer
than the current fashion. Therefore, transfers to the Disklavier,172 for example,

Sitsky, “Introduction,” xl.
Philip, Performing Music in the Age of Recording, 32.
Bauer, Harold Bauer, 175–76.
Hall, “A Window in Time,” 10.
“By the 1990s the Yamaha Corporation, a Japanese piano manufacturer, had introduced the
‘Disklavier,’ an acoustic player piano equipped with a computer that, by reading data on a floppy disc or
compact disc, could re-create on the piano virtually every nuance of a performance—the tone, touch,
timing, and dynamic range of a real performance”; in Encyclopedia Britannica, available: http://www.
Early R ecording s 37

are very distorted with regard to the piano tone. And recordings made by push-
ups (Vorsetzers) operating concert grand pianos are bound to be a compromise.
Lawson acknowledges, however, that the latter can work in concert, when other
emotional and dramatic factors come into play.173
Playback on instruments that are not functioning well or that are very differ-
ent from the original do not do justice to the recorded pianist, or to the original
recording. Even in its heyday, the reproducing piano needed continual mainte-
nance. For this reason, companies went to the trouble of supplying test rolls.
These were used to try to ensure that the equipment was working properly.174 But
they could not verify every aspect of the mechanism.
Once their popularity waned, however, reproducing pianos gradually fell into
disuse, and became defunct. Many instruments were dismantled or their intricate
parts were left to deteriorate. Perhaps worst of all, they were allowed “to play
badly.”175 To make matters worse, much vital information and expertise about
maintenance was diluted or, as in the case of the Welte Company, lost during
World War II. When dilapidated reproducing pianos were later used to replay
piano rolls, the understandably strange and bizarre effects were regarded with
disdain. As Sitsky explains, in spite of the recent resurgence in reproducing
pianos, “many musicians regard them with scorn.” They have yet “to be convinced
of their value.”176
In recent times, much criticism has been directed at roll recordings that prom-
ise great results but deliver far less. In these recordings, the necessary steps to
render the reproducing piano in “tip-top” condition have not been undertaken.
According to Benko, those presenting the rolls have done a poor job: it is impos-
sible to make a proper judgment about the performances they purport to pre-
serve. He acknowledges, however, that sometime down the track, when more
careful presentations take place, the merits of the process may be determined.177
It stands to reason that under the right circumstances the rolls will yield much
important information. As Sitsky suggests, “With infinite care, with the resources
of first class grand pianos and with unlimited time to work on the machines and
perfect the replay, startling results may be obtained.”178 There are certainly some
fine transfers commercially available now. Listen to, for example, transfers of
various Welte rolls on the TACET label.

Lawson, pers. comm., e-mail, Feb. 5, 2008.
Sitsky, “Introduction,” xiv–v.
Sitsky, “Introduction,” xiii–iv.
Sitsky, “Introduction,” xiii–iv.
Benko, “More on Piano Rolls,” 81.
Sitsky, “Introduction,” xl; Sitsky particularly commends the BBC’s Ampico rolls, the Klavier
label records, and the Welte Treasury records.
38 off the record

A Matter of Perception
When it comes to the playback of piano rolls, the difference in sound between one
piano and another may dramatically affect the perception of the pianist’s rubato
and tempo and therefore its success. The English pianist and composer David
Wilde (b. 1935) came to this conclusion based on his own experiments playing the
same repertoire on different pianos. He suggests that “some rubati which sound
grotesque on piano rolls would sound perfectly acceptable if the recording per-
mitted us to hear the tonal subtleties that gave rise to them.”179 Unless piano rolls
are replayed under strictly controlled conditions, certain otherwise sophisticated
features may come across as unsophisticated or distorted.
This perceived distortion has often led to criticism of the rolls themselves. But
comparison between acoustic recordings and piano roll transfers of the same work
by the same pianist reveal a close relationship in performance features such as
rubato and tempo. The tempo variations in Saint-Saëns’s performance of his Valse
mignonne Op. 104 are recognizably similar in both his 1904 G & T disc recording
(2008 Marston transfer) and his 1905 Welte-Mignon roll (2008 TACET transfer).
For example, the tempo remains steady from bars 1 to 48, after which there
is a recognizably similar accelerando. And there are other tempo changes that
sound similar in both performances. Grieg’s two 1906 Welte-Mignon rolls of his
Bridal Procession Op. 19 No. 2 (1992 SIMAX transfer) both preserve very similar
traits to his 1903 G & T disc recording of it (2008 Marston transfer). For example,
the characteristic rhythm—quaver-crotchet—that appears as a phrase ending
throughout the work is often subverted in both types of recording, so that the
crotchet is noticeably delayed. Many tempo modifications are mirrored, such as
the sudden hastening at bar 25 and the broadening in bars 31 and 32. Hall has
presented other useful discussion on this subject. Comparing Busoni’s 1922 disc
recording for Columbia and his 1923 Duo-Art piano roll of Chopin’s Prelude
Op. 28 No. 7, he states the following:

The one Prelude which appears on disc as well as roll (no. 7 in A) is

remarkably similar in both versions. Busoni plays it twice, bringing out
different features in the music on each occasion. The emphasis of the
melody at the beginning of the repeat is clearly there, as is the accenting
of the first chord in bar 12. The treatment of the alto line from bars 4 to
10 is not so obvious on the roll although the dynamic coding shows
that Reynolds [the editor] was aware of what Busoni had played even
if he did not translate the effect successfully to the roll. Nonetheless, the
similarities are very marked; the two performances are quite clearly by
the same pianist.180

Philip, Performing Music in the Age of Recording, 32.
Hall, “Duo-Art Rolls,” 34.
Early R ecording s 39

Hall is also convinced that when piano rolls are played under the right condi-
tions, many aspects of the original performance are indeed reproduced. This is
supported by comparison between the Duo-Art rolls of certain pianists and their
disc recordings. As examples, he singles out Paderewski’s rolls of The Maiden’s
Wish (Chopin/Liszt), the Nocturne Ragusa (Schelling), and Reflets dans l’eau
(Debussy) “as instantly being recognizably” the Paderewski on disc recordings. In
these, the phrasing, dynamic effects, and pedaling are identical. Hall also explains
that Bauer’s general style—the style that comes through in his disc recordings
distinctly—is unmistakable on his Duo-Art rolls. This includes his characteristic
“singing treatment of melody-lines, with the left hand not quite together with the
right” as well as “his forthright playing of rhythmic passages.”181 Schonberg high-
lights other examples by comparing the 1966 Argo transfers of various pianists’
Ampico rolls with their disc recordings. For him, Rachmaninoff ’s performances
are recognizably the Rachmaninoff of his Victor and HMV records—“the same
control, charm and aristocracy” are clearly evident. And Rosenthal’s style, too,
“comes faithfully through” on his Ampico rolls.182
There will inevitably be more debate about the value of reproducing piano
rolls and the success of their playback. To take into account their shortcomings is
judicious, but to be overly dismissive is unnecessary and misguided.

Reproducing Piano Rolls: Summary

The evidence preserved in reproducing piano rolls is of undeniable importance. It
can tell us much about the playing style of great pianists born as far back as the
middle of the nineteenth century. These rolls preserve many important features
of the original performance. They are a snapshot of the note pitches, durations,
positions, and the tempo, which were registered automatically. From these fea-
tures alone, it is often possible to recognize the playing of particular pianists.183
When editing did not take place—in earlier Welte-Mignon roll production, for
example—rolls provide very precise information about such features. Where edit-
ing did take place, with or without the supervision of the artist, we are hearing an
impression—a version that may be somewhat modified or tidied up—but one
that still retains many important features of the initial performance. It was
certainly possible to modify any note pitch, duration, or position. Yet, often
published rolls preserve features that sound untidy, erratic, or downright wrong
to modern ears. Although these are not in accordance with modern taste, it is
clear that they were once considered the essence of high art in piano playing.

Hall, “Duo-Art Rolls,” 49–50.
Schonberg, “The Ampico-Argo Piano Rolls,” 308.
Hall, “The Reproducing Piano,” 15.
40 off the record

Dynamics and pedaling are the less reliable aspects of piano roll production
and playback. Think of dynamics as “the icing on the cake.”184 These were ulti-
mately the most difficult feature to encode and were either rather bland—as in
the early Welte-Mignon rolls—or created and/or enhanced by a roll editor. And
although both sustaining and soft pedal data were recorded automatically as the
pianist played them, such features could be modified in the production stage for
a variety of technical reasons and to create successful effects. It is not always
possible, therefore, to be absolutely certain what the pianist actually did.
Reproducing piano rolls are an important aid to the study of historical per-
forming practices being “a repository of a valuable but lost performing tradition.”185
Clearly, they cannot tell us exactly how particular pianists sounded, but they can
augment our understanding of pianists who made only preelectrical recordings.
And, importantly, they give us some idea of the playing style of several important
pianists who did not record at all. We can, for example, gain a useful impression of
Leschetizky’s own playing style from his Welte-Mignon rolls. It may or may not
come as a surprise to hear him playing one hand after the other and arpeggiating
chords in Chopin’s Nocturne Op. 27 No. 2, but what I find utterly striking is the
manner in which he applies these devices, their audible effect, and the frequency
of their occurrence.
Acoustic piano recordings and reproducing piano rolls provide an all-important
window into the last hundred years or so. But the information that I feel can
safely be extrapolated from them at this stage concerns practices that are not
directly influenced by dynamics, tone, touch, and pedaling. These practices are
investigated in the chapters that follow.

Hall, “The Reproducing Piano,” 15.
Sitsky, “Introduction,” xiv.

Playing One Hand after the Other

When I listened to the first record of my own playing, I heard
things which seem unbelievable to me. Was I, after years of public
playing, actually making mistakes that I would be the first to con-
demn in any one of my pupils? I could hardly believe my ears, and
yet the unrelenting machine showed that in some places I had
failed to play both hands exactly together and had been guilty of
other errors no less heinous, because they were trifling. I also
learned in listening to my own playing, as reproduced, that I had
unconsciously brought out certain nuances, emphasized different
voices and employed special accents without the consciousness of
having done so. Altogether it made a most interesting study for
me, and it became very clear that the personality of the artist
must permeate everything that he does.
—Max Pauer (1866–1945)1

Preamble: The Tyranny of Synchronous Playing

Acoustic piano recordings and piano rolls made around the turn of the twentieth
century preserve a style of playing that is often vastly different from the charac-
teristic style of piano playing today. Some exhibit frequent use of such expres-
sive devices as dislocation of melody from accompaniment2 and unnotated

Max Pauer, “Modern Pianistic Problems,” in James F. Cooke, Great Pianists on Piano Playing:
Study Talks with Foremost Virtuosos (1913, reprint, Philadelphia: Theodor Presser, 1917), 201–2. Like
his father Ernst Pauer (1826–1905), Max Pauer was a celebrated concert pianist, teacher, composer,
and music editor who held prestigious posts at the Cologne and Stuttgart Conservatories and in
Referred to as “rhythmic dislocation of melody from accompaniment” in Philip, Early Recordings,
47; “breaking of hands” in Richard Hudson, Stolen Time: The History of Tempo Rubato (Oxford: Oxford
University Press, 1994), 334; and “the splitting of hands” in Sandra P. Rosenblum, “The Uses of Rubato
in Music, Eighteenth to Twentieth Centuries,” Performance Practice Review vol. 7, no. 1 (1994); and
variously as “dislocation of the hands” and “asynchronization” in Hamilton, After the Golden Age.


Peres da Costa-Ch02.indd 41 3/14/2012 6:01:39 PM

42 off the record

chordal arpeggiation,3 enhancing the expressive poignancy and/or the rhythm

and texture of the music; whereas the trend during the past sixty or seventy years
has been toward little or no use of these whatsoever. The twentieth century saw a
growing obsession with absolute faithfulness to the musical notation, which has
undoubtedly made its mark. Piano playing has become increasingly neat and tidy
to the extent that absolute synchrony is now one of the benchmarks of virtuosity
in piano playing. “Most pianists today,” explains Crutchfield, “are flabbergasted at
the suggestion” that these previously popular techniques were “anything but a
meretricious sugar coating.”4 This dramatic change in attitude and practice is one
of many reasons why early piano recordings often come across as curiously dis-
jointed by present standards.5
The synonymy between synchronous playing and stylistic or good taste is
clearly exemplified in the horrified, if somewhat amusing, reaction of a trained
musician to pianist Melvyn Tan’s introduction of an unnotated arpeggio to a
hallowed early-nineteenth-century work. In a review published in the Sydney
Morning Herald (November 23, 1998), the critic takes obvious exception to Tan’s
interpretation of the opening chord of Beethoven’s Piano Concerto No. 4 Op. 58
(first published in Vienna, 1808) in a performance with the Australian Chamber

Then there was the vexatious question of the first chord. For those unac-
quainted with the frailties of modern pianism, the first chord of
Beethoven’s Fourth Piano Concerto is generally held to be the most dif-
ficult chord in the repertoire. Playing it is like performing brain surgery
on Stephen Hawkings [sic]: if you don’t get the exact balance, the exact
pressure, the exact weight, you risk killing one of the most sublime cre-
ations of the human spirit. Tan rolled it! He arpeggiated it (i.e., he played
the notes one after the other, rather than together). Many would regard
this as the greatest dereliction of civic duty since Pontius Pilate. For
Tan and Tognetti, however, it seemed to be an attempt to introduce
some of the freedoms of 18th-century performance practice to this early
19th-century work.6

When I first read this review, I was astonished by the severity of the objection
and its attendant analogies. But I have since discovered that others share

Here, chords are composed of two or more notes aligned vertically in the notation.
Crutchfield, “Brahms, by Those Who Knew Him,” 18.
Henceforth, the term early piano recordings is used to describe wax cylinder, wax disc, and
reproducing piano roll recordings and transfers.
Peter McCallum, “When Seeking a Good Tan, Turn before You Burn,” Sydney Morning Herald,
Nov. 23, 1998, 15.
Play ing One Hand after the Other 43

such opinions. In 2001, for example, the music critic of the Frankfurter Allgemeinen
Zeitung censured the pianist Lars Vogt for playing the opening chord of Beethoven’s
Fourth Piano Concerto with an “audacious arpeggio not intended by the com-
poser.” In response to this, Anselm Gerhard has quite rightly queried how the
critic knows what Beethoven really intended.7 In any case, I fail to see why some-
thing as simple and as potentially beautiful as an arpeggio should cause so much
Should one arpeggiate this chord, and if not, why not? Pianists of Beethoven’s
era and later considered arpeggiation and other practices indispensable for highly
expressive playing, a fact that is well documented in a plethora of historical
sources. Here, one piece of evidence stands out as worthy of consideration. When,
in 1846, Carl Czerny inserted an arpeggio sign next to the first chord of Beethoven’s
Fourth Piano Concerto, he may well have been preserving either Beethoven’s
own practice as he remembered it or one that was commonly heard during the era
(Fig. 2.1).8 Some might argue, as George Barth has in The Pianist as Orator (1992),
that Czerny’s additions and/or changes to Beethoven’s texts—particularly slurs
and articulations—had a modernizing effect, bringing them into line with
mid-nineteenth-century taste.9 Nevertheless, there is little reason to believe the
arpeggiation of a chord would have sounded modern in 1846. Indeed, the evi-
dence supports quite the opposite: it is more likely that during the mid-nineteenth
century a nonarpeggiated chord would have sounded unusual and or even

Figure 2.1 Beethoven Piano Concerto Op. 58, first movement, bars 1 to 5, annotated
by Czerny.

“Jürgen Otten, über ein Konzert in Berlin mit Lars Vogt und dem Orchester der Komischen Oper
Berlin unter Leitung von Shao-Chia Lü,” Frankfurter Allgemeinen Zeitung (March 2001), in Anselm
Gerhard, “Willkürliches Arpeggieren: Ein Selbstverständliches Ausdrucksmittel in der Klassische-
Romantischen Klaviermusik und seine Tabuisierung im 20. Jahrhundert,” Basler Jahrbuch für
Historische Musikpraxis vol. 27 (2003): 123: “Auch in den Ecksätzen ist dieser Beethoven, den der
Pianist mit einem kühnen—von Komponisten nicht intendierten—Arpeggio in der Haupttonart fast
lapidar, beiläufig eröffnet hat, zumeist von erlesener Qualität.”
Czerny, Art of Playing, 109.
See George Barth, The Pianist as Orator (Ithaca, N.Y.: Cornell University Press, 1992).
44 off the record

Almost a half-century ago, Paul Badura-Skoda cast a shadow of doubt on

Czerny’s arpeggio sign, reasoning that “it is present neither in the copy corrected
by Beethoven (Gesellschaft der Musikfreunde, Vienna) nor in the first edition.”10
Clearly, this level of text faithfulness muddies the waters. In this second decade of
the twenty-first century, there is widening acceptance that musical scores (even
those emanating from godlike figures such as Beethoven) provide only the first
steps—the notes and some musical signposts—in appreciating what took place in
performances of the past. If we entertain the idea that Czerny was simply letting
future generations know that this was one significant place (perhaps among
many) that was worthy of such treatment, his arpeggio sign is not curious at all.
In fact, it makes perfect musical sense, particularly to enhance the feeling of piano
and dolce. The gentle harp-like aural effect of such an arpeggio for the first chord
of Beethoven’s Fourth Piano Concerto is heard on Steven Lubin’s 1988 recording
with the Academy of Ancient Music (Audio Ex. 2.1 ).

The Practices of a Receding Epoch

One occasionally meets people who remember hearing pianists of an older
generation (sometimes their grandparents or the like), who regularly played one
hand before the other. Yet during the second half of the twentieth century, this
practice—which results in the separation of melody from accompaniment—was
often met with severe disapproval. Gerald Thompson (b. 1933), a semiprofes-
sional pianist, recounted his consternation at hearing (around 1975) a recording
of Schumann’s Frauenliebe und -leben Op. 42 by the singer Kathleen Ferrier and
the pianist Bruno Walter:11

I realized there was something wrong, that Bruno Walter was putting
down his left hand before his right hand, and this continued to the extent
that I don’t think that I could listen to it to the end, I felt really so
distressed, almost ill, and I haven’t replayed it for twenty-six years.12

Others have expressed similarly strong reactions to this practice. In Speaking of

Pianists (1957), Abram Chasins praises Paderewski’s “beautiful tone and poetic
feeling” in a performance of Schumann’s Warum? Op. 12 No. 2. But he bemoans
“the unhappy traits of melodramatic expressivity—the agonized cantilena,

Paul Badura-Skoda, “Kommentar,” Carl Czerny: Uber den richtigen Vortrag der sämtlichen
Beethoven’schen Klavierwerke (1963); trans. as “Commentary,” On the Proper Performance of All
Beethoven’s Works for the Piano (Vienna: Universal, 1970), 11.
Kathleen Ferrier and Bruno Walter, Schubert, Schumann and Brahms, BBC live recording from the
1949 Edinburgh Festival (Decca, Mono 6BB197-8).
Transcript of part of an interview with Gerald Thompson (September 15, 2001). Thompson con-
ducted a semiprofessional career as a piano accompanist in Wakefield, West Yorkshire.
Play ing One Hand after the Other 45

the torn-to-tatters meter, the hands played one before the other.”13 Like
Paderewski, Rosenthal received criticism for this manner of playing. In 1962,
Sackville-West observed that Rosenthal “shared with Paderewski an indifference
to synchronizing the hands” but recognized that he “probably thought it more
expressive not to do so.”14 Clearly, asynchrony of the hands was part and parcel of
Rosenthal’s pianistic style.
By the second half of the twentieth century, playing one hand after the other
(henceforth called dislocation) was firmly a thing of the past. In 1983, Moore
recounted the English music critic Neville Cardus’s (1888–1975) description of
Paderewski as “a visitant from a receding epoch.” He attributed this to Paderewski’s
practice of dislocating the hands—“a relic of bygone days.” Through the course of
the twentieth century, musical tastes had changed dramatically. Moore regarded
such a practice as amateurish, the result of not listening to oneself, and the first
fault that a great teacher tries to correct. Importantly, however, he acknowledged
that dislocation was possibly “the apogee of expressiveness” a century or more

Dislocation describes a momentary separation between the left and right hands.
This expressive technique is not exclusive to, but is particularly noticeable on,
early recordings of solo pianists.16 The usual method is to delay a melody note in
the right hand, placing it directly after the corresponding accompaniment note in
the left hand. In fewer cases, the right hand precedes the left. Dislocation resem-
bles the technique referred to in chapter 4 as metrical rubato—the rhythmic
alteration of melody notes (often in an extended way) over an accompaniment
that preserves a sense of the pulse. It also bears similarity with the practice of
unnotated arpeggiation (dealt with in chapter 3)—the separation of notes that
should, when their vertical alignment in the musical text is interpreted literally,
be synchronized. Some might argue that all of these techniques amount to one
and the same thing: a way of creating rubato or displacement of time that influ-
ences phrase-shape, texture, and dynamic. Indeed, on early piano recordings,
these are often introduced in combination resulting in highly expressive effects.
However, when I use them in my own playing, I do have to think about them
separately, to decide how and when to dislocate my hands, add arpeggios, and

Abram Chasins, Speaking of Pianists, 3rd ed. (New York: Da Capo, 1981), 90.
Sackville-West, “Rosenthal,” 214.
Moore, Furthermoore, 396–97.
There are examples in recordings of singers, string players, and chamber ensembles, cited later.
46 off the record

make metrical displacements. I feel, therefore, that such complexity necessitates

discussion of these as separate techniques.
On early piano recordings, dislocation is heard much more often in music
of a slow and expressive character than in fast music. Often in compositions of
varying characters, it is reserved for the most expressive part. Some pianists,
however, apply it universally. Typically, dislocation occurs at the beginnings of
phrases, beginnings of bars, or moments that are harmonically strong or disso-
nant. In some examples, it can be heard on practically every beat in a bar.
Dislocation is effected in all permutations of single notes and chords (either notes
struck together or arpeggiated) in both hands, the underlying criterion being the
separation of the hands. This is illustrated in Table 2.1.
The aural effect of dislocation is that sometimes the accompaniment sounds
aligned with the beat or pulse, with the melody note sounding late. At other times,
it is the melody note that sounds aligned with the pulse, and the accompaniment
sounds early.17 It is not always clear what relationship the right and left hands
have to a pulse, especially when dislocation occurs in conjunction with a modifica-
tion of tempo.
Dislocation is preserved on piano recordings from as early as 1889 and contin-
ues well into the second half of the twentieth century, though with significantly
declining incidence after the 1930s.18 One hears it only very occasionally nowa-
days in modern piano recordings and live performances. But its use has been
long-standing in the early keyboard scene—particularly by harpsichordists and
clavichordists. And it is gradually infiltrating fortepiano playing. Table 2.2 charts
some examples from early piano recordings in which dislocation is prominent

Table 2.1 Types of Dislocation Preserved in Early Recordings

Left Hand Right Hand

Single accompaniment note Single melody note

Chord (notes arpeggiated) Single melody note
Chord (notes struck together) Single melody note
Single accompaniment note Chord (notes arpeggiated)
Chord (notes arpeggiated) Chord (notes arpeggiated)
Chord (notes struck together) Chord (notes arpeggiated)
Single accompaniment note Chord (notes struck together)
Chord (notes arpeggiated) Chord (notes struck together)
Chord (notes struck together) Chord (notes struck together)

Also known as “bass-note anticipation.”
For further discussion see Philip, Early Recordings.
Play ing One Hand after the Other 47

and some in which it is almost completely absent.19 The pianists listed include
those who used dislocation in a significant number of recordings, as well as those
in whose playing it might reasonably have been expected because of their age.
Other pianists, not listed, whose recordings reveal that they used dislocation
occasionally, or infrequently, include Emil Sauer (1862–1942), Alfred Cortot
(1870–1962), Leopold Godowsky (1870–1938), Harold Bauer (1873–1951), Sergei
Rachmaninov (1873–1943), Alexander Goldenweiser (1875–1961), Maurice Ravel
(1875–1937), Ernö Dohnanyi (1877–1960), Ossip Gabrilowitsch (1878–1936),
Severin Eisenberger (1879–1945), Olga Samaroff (1880–1948), Ignaz Friedman
(1882–1948), Elly Ney (1882–1968), Wilhelm Backhaus (1884–1969), Edwin
Fischer (1886–1960), Ethel Leginska (1886–1970), Myra Hess (1890–1965),
Benno Moiseiwitsch (1890–1963), and Alexander Brailowski (1896–1976).
The recordings presented in Table 2.2 show that between the late nineteenth
century and at least the 1950s, many pianists made use of dislocation. It was
employed in slower expressive compositions of Classical and Romantic repertoire,
less so in later nineteenth century and more contemporary repertoire, or in music
that was fast or required a more incisive and sharp rhythm and attack. Notably,
several pianists seem to avoid its use altogether: their playing sounds much more
synchronized. Reasons for this are explored later.

Table 2.2 Dislocation Preserved in Early Recordings

Pianists Compositions with Frequent Compositions with Infrequent or
Dislocations No Dislocations

Carl Reinecke Schumann Warum? Op. 12 No. 3

(1824–1910) and Mozart Larghetto from Piano
Concerto K 537 arr. Reinecke, 1905,
piano roll
Theodor Chopin Nocturne Op. 27 No. 2,
Leschetizky Mozart Fantasia K 537, Leschetizky
(1830–1915) Les deux alouettes and Barcarole,
1906, piano roll
Johannes Brahms Hungarian Dance
Brahms No. 1, 1889, wax cylinder
Camille Chopin Nocturne Op. 15 No. 2 and Saint-Saëns Valse mignonne
Saint-Saëns Beethoven Sonata No. 16 Op. 31 No. 1 Op. 104, 1919; Valse mignonne
(1838–1921) (2nd movement), 1905, piano roll Op. 104, 1905, piano roll

Unless otherwise stated, recordings made before 1924 are acoustic; recordings from 1924
onward are electrical. All piano roll transfers are indicated.
Table 2.2 Cont’d
Pianists Compositions with Frequent Compositions with Infrequent or
Dislocations No Dislocations

Francis Planté Chopin Etude Op. 25 No. 1, Etude

(1839–1934) Op. 25 No. 2; Gluck arr. by Brahms
Gavotte; Schumann Romance Op. 32
No. 3, 1928
Edvard Grieg Grieg Butterfly Op. 43 No. 1,
(1843–1907) To Spring Op. 43 No. 6,
Remembrances Op. 71 No. 7, Alla
Menuetto and Finale from Piano
Sonata Op. 7, Gangar Op. 54
No. 2, Wedding Day at Troldhaugen
Op. 65 No. 6, Humoresque Op. 6
No. 2, and Bridal Procession Op. 19
No. 2, 1903
Vladimir de Chopin Nocturne Op. 9 No. 2 and Liszt Polonaise No. 2, 1915;
Pachmann Mazurka Op. 64 No. 4, 1915, and Chopin Etude Op. 25 No. 3,
(1848–1933) Nocturne Op. 27 No. 2, 1916 1915; Schumann Grillen
and 1925; Liszt Liebestraum No. 3 Op. 12 No. 4, 1915
S. 541, 1916
Alfred Chopin Nocturne Op. 32 No. 2, J.S. Bach Gavotte from English
Grünfeld 1911; Schumann Träumerei Op. 15 Suite No. 6, 1908; Grieg
(1852–1924) No. 7, 1913; Wagner-Liszt Isolde’s Schmetterling Op. 43, No. 1,
Liebestod, 1909 1899, Vöglein Op. 43 No. 4,
1907, Sie Tanzt Op. 57 No. 5,
1907; Debussy Golliwog’s
Cakewalk, 1914
Raoul Pugno Chopin Nocturne Op. 15 No. 2, Scarlatti Sonata in A Major;
(1852–1914) Impromptu Op. 29, Handel Gavotte and Variations
Marche funèbre from Sonata Op. 35, from Suite No. 14;
Valse in A flat Op. 34 No. 1, and Mendelssohn Scherzo Op. 16
Berceuse Op. 57, 1903 No. 2, Spinning Song Op. 67
No. 4 and Hunting Song Op. 19
No. 3; Weber Rondo brilliante in
E flat Op. 62; Massenet Valse
folle; Chabrier Pièces pittoresques;
Pugno Valse lente, Sérénade à la
lune, Impromptu, 1903

Table 2.2 Cont’d
Pianists Compositions with Frequent Compositions with Infrequent or
Dislocations No Dislocations

Ignacy Jan Haydn Andante & Variations in F, All faster works of Chopin such
Paderewski 1937; Mozart Rondo K 511, 1937; as Etude Op. 10 No. 12, 1928,
(1860–1941) Beethoven Sonata Op. 27 No. 2, Etude Op. 10 No. 5, 1928,
1937; Schubert Impromptu D. 935 Polonaise Op. 26 No. 2, 1930,
No. 2, 1926 and Impromptu D. 935 Waltz Op. 34 No. 1, 1912; and
No. 3, 1924; Chopin Nocturne faster works of Liszt such as La
Op. 15 No. 2, 1917, acoustic, leggierezza from 3 Etudes de
Nocturne Op. 9 No. 2, 1930, Etude Concert, and La Campanella from
Op. 10 No. 3, 1928, Waltz Op. 18, 6 Etudes d’exécution transcendent
1928, Mazurka Op. 17 No. 4, d’après Paganini, recording dates
Prelude Op. 28 No. 15, 1928, unknown
electrical; Paderewski Menuet célèbre
Op. 14 No. 1, 1937
Fanny Davies Schumann Kinderszenen
(1861–1934) Op. 15 No. 1, 1929, Concerto Op. 54,
1928, acoustic
Moriz Chopin Nocturne Op. 9 No. 2, 1935
Rosenthal and 1936, Nocturne
(1862–1946) Op. 27 No. 2, Valse Op. 64 No. 2,
Mazurka Op. 50 No. 2, 1936;
Schubert Moments musicals No. 3 D.
780, 1937; generally in the slower
sections of works in fast tempo
Feruccio Chopin Prelude Op. 28 No. 7, 1922 Not in works of Chopin that are
Busoni fast or in works of Bach
Carl Friedberg Chopin Nocturne Op. 15 No. 2, Faster music of Beethoven,
(1872–1955) 1949; Schumann Etudes Brahms, Chopin, and
Symphoniques Op. 13, Romance Mendelssohn
Op. 28 No. 2;
Beethoven Sonata Op. 14
No. 2, 1953
Adelina de Lara Brahms Rhapsody Op. 79
(1872–1961) No. 2, Intermezzo Op. 117 No. 1,
1951; Schumann Fantasiestücke Op.
12 Nos. 1 and 2; Kinderszenen Op. 15
No. 1, 1951; Arabeske Op. 18, 1951

Table 2.2 Cont’d
Pianists Compositions with Frequent Compositions with Infrequent or
Dislocations No Dislocations

Ilona Brahms Waltzes Op. 39 No. 2 and Scarlatti Sonatas in E and G,

Eibenschütz No. 15, 1903; Ballade Op. 118 No. 3, 1903; Beethoven Sonata
(1873–1967) middle section only, 1903; Op. 109 2nd movement,
Intermezzo Op. 119 No. 2, 1952, c. 1950
Waltz Op. 39 No. 15, 1962
Landon Ronald Wagner Die Meistersinger Overture, Grieg Bridal Procession Op. 19
(1873–1938) Grieg Dance Caprice Op. 28 No. 3, No. 2, 1900
Chopin Polonaise Op. 40 No. 1, 1900;
various accompaniments to vocal
works with Adelina Patti including
Mozart “Voi che sapete” (Le Nozze di
Figaro), 1905
Ricardo Viñes Scarlatti Sonata L. 461; Gluck
(1875–1943) arr. Brahms Gavotte; Debussy
Soirée dans Grenade, 1930
Joseph Chopin Nocturne Op. 9 No. 2; Not in works of Schubert, Liszt,
Hofman Schumann Warum? Op. 12 Mendelssohn that are fast or in
(1876–1957) No. 3, 1912; Chopin Valse Op. 64 Beethoven Moonlight Sonata
No. 2 1916 Op. 27 No. 2, 1912
Etelka Freund Brahms, Sonata Op. 5, Intermezzo
(1879–1977) Op. 116 No. 2 and Intermezzo Op.
117 No. 2, 1953; Capriccio Op. 76
No. 1, 1950; opening of J. S. Bach
Prelude in E flat minor BWV 853, 1957
Frank la Forge Chopin Nocturne Op. 27 No. 2,
(1879–1953) 1912
Mark Chopin Nocturne Op. 55 No. 1, Bach-D’Albert Organ Prelude in
Hambourg 1921, Nocturne Op. 9 No. 2, 1927; D major BWV 532, 1921
(1879–1960) Beethoven Concerto
Op. 37 2nd movement, 1929;
Mendelssohn-Liszt On Wings of Song,
John Powell Chopin Nocturne Op. 27 No. 2, Chopin Scherzo Op. 20 No. 1,
(1882–1963) 1929, piano roll 1921, piano roll
Walter Beethoven Andante from Sonata Not in works of J. S. Bach or
Gieseking Op. 109; Brahms Intermezzo other works of Beethoven or
(1895–1956) Op. 117 No. 2, 1939–40; Brahms that are fast, or in works
Mendelssohn Andante and Rondo of Debussy, Poulenc, Fauré, or
Capriccioso, 1956 Scriabin

Peres da Costa-Ch02.indd 50 3/14/2012 6:01:40 PM

Play ing One Hand after the Other 51

Written Sources and Historical Precedents

Considering its widespread use around the turn of the twentieth century, men-
tion of dislocation in contemporary literature appears somewhat scant. As far as
I can tell, many of the detailed late-nineteenth-century pedagogical texts, includ-
ing those by Mathis Lussy (1828–1910) and Hugo Riemann (1849–1919), fail to
mention it directly.20 Interestingly, however, Lussy (and his coauthor) show aware-
ness of the technique in Histoire de la notation musicale (1882). They express a
desire to retain in modern music a sign used by François Couperin (1668–1733),
which expresses the suspension or retardation of a note until after the correspond-
ing bass note is sounded.21 Another influential pedagogue, Adolphe Christiani
(1836–85), may well have been referring to a range of practices including disloca-
tion in his discussion of rubato, but this is not clear.22
What is clear is that the practice of dislocation featured strongly in the
Leschetizky school. Leschetizky’s importance as a nineteenth-century performer
and pedagogue is obvious. He claimed to uphold the precepts of Czerny (with
whom he studied in Vienna) throughout his highly successful concert and teach-
ing career. And he was also apparently influenced by the sonorous, cantabile, and
legato style of the pianist and salon composer Julius Schulhoff (1825–98).23 From
1862 to 1878, he served as head of the piano department at the St. Petersburg
Conservatory, founded by Anton Rubinstein. In 1878, Vienna became Leschetizky’s
home. He played a central role in establishing its Tonkünstlerverein in 1881, a
society that met weekly providing an opportunity to hear the crème de la crème of
local and international talent as well as new works. Leschetizky soon became the
most sought-after teacher of the day. About 1,500 pianists benefited from his
guidance. While he devoted himself to developing the higher artistic ideals of
those students who were already capable, his teaching assistants did important
preparatory work with those who were less so. Leschetizky’s strength lay in
pinpointing weaknesses and advising of suitable remedies while respecting the
individuality of each student. His world-famous fortnightly piano class was a
very popular, if somewhat nerve-racking, semipublic forum at which visiting vir-
tuosi (some of them Leschetizky’s former students), as well as rising stars and
talented young recruits, would perform.24 For this, Leschetizky purposely made

Mathis Lussy, Traité de l’expression musicale, accents, nuances et mouvements dans la musique vocale
et instrumentale (Paris: 1874); Hugo Riemann, Der Ausdruck in der Musik (Leipzig, Germany: 1878) and
Katechismus des Klavierspiels (Leipzig, Germany: 1888).
Ernest David and Mathis Lussy, Histoire de la notation musicale depuis ses origines (Paris:
1882), 171.
Adolphe Christiani, The Principles of Expression in Pianoforte Playing (New York: Harper, 1885),
See Hamilton, After the Golden Age, 153.
Aniela Potocka, Theodor Leschetizky: An Intimate Study of the Man and Musician, trans. Geneviève
Seymour Lincoln (New York: Century, 1903), 295–302. Countess Aniela Potocka was Leschetizky’s
sister-in-law and had firsthand experience of his teaching.
52 off the record

the conditions as difficult as possible.25 Among those of his students who went on
to conduct active concert and teaching careers were Paderewski, Gabrilovich, Ney,
Moiseiwitsch, Friedman, Hambourg, Brailowsky, and Schnabel.
An informative reference about dislocation appears in Die Grundlage der
Methode Leschetizky (1902), by Malwine Brée (b. 1861), translated in the same
year as The Groundwork of the Leschetizky Method. Brée, a former Leschetizky
student, eventually became one of his eminent teaching assistants.26 Although
Leschetizky himself had no desire to produce a pedagogical text, he verbally
applauded Brée’s efforts in exposing thoroughly his personal views and he sub-
scribed “word for word” to everything in it.27 Regarding dislocation, she advises
the following, giving the example shown in Figure 2.2:

Neither should bass tone and melody-note always be taken precisely

together, but the melody-note may be struck an instant after the bass,
which gives it more relief and a softer effect. However, this can be
done only at the beginning of a phrase, and usually only on important
notes and strong beats. (It is better for the hands to coincide precisely

Figure 2.2 Chopin Nocturne Op. 27 No. 2, annotated by Brée.

Ethel Newcombe, Leschetizky as I Knew Him (New York: Appleton, 1921), 14–15. Ethel
Newcombe, a former Leschetizky student, opined that it was impossible to exaggerate the importance
of these classes.
Potocka, Theodor Leschetizky, 304.
Malwine Brée, Die Grundlage der Methode Leschetizky (1902), trans. Dr. Theodor H. Baker as
The Groundwork of the Leschetizky Method (New York: Schirmer, 1902), iv.
Play ing One Hand after the Other 53

on weak beats.) The melody-note must follow so swiftly as to make the

pause hardly noticeable for the uninitiated.28

Brée clearly describes dislocation as a localized event that results in a particular

expressive effect. As we will see later, her prescription for its use corresponds
quite closely to Leschetizky’s own practices as well as those within and outside
his circle.
Verification that Leschetizky strongly favored the use of dislocation comes
from the reminiscences of Frank Merrick (1886–1981), an English pianist who
studied with Leschetizky around 1905. During a lecture given in 1961, Merrick
recalled his teacher’s advocacy of practices that later become outlawed. The hands
were not always played together: indeed, in particular places “the right hand
should be played slightly before the left.” Merrick explains that at the turn of the
twentieth century, these things were regarded as intensifying expression, but
later as oversentimental.29 For successive generations, rhythmic precision and
stricter adherence to the score became increasingly the hallmark of the virtuoso.
Merrick remembered that for many years after his lessons with Leschetizky, he
was often accused of making dislocations “to an undesirable extent” but he was
not always aware that he was doing it. No doubt, like Pauer who unconsciously
played one hand after the other, dislocation was well assimilated into Merrick’s
palette of expressive techniques.30 Eventually, Merrick succumbed to the tide of
fashion, adopting a more judicious approach to its use and those of other expres-
sive freedoms.31
Dislocation was by no means a practice peculiar to the turn of the twentieth
century. It was commonly employed during the nineteenth century, very probably
in a more exaggerated fashion. Validation of this comes from the virtuoso pianist
Sigismund Thalberg (1812–71) in the mid-nineteenth century. In his L’Art du
chant appliqué au piano, Op. 70 (Paris, c. 1853)—a work that provides arrange-
ments of opera arias as instructive pieces for the piano—Thalberg cites eleven
rules to enlighten pianists about how literally to sing at the piano.32 In the fifth,
he describes succinctly the “pros and cons” of dislocation. There is a noticeable
resemblance in part to Brée’s description of dislocation fifty years later:

It will be indispensable to avoid, in playing, the ridiculous habit and in

bad taste, of withholding with exaggeration the production of the notes

Brée, The Groundwork, 72–73.
Frank Merrick, “Memories of Leschetizky,” Recorded Sound: The Journal of the British Institute of
Recorded Sound vol. 18 (1965): 336.
See note 1, page 41.
Merrick, “Memories of Leschetizky,” 336.
See Hamilton, After the Golden Age, 155–62.
54 off the record

of the melody a long time after those of the bass [have been sounded];
and thus of producing, from the beginning to the end of a composition,
the effect of continuous syncopations. In a slow melody written in notes
of long duration, it produces a good effect, above all on the first delivery
of each measure, or at the commencement of each period or phrase,
to sound the melody after the bass, but only with an interval nearly

For those interested in Romantic piano style, Thalberg’s rule is a golden nugget. It
documents the practice of dislocation in the mid-nineteenth century as a valued
expressive device used locally particularly in slow melodies. It also documents its
apparent abuse. Thalberg implies that some players used dislocation in both fast
and slow music, and that they applied it overly frequently with excessive time
lapses between melody note and accompaniment. No doubt these would have
sounded monotonous, clumsy, and ineffective. Notably, there is no mention of
melody notes sounding before the bass. Although Thalberg’s disapproval is unmis-
takable, there is little reason to believe that this was universally shared. In any
case, without audible evidence the nature and extent of such dislocations remains
Dislocation is positively advocated in another influential and widely circulated
pedagogical text contemporary with Thalberg’s. In their Grosse theoretische-prak-
tische Klavierschule (1858), Sigmund Lebert and Ludwig Stark advise that to give
emphasis, “one is allowed, and even should in most cases, play the melody notes
imperceptibly later than the accompaniment, which leads to a kind of ‘arpeggio.’”34
This explanation hints at an important distinction, albeit a subtle one, between
dislocation and chordal arpeggiation. Dislocation produced the effect of (a kind
of) arpeggio but was created by a means different from chordal arpeggiation. The
impression, too, is that dislocation was to be applied habitually.
Although the advice presented so far provides strong evidence of the use
of dislocation, the nature of the descriptive language raises some important
questions. What would a trained musician of the mid-nineteenth century have

Sigismund Thalberg, L’Art du chant appliqué au piano, Op. 70, 2 series (Paris: Heugel, 1853),
series 1, unpaginated 2: “Il sera indispensable d’éviter, dans l’exécution, cette manière ridicule et de
mauvais goût de retarder avec exagération le frappement des notes de chant longtemps après celles
de la basse, et de produire ainsi, d’un bout à l’autre d’un morceau, des effets de syncopes continues.
Dans une mélodie lente écrite en notes de longues durée, il est d’un bon effet, surtout au premier
temps de chaque mesure ou en commençant chaque période de phrase, d’attaquer le chant après la
basse, mais seulement avec un retard presque imperceptible.”
Sigmund Lebert and Ludwig Stark, Grosse theoretische-praktische Klavierschule für systematischen
Unterricht nach allen Richtungen des Klavierspiels vom ersten Anfang bis zur höchsten Ausbilding, 3 parts
(Stuttgart, Germany: 1858), part 3, 3: “man darf also und soll sogar in den meisten Fällen 1) die
Melodie unmerklich später anschlagen, als die Begleitung, was eine Art Harpeggio bewirkt.”
Play ing One Hand after the Other 55

perceived as a brief, imperceptible, or hardly noticeable delay between melody and

accompaniment? And how might this relate to the modern meaning of such
terms? Can we fully appreciate, today, the meaning of descriptive terminology of
Through the ages, instrumentalists have been advised to imitate the art of
good singers. So it is pertinent to consider any link between dislocation in piano
playing and the practices of singing. As it turns out, the celebrated nineteenth-
century singing teacher Manuel García (1805–1906) prescribed a type of disloca-
tion in vocal music. In García’s New Treatise on the Art of Singing (London, 1857),
another text contemporary with Thalberg’s, García advises that a dramatic change
of dynamic—forte to piano or pianissimo, for example—could be greatly enhanced
if the first melody note at the change is sung “an instant after the bass.” This
would provide relief after “the loud notes” that preceded and would prepare the
ear for the following effects no matter how delicate or subtle. Particular success
would be achieved when the first consonant after the rest was sung with vigor.35
The correlation between dislocation and the effect of relief and softening made by
García is particularly interesting because these were the two effects highlighted
by Brée.
To illustrate his point, García provides an annotated example (Fig. 2.3) from
Rossini’s opera Otello.36 His advice to “strike the C after the bass” refers to the
beginning of the word consolar. There is an unmistakable similarity between this
expressive vocal technique and dislocation in piano playing. Notably, García pre-
scribes it as a means of emphasizing a change to a softer dynamic. There is no
mention, as one might expect, of rubato. In fact, he discusses rubato elsewhere, in
a section all of its own. García seems to have considered this apparently localized
effect to be different to rubato despite its outward appearance. This may hold an
important key to understanding the origin of dislocation in piano playing.

Figure 2.3 Rossini Otello, annotated by García.

Manuel García, García’s New Treatise on the Art of Singing: A Compendious Method of Instruction
(London: Beale & Chappell, 1857), 55.
García, García’s New Treatise, 55.
56 off the record

Although many now think of it as a bastardized rubato effect, its purpose is more
likely to have been to enhance dynamic rather than rhythmic variation.

Older Traditions: Vocalists, Lutenists, and Clavecinistes

Making expressive silences by delaying the melody note may, as Julian Rushton
suggests, have derived from a vocal device.37 That is certainly the impression
gained from Sébastien de Brossard’s Dictionnaire de la Musique (1701). In the third
edition (c. 1708), Brossard gives an explanation of the mezzo-sospiro (an Italian
singing term or ornament) known in French as the demi soupir (the latter was
sometimes used to describe the French ornament called suspension). Initially,
Brossard states that this ornament is shown by a figure (resembling a modern
quaver rest) that is used in all time signatures and is realized by “keeping silent for
the eight part of a bar.” Later in the explanation he adds that “it is better to say
that you must be silent for the equivalent value of a quaver depending on what
time signature we sing or play.”38
If you are a lutenist or an early-keyboard player you might well be nodding
knowingly. Dislocation between melody and accompaniment is a technique well
documented in seventeenth- and eighteenth-century texts, particularly from
France. And it is certainly well and alive in lute and early keyboard playing today.
Yet the notion that the type of dislocation heard in early piano recordings may be
rooted in this older tradition has been fairly understated and warrants detailed
During the second half of the seventeenth century or earlier, French lutenists
sometimes separated a melody note from its corresponding bass note for
special expressive effect. In the Livre de tablature des pièces de luth (c. 1672), Denis
Gaultier (1603–72) and Ennemond Gaultier (c. 1575–1651) marked an oblique
line between two or more notes signifying that they “must be separated one after

Julian Rushton, “Suspension,” The New Grove Dictionary of Music and Musicians, 2nd ed., ed.
Stanley Sadie (London: Macmillan, 2001): vol. 24, 733. Donington has pointed out the resemblance
between the suspension and the hocket, a medieval vocal compositional technique in which the notes
of a melody are shared between multiple (sometimes more than two) voices bringing about the alter-
nation of the voices—one sounds while the other rests. See Robert Donington, The Interpretation of
Early Music, new rev. ed. (New York: W. W. Norton, 1992), 281.
Sébastien de Brossard, It. mezzo-sospiro; Fr. demi soupir, Dictionnaire de la Musique (Paris: 1701;
3rd ed. Amsterdam: c. 1708; reprint, Geneva: Minkoff, 1992), 59: “Il marque qu’il faut se taire pendant
la huitième partie d’une mesure. Quand c’est un mouvement de quatre temps, it faut se taire pendant
la huitième partie d’une mesure, mais si c’est un mouvement de 3 ou 3/4 il ne faut se taire que pen-
dant une sixième partie de la mesure et si c’est un mouvement de 6/4 il ne faut se taire que pendant la
douxième partie et ainsi de divers autres mesures; je crois dont qu’il vaut mieux dire qu’il faut se taire
pendant la valeur d’une croche figurée ainsi quelque mesure que l’on chante ou que l’on joue.”
Play ing One Hand after the Other 57

Figure 2.4 Perrine, Harpègement or séparation in lute playing.

the other.”39 In his Pièces de luth en musique (Paris, 1680), the theorbist and lute
teacher Perrine (b. seventeenth century) also adopted the oblique line notation
(Fig. 2.4), offering the same explanation as the Gaultiers.40 He called the effect
harpègement or séparation.41
Similar use of the oblique line notation can also be found throughout the
extensive collection of lute pieces by diverse authors entitled Manuscrit Barbe
(c. 1690).42 In all of these examples, the notation says nothing of the degree of
dislocation, which was presumably left to the taste of the player, the needs of the
music, and/or the circumstance of the performance.
During the second half of the seventeenth-century, French keyboard players
(those who played harpsichord, spinet, virginals, or clavichord) followed suit.
Regarded as an invaluable expressive device, dislocation was notated with various

Denis Gaultier and Ennemond Gaultier, Livre de Tablature des Pieces de Luth (c. 1672; reprint,
Geneva: Minkoff, 1975), 4: “La ligne qui se trouve marquée de travers ou qui est tirée obliquement
entre deux ou plusieurs lettres, signifie que les deux lettres doivent estre [sic] separées l’une apres [sic]
lautre [sic] en les touchant.”
Perrine, “Introduction” to Pièces de luth en musique (Paris: 1680; reprint, Geneva: Minkoff, 1982),
6; for a possible realization of this, see Hudson, Stolen Time, 24.
Perrine, Pièces de luth en musique, 6: “La ligne obliquement tirée entre les notes comme [Fig. 2.4]
signifie qu’il les faut toucher l’une après l’autre.”
Manuscrit Barbe: Pièces de luth de différents auteurs en tablature française (Paris: c. 1690;
reprint, Geneva: Minkoff, 1985).
58 off the record

symbols in French keyboard music of the period.43 Like Perrine, Gaspard Le Roux
(c. 1660–c. 1707) adopted the oblique line to show both separation of notes in a
chord and dislocation between the hands in the Sarabande from Suite in F Major
(Fig. 2.5 ),44 the Chaconne from Suite in D Minor, and “La Favorite” from his
Pièces de clavecin (1705).
More popular, however, was the notation of the suspension. In his first book of
Pièces de clavecin (1713), François Couperin gave a pictorial explanation for the
suspension showing the delay of a melody note (Fig. 2.6).45 Four years later, in his
L’Art de toucher le clavecin (1717), Couperin spoke of the difficulty of giving soul
to the harpsichord in order to make it expressive. He attributed the success of
his own playing to “the cessation and suspension of tones, made opportunely
and according to the character of the melody of the preludes and pieces.”46 To
achieve this, he suggests the use of two ornaments, the aspiration (shortening
notes to create breathing points) and the suspension, explaining that “these two
ornaments by their contrast leave the ear undetermined in such a way that,
in those places where the bowed instruments would increase [swell] their tone,

Figure 2.6 Couperin, pictorial explanation of the suspension. Source: Extract of the
facsimile edition “Pièces de clavecin—Premier livre” of François Couperin (Réf. 1982)—Anne
Fuzeau Production—

More recently, the effect has been termed “staggered execution” and “hands-apart playing.”
See Mark Kroll, Playing the Harpsichord Expressively: A Practical and Historical Guide (Lanham, Md.:
Scarecrow Press, 2004), 44.
Gaspard Le Roux, “Sarabande from the Suite in F major,” Pièces de clavecin (Paris: 1705; reprint,
Geneva: Minkoff, 1982), 19.
François Couperin, Premier livre de pièces de clavecin (Paris: 1713; reprint, Courlay, France: Fuzeau
1988), unpaginated 75.
François Couperin, L’Art de toucher le clavecin (Paris: 1713; reprint Geneva: Minkoff, 1986), 16:
“la cessation; et à la suspension des sons, faites à propos; et selon les caractères qu’èxigent les chants
des préludes, et des pièçes.”
Play ing One Hand after the Other 59

the suspension at the harpsichord, by a contrary effect, seems to produce this

desired result.”47
Clearly, Couperin employed the suspension to give the impression of dynamic
shading. Being a harpsichordist myself, I can confirm that dislocating the melody
from the bass draws special attention to it. When skillfully done, it can give the
illusion of emphasis or increased dynamic to the melody note. And it can also have
a softening effect, though Couperin did not mention that specifically.48 Giving
dynamic expression to a single melody note is otherwise difficult or impossible on
the harpsichord. Couperin goes on to explain that the suspension is rarely used,
other than in tender and slow works. Importantly, the length of silence that pre-
cedes the note marked with the suspension “must be regulated by the good taste”
of the player.49
Couperin himself notated the suspension, rather sparingly, in slow pieces of
tender character such as Les Sentimens (Fig. 2.7 ),50 Les Laurentines and La
Tendre Fanchon from the first book of Pièces de clavecin, and in La Princese de Sens
and La Castelane from the second book (1722). Curiously, he no longer notated
the suspension in the third book (1722), nor in the fourth book (1730). Perhaps, as
Peter Le Huray suggests, Couperin concluded that the use of the suspension should
be guided by the players’ instinct.51 With that in mind, it is likely that harpsi-
chordists employed the suspension as many times as they felt necessary. And they
probably used it in a variety of situations too.
Other French Baroque composers also made use of the suspension. Jean-
Philippe Rameau (1683–1764) included it in his table of ornaments and notated
it, again somewhat sparingly, in Les Soupirs and L’Entretien des Muses from his
Pièces de clavecin (1724, revised in 1731) and in the Premier Minuet from the
Deuxième Concert of the Pièces de clavecin en concerts (1741). Suspension is notated
fairly frequently in Joseph-Nicolas-Pancras Royer’s (c. 1700–1755) Les tendres
Sentiments (Fig. 2.8 ) from his Pièces de clavecin (1746).52 Here, the suspension
highlights moments of poignant dissonance and or dramatic effect.

Here, the word suspension appears to refer both to the cessation and delay of sounds effected by
the aspiration and the suspension, respectively.
See also Richard Troeger, Technique and Interpretation on the Harpsichord and Clavichord
(Bloomington: Indiana University Press, 1987), 133–38.
François Couperin, L’Art de toucher, 18: “A l’égard de la suspension! elle n’est gueres usitée que
dans les morceaux tendres, et lents. Le silence qui précéde [sic] la note sur laquelle elle est marquée
doit être réglé par le goût de la personne qui exècute.”
Couperin, “Les Sentimens,” Premier livre de pièces de clavecin, 11.
Peter Le Huray, Authenticity in Performance: Eighteenth-Century Case Studies (Cambridge:
Cambridge University Press, 1990), 54.
Joseph-Nicolas-Pancras Royer, Les tendres Sentiments, Pièces de clavecin (Paris: 1746; reprint,
Geneva: Minkoff, 1981), 21.
60 off the record

Figure 2.9 Foucquet, pictorial explanation of the suspension combined with the trill
and mordent.

Like Couperin and Rameau, Pierre-Claude Foucquet (c. 1694–1772) gave a

pictorial explanation for the suspension in the table of ornaments in his Premiere
et deuxième livre de pièces de clavecin (Paris, 1751). And he notated it sparingly in
La Soeur Agnes ou La novice. Interestingly, his table of ornaments also shows the
cadence et pincé suspendu—the combination of suspension with both the trill and
the mordent (Fig. 2.9).53 Although I cannot find an example of either in his music,
it is likely that this type of delay was to be applied to both types of ornaments
(trill and mordent) at the discretion of the player.
Foucquet was not alone in advocating this type of delay to an ornament. In his
Pièces de clavecin (1735), Louis-Claude Daquin (1694–1772) gave the example
shown in Figure 2.10,54 explaining that “to play a port de voix properly, it is
indispensable when the little note [grace note] is slurred [to the main note], to
play the bass note a little before the grace note in the top part and to lean on the
grace note a little more strongly before playing the mordent.”55 Daquin’s use of the
port de voix with the pincé (mordant) is particularly noticeable in Les trois Cadances
and in the tender pieces La Favorite and La Tendre Silvie.
Other significant evidence of dislocation is found in the many examples of
French unmeasured prelude or prélude non mesuré. As Davitt Moroney has rightly
pointed out, these are extremely instructive, providing information about “articu-
lation, ways of spreading a chord, arpeggiation, resolving trills, and the life which

Pierre-Claude Foucquet, Premiere et deuxième livre de pièces de clavecin (Paris: 1751; reprint,
Geneva: Minkoff, 1982), 5.
Louis-Claude Daquin, “Avertissement,” Premier livre de pièces de clavecin (Paris: 1735; reprint,
Geneva: Minkoff, 1982), unpaginated.
Daquin, “Avertissement,” unpaginated: “Mais je dois observer que pour bien faire un port de voix
il est indispensable quand la petite notte est lieé de toucher la notte de la basse un peu devant la petite
notte du dessus et d’appuyer la petite notte du dessus un peu plus fort avant que de faire le pincé.”
Play ing One Hand after the Other 61

Figure 2.11 Couperin Unmeasured Prelude in C Major.

can infuse the port de voix, the aspiration, and above all the suspension.”56 A highly
arpeggiated style incorporating not only arpeggiated chords, but also many
instances in the notation in which a melody note is placed slightly after the bass
note, is revealed in the Unmeasured Prelude in C Major (Fig. 2.11) by Louis
Couperin (c. 1626–61).57 Indeed, in his unmeasured preludes from the Premier
livre de pièces de clavecin (Paris, 1704), Louis-Nicolas Clérambault (1676–1749)
went a step further by notating dotted lines to show where the right and left
hands should be either synchronized or played apart (Fig. 2.12).58 The frequency
with which such dislocation occurs in these and other unmeasured preludes pro-
vides important clues to the general style of harpsichord playing in France during
this period.
The general sparseness in which dislocation was notated using the suspension
in French harpsichord music gives the outward impression that it was used spar-
ingly for special effect. However, it is probable that French composers bothered to
notate dislocation (in the ways discussed earlier) only at places where the effect
was considered absolutely essential, expecting its addition elsewhere at the will of

Figure 2.12 Clérambault Unmeasured Prelude in C Major.

Davitt Moroney, “The Performance of Unmeasured Harpsichord Preludes,” Early Music vol. 4,
no. 2 (Apr. 1976): 151.
Louis Couperin, “Prelude in C Major,” Bayn Manuscript (c. 1658), ed. and typeset Steve Wiberg
(Due West Editions, 2009), 33.
Louis-Nicolas Clérambault, “Prelude in C Major,” Premier livre de pièces de clavecin (Paris: 1704;
reprint, Geneva: Minkoff, 1982), 1.
62 off the record

Figure 2.13 Duphly Les Grace, bar 14.

the player. Some important evidence supports this notion. By the middle of the
eighteenth century in Paris, a more liberal application of dislocation seems to
have been sanctioned, at least for particular repertoire. In this respect, the
Troisième livre de pièces de clavecin by Jacques Duphly (1715–89) provides some
interesting examples. For the tender piece Les Graces (Fig. 2.13),59 Duphly explains
that “the dots which are on (or over) the bass notes signify that it is necessary to
play them [literally they must pass] before those of the upper part (melody).”60 In
bar 14, for example, he marks the dots on several adjacent bass notes. Additionally
and somewhat curiously, one suspension is marked on the fifth melody note. Did
these two signs (the dot and the suspension) signify differing amounts of delay?
Duphly must have intended the same effect in the tender gavotte La De Villeneuve
where the dots are notated again.
Jean-Baptiste Forqueray’s (1699–1782) arrangements for harpsichord of some
of the Pièces de viole by his father Antoine Forqueray (1671–1745) provide more
examples in which dislocation occurs frequently. For a successful interpretation
of the Sarabande La D’aubonne (Fig. 2.14 ) from Suite No. 4,61 Forqueray advises
“a great deal of taste and sentiment.” He adds that

to make it intelligent [presumably for the proper interpretation] I have

marked some little crosses which signify that it is necessary that the
chords of the bass be played [literally—pass] before those of the melody;

Jacques Duphly, Troisième livre de pièces de clavecin (Paris: 1756; reprint, Geneva: Minkoff,
1987), 16.
Duphly, “Les Grace,” Troisième livre, 16: “Les points qui sont sur les notes de basse signifient
qu’il faut les passer avant celles du dessus.”
Jean-Baptiste Forqueray, “La D’aubonne,” Pièces de viole mises en pièces de clavecin (Paris: 1747;
reprint, Geneva: Minkoff, 1987), 24.
Play ing One Hand after the Other 63

and at all those places where there are no crosses, the melody must be
played before the bass.62

Forqueray also showed what he intended by misaligning the right and left hands
in the notation.
It is extraordinary that if Forqueray’s instruction were followed to the letter,
not many beats in La D’aubonne would be synchronized between the hands. But it
is also rather astonishing to find that the right hand would precede the left just as
many times as the other way. This is particularly significant when we recall that
over a century later, Leschetizky advocated right hand before left in certain situa-
tions. It is possible, of course, that this style of dislocation for the harpsichord
arrangement of La D’aubonne was in fact unusual. Perhaps it helped to reflect in
some way the natural placement of the melody note (before or after the bass and
other accompanying notes in this highly chordal piece) created by the push fol-
lowed by pull strokes of the viola da gamba. On the other hand, given that harpsi-
chordists may well have made these types of dislocations as a matter of course,
Forqueray might only have bothered to leave such specific instructions as a means
of organizing and varying their use in this unusually complex texture.
A rather striking example of the practice of dislocation is found in the Sarabande
marked tendrement (tenderly) entitled La Léon from Forqueray’s Suite No. 5
(Fig. 2.15 ).63 Here, again, Forqueray provides a visual portrayal of dislocation
by purposely misaligning the right and left hands in the notation. He explains
that “to play this piece in the taste” that he would like, attention must be paid
to the manner of writing—“the upper part is hardly ever found exactly together
with the bass.”64
Yet more telling evidence exemplifying the frequent use of dislocation is
found in the preface to Foucquet’s Deuxième livre de pièces de clavecin (1750–51).
He prescribes its general use, stating emphatically that “in all pieces of a gracious
or tender execution, one should play the note of the bass, before that of the upper
part, without altering the beat, which produces a suspension on each note of the
upper part.”65 Included are a number of pieces of gracious or tender feeling to

Forqueray, Pièces de viole mises en pièces de clavecin, 24: “Cette pièce doit être jouée avec
beaucoup de gout et de sentiment: pour en donner l’intelligence, j’ay marqué des petites croix qui
signifient qu’il faut que les accords de la basse, passent avant ceux du dessus; et à tous où ils ne s’en
trouvera point, le dessus doit passer avant la basse.”
Forqueray, “La Léon,” Pièces de viole mises en pièces de clavecin, 30.
Forqueray, Pièces de viole mises en pièces de clavecin, 30: “Pour jouër cette piece dans le gout que
je souhaiterois qu’elle fut jouée, il faut faire attention à la façon dont elle est écrite, le dessus ne se
trouvant presque jamais d’aplomb avec la Basse.”
Foucquet, “Preface,” Deuxième livre, unpaginated: “Dans toutes les pièces d’éxécution gracieuse
ou tendre, on doit toucher la note de basse, avant celle de dessus, sans altérer la mesure, ce qui opère
une suspension sur chaque note du dessus.”
64 off the record

Figure 2.16 Jollage La Marais. Source: Extract of the facsimile edition “Premier livre de pièces
de clavecin” of Charles Alexandre Jollage (Réf. 50105)—Anne Fuzeau Production—

which this would surely apply. But, as if to reinforce the idea, Foucquet went to
the further trouble of verbally prescribing a tender and arpeggiated style of
playing with the bass played before the melody in L’Angelique and La Mimi.66
If such dislocation was intended for every beat without exception, then clearly the
effect (or something like it) mentioned by Thalberg a century later, of continuous
syncopation caused by dislocation, was already heard in Foucquet’s time. This
notion is further corroborated by the notation of the suspension in both La Marais
(Fig. 2.16)67 and La Tranquille from the Premier livre de pièces de clavecin (Paris,
1738) by Charles-Alexandre Jollage (d. 1761). In these, the effect of continuous
dislocation between the hands is unmistakable.
The picture so far is that localized dislocation (both anticipation and delay) was
practiced extensively in France during the Baroque era. Sometimes it was used to
an extent that caused continuous asynchrony between consecutive notes in a
melodic sequence and their corresponding bass notes. It is probable, too, that it
was heard more often and in a wider range of contexts than was notated.

Other Countries
Elsewhere the picture is very unclear. There is a surprising lack of written evi-
dence about dislocation in English, German, or Italian sources. In recent times,
particular scholars have linked the French suspension or dislocation style (which
they call “fringing”) with certain comments by the Englishman Roger North
(1653–1734), who wrote on a wide range of subjects including music. In his essay
The Musicall Gramarian (c. 1726), North describes what might seem to us now an
extraordinary practice in violin playing, a way of “gracing” or giving life to passing
notes producing in them “a fullness scarce describable.” This life is achieved not by
the usual means of adding a trill, mordent, or alternative ornament, but by “a sort
of shock or concussion of the finger upon every touch, as if every note were torne

Foucquet, Deuxième Livre, 15–16: “Tendrement et Harpegé—La basse avant le dessus.”
Charles-Alexandre Jollage, La Marais, Premier livre de pièces de clavecin (Paris: 1738; reprint,
Geneva: Minkoff, 1985), 14–15.
Play ing One Hand after the Other 65

discerptìm [tearing in pieces] from another.” North explains that this results in a
very short mingling of each note with the one before or after, “which doth not
corrupt or stay, but rather fringes the tone, as colours are seen onely at the
entrance of a refraction.”68 It is difficult to know exactly what effect was intended
here, though it seems that adjacent melody notes would sound overlapped. There
is no mention of anticipation or delay between melody and bass. Elsewhere in The
Musicall Gramarian, North refers to the performance on two violins of the same
“lesson” (or music), advising that the musical result is better when one part is
played “a little before or behind the other,” rather than when both play “zealously”
together. North adds that in the latter, nothing is gained by the precise doubling
than a little extra volume. While in the “frequent dissonances” (mingling of
sounds) created by the former, “there is a pleasant seasoning obtained.” He also
speaks of this type of mingling of sounds in consort playing.69 As before, there is
no mention of anticipation or delay between melody and bass note. It is not cer-
tain whether these two separate descriptions refer to the same practice, though
there are some obvious similarities.
In 1967, violinist Sol Babitz (1911–82) linked North’s finger concussion in violin
playing that “fringes” or mingles the tone with what he described as “an expressive
spoken effect” in keyboard (presumably harpsichord) playing. To illustrate the
point he gave an annotated example from J. S. Bach’s Invention BWV 772 with
dotted line arrows to show “the merest hint of non-simultaneous attack” between
melody and bass.70 Later, in 1976, Babitz prescribed fringing, among other things,
as an antidote to the modern “sewing machine” style of performance:

Modern “as-written” precision stemming from a reaction to the excesses

of late Romanticism sounds dry even when performed by those who try
to transcend the notes and must be replaced with the Swing, Snaps,
“Fringing” and other humanizing devices used since the Middle Ages to
make equally written notes expressive.71

In 1987, Robin Stowell amplified Babitz’s term “Fringing,” explaining it as “note-

displacement resulting from a kind of syncopated interpretation.”72

John Wilson ed., Roger North on Music: Being a Selection from His Essays Written during the Years
c. 1695–1728 (London: Novello, 1959), 169; the reference appears in North’s The Musicall Gramarian
or A pratick Essay upon Harmony, plain and artificial etc (c. 1726).
Wilson, Roger North, 172–73. According to Wilson, North “had a fondness for diatonic clashes
[dissonances] of various kinds” and “regarded all diatonic intervals as being concordant in some degree.”
Sol Babitz, “Concerning the Length of Time that Every Note Must Be Held,” The Music Review
vol. 28 (1967): 34.
Sol Babitz, “Style and the Performer,” Early Music vol. 4, no. 4 (Oct. 1976): 491–93.
Robin Stowell, “Bach’s Violin Sonatas and Partitas,” The Musical Times vol. 128, no. 1731 (May
1987): 253.
66 off the record

Given North’s explanation, Babitz’s link does not seem to me to be valid. I am

more inclined to accept John Wilson’s explanation that North “seems to describe
a combination of forceful bowing and smart finger-stopping, but with the finger-
change before or after the bow-change.”73 Be that as it may, Babitz’s explanation
apparently gave license for the term “fringing” to be applied interchangeably with
suspension. Subsequently, Anthony Newman linked both North’s finger concus-
sion and his advice about two violins playing in a nonsimultaneous manner to the
suspension. And Richard Troeger and Igor Kipnis have also mentioned fringing
and suspension in the same breath.74
In the end, it is difficult to make any concrete conclusion about the correspon-
dence between North’s writings and dislocation in English keyboard playing of
his era. Despite this, one might readily assume that something like the French
practice was adopted in post-Restoration England. It has been established that
keyboard composers of that era were strongly influenced by French harpsichord-
ists. As Michael Burden explains, this was “not only in their ornamentation, but
also (and even more obviously) in their wide-spread use of contemporary French
dance forms and stile brisé techniques.”75
This French influence also extended to German musicians.76 We know that
J. S. Bach (among others) incorporated many French stylistic elements into his
compositions. This is exemplified in Bach’s ornament signs and their interpreta-
tion as given in the Clavier-Büchlein vor Wilhelm Friedemann Bach (1720) that cor-
respond closely with those of Bach’s French counterparts. We also know that Bach
was enamored of the French style of harpsichord playing. Throughout his career,
he had close contact with French musicians, some of them notable, including
Louis Marchand (1669–1732), who was famous for his “beautiful and very correct
style of playing” and whose harpsichord touch Bach very much admired.77 In this
respect, Moroney has quite rightly asked us to consider what might be “essentially
French” about Bach’s Germanic keyboard style, not only his “intellectual appro-
priation of compositional styles,” but also his “Spielmanieren and his specific way
of playing, of touching the instrument.”78

Wilson, Roger North, 169 n. 32.
Anthony Newman, Bach and the Baroque: A Performing Guide to Baroque Music with Special Emphasis
on the Music of J. S. Bach (New York: Pendragon Press, 1985), 87; Troeger, Technique and Interpretation,
133; Igor Kipnis ed., The Harpsichord and Clavichord: An Encyclopedia (New York: Routledge, 2007),
Michael Burden, Performing the Music of Henry Purcell (Oxford: Clarendon, 1996), 86.
Troeger, Technique and Interpretation, 169–70.
Davitt Moroney, “Couperin, Marpurg and Roeser: A Germanic Art de Toucher le Clavecin,
or a French Wahre Art?” in The Keyboard in Baroque Europe, ed. Christopher Hogwood (Cambridge:
Cambridge University Press, 2003), 119.
Moroney, “Couperin, Marpurg and Roeser,” 118.
Play ing One Hand after the Other 67

Did Bach make use of the French suspension in his harpsichord playing? The
answer based on informed intuition, is almost certainly yes. That said one might
expect to find some mention of the device in his son Carl Philip Emmanual Bach’s
(1714–88) influential Versuch über die wahre Art das Clavier zu spielen (1753/1762),
but as far as I can see, this is not the case.79 Nor is it discussed directly in Daniel
Gottlob Türk’s Klavierschule (1789).80 On the other hand, in both Anleitung zum
Clavierspielen (1755) and Principes du clavecin (1756), Friedrich Wilhelm Marpurg
(1718–95) prescribes the French suspension sign as an alternative way of notating
Italian-style rubato involving the delay of consecutive notes in a melodic line.
Reportedly, Marpurg spent a significant time in Paris in 1746, during which he
would almost certainly have earwitnessed the French suspension. As Hudson
explains, Marpurg seems to have rationalized “two different concepts of synco-
pated displacement.”81 The first is the Italian type, discussed by Johann Joachim
Quantz (1697–1773) and others of the Berlin school involving the shortening,
lengthening, and/or addition of rests (silences) to consecutive notes in a melody
(see chapter 3). The second is the French suspension style of Couperin and others,
which as we have seen was generally prescribed for localized effect though also
sometimes affected consecutive notes in a melody. Marpurg was seemingly con-
vinced that these practices amounted to one and the same thing. He explains that
certain composers employed the suspension sign instead of the rest “in order to
mark the tempo rubato of a note.” But in order “to avoid unnecessary signs,” he
advises the use of the latter (the rest).82 There is, however, no clear evidence that
the French considered the suspension as a type of tempo rubato in the Italian
sense. Any theory about the similarity between the two remains conjectural.
Presumably Italian keyboard players of the Baroque era also employed disloca-
tion. Indeed, they may have been the first to use it in imitation of the Italian vocal
ornament the mezzo-sospiro.83 When, in 1615, Girolamo Frescobaldi (1583–1643)
advised that to fill out the sound on plucked keyboard instruments the begin-
nings of his Toccatas “should be played slowly and arpeggiated,” he may well have
envisaged, in addition to chordal arpeggiation, the use of dislocation (a kind of

Carl P. E. Bach, Versuch über die wahre Art das Clavier zu spielen, vol. i (Berlin: 1753, rev. 2nd ed.
1787), vol. ii (Berlin: 1762, rev. 2nd ed. 1797); repr. of 1st ed., incl. revs. of 1787 as a separate sec-
tion (Leipzig: 1787); trans. and ed. William J. Mitchell as Essay on the True Art of Playing Keyboard
Instruments (New York: Norton, 1949).
Daniel G. Türk, Klavierschule oder Anweisung zum Klavierspielen für Lehrer und Lernende mit
kritischen Anmerkungen (1789), 2nd enlarged ed. (1802), trans. Raymond H. Haggh as School of Clavier
Playing (Lincoln: University of Nebraska Press, 1982).
Hudson, Stolen Time, 115–16.
Friedrich W. Marpurg, Principes du clavecin (Berlin: 1756; reprint, Geneva: Minkoff, 1974), 52.
See my earlier discussion (this chapter) on the mezzo-sospiro.
68 off the record

arpeggiation) from time to time.84 Such an interpretation might also be implied in

the foreword to Frescobaldi’s Capricci (1624) where he advises that “in some dis-
sonances one should pause, arpeggiating them, so that the following passage
comes out more spiritedly.”85
Certain rhythmic notations in particular compositions may also lend support
to the idea that dislocation was used widely. The arpeggiated or stile brisé type of
figurations in Henry Purcell’s highly expressive Grounds86 could be seen as writ-
ten out dislocation between right and left hands (Fig. 2.17 ). The same could be
said of Wolfgang Amadeus Mozart’s staggered rhythms for the sighing figures in
bars 86 and 87 of his poignant Rondo in A minor K 511 (Fig. 2.18 ). And Johann
Sebastian Bach’s notation in bars 194–199 in the third movement of his D Minor
Harpsichord Concerto BWV 1052 (Fig. 2.19 ) certainly produces the effect of
dislocation and visually matches Couperin’s explanation of the suspension. In all of
the foregoing, the rhythms would no doubt have been interpreted more flexibly
than notated. Moreover, such effects were probably not restricted to the places in
which they appear. At the very least, these and other examples suggest that a
dislocated style was by no means foreign to keyboard playing of the seventeenth
and eighteenth centuries.
The paucity of written reference about dislocation (save in France) is somewhat
mysterious. Perhaps the practice had already become so integrated and “normal”
that it needed little or no discourse. The French may simply have codified (as they
so often did with other types of ornaments) a practice that was prevalent for some
time in Europe. After all, plucked keyboards and clavichords had been around for
a significant period before any mention of the suspension in French sources. In
the end, we will have to make up our own minds regarding the use of dislocation
outside of France.
The effect of dislocation is so expressive and seductive that many harpsichord-
ists today, including myself, use it frequently and in a wide range of repertoire.
This is sometimes met with disapproval. “He employs Couperin’s expressive
suspension—non-synchronization of the hands—to a degree that I find much

Girolamo Frescobaldi, “Al Lettore,” Toccate e partite d’intavolatura di cimbalo et organo . . . libro
primo (Rome: 1615; reprint, Florence: SPES, 1980), unpaginated: “Li cominciamenti delle toccate sieno
fatte adagio, et arpeggiando.” See also Luigi F. Tagliavini, “The Art of ‘Not Leaving the Instrument
Empty’: Comments on Early Italian Harpsichord Playing,” Early Music vol. 11, no. 3 (Jul. 1983):
Girolamo Frescobaldi, Il primo libro di capricci fatti sopra diversi soggetti et arie in partitura (Rome:
1624). Tagliavani suggests that dissonant chords with suspended notes “often mark the beginnings of
brief improvisatory transitional episodes between the sections of capriccios and canzoni; such disso-
nances are generally followed by a fanciful passage, which should be played ‘spiritoso.’” See Tagliavani,
306–7 n. 8.
See Purcell’s Ground in C Z. T681; Ground in c Z. D221; Ground in e Z. T682 variant; A new
Ground in e Z. T682.
Play ing One Hand after the Other 69

overdone, particularly in the allemandes and sarabandes,” explains John Kitchen

of American harpsichordist David Coates’s 2003 recording of J. S. Bach’s French
Suites BWV 812–817.87 This is, of course, a matter of taste. Be that as it may, such
a critique lacks a point of reference.

The Ascendancy of the Piano

As the piano—in all its early forms—gained popularity through the course of the
eighteenth century, we can assume, as Kroll has, that the practice of dislocation
was taken over and continued on without interruption through the nineteenth
century and into the twentieth.88 It surely remained an indispensable means of
giving heightened expression to particular melody notes on instruments that
were softer and thinner in tonal quality, particularly in the treble region, than
the fuller, more rounded sounding pianos that developed during the nineteenth
century. With that in mind, Thalberg’s censuring of the “effect of repeated synco-
pations” may signal, in the mid-nineteenth century, a move toward a more
judicious use of dislocation than was previously the case.
Significantly, some early-nineteenth-century references to the articulation
sign now commonly described as “portato” or “slurred staccato” provide some-
thing of a missing link. Initially, the sign was associated with clavichord playing.
C. P. E. Bach, Türk, and others refer to it as Tragen der Töne or appoggiato and
describe an effect related to the bebung—a type of vibrato on the clavichord. For
C. P. E. Bach, it is a drawing out of the notes in which each “receives in the same
time a perceptible pressure.”89 Türk explains that “the dot indicates the pressure
which every key must receive” and the curved line reminds the player “to hold the
tone out until the duration of the given note has been completed.”90 A character-
istic that sets the clavichord apart from all other keyboard instruments is the
possibility of maintaining contact with and applying extra pressure to the string
after the tangent has struck it. This extra pressure causes a slight bending of the
pitch producing a highly expressive effect that is not possible on the harpsichord
or on the piano.
The portato sign was adopted into piano repertoire from the late eighteenth
century onward. Friederich Starke’s Wiener Pianoforte-Schule (1819) explains that
the portato or appoggiato “is the least staccato” for which the notes are carried
equally. To achieve this, each note must be played with gentle pressure and held

John Kitchen, “Harpsichord Music from England and Germany,” Early Music vol. 33, no. 3
(Aug. 2005): 531–32.
Kroll, Playing the Harpsichord Expressively, 44.
Bach, Versuch, 126; see also David Rowland, Early Keyboard Instruments: A Practical Guide
(Cambridge: Cambridge University Press, 2001), 67–68.
Türk, Klavierschule, 343.
70 off the record

for three-quarters of its notated length.91 Johann Nepomuk Hummel (1778–

1837) stated in 1828 that the portato sign usually appears in music of a singing
character and that the notes must “be gently detached” and “receive a certain
increasing of emphasis.”92 Like Starke, Ignaz Moscheles (1794–1870) advised in
1827 that notes marked portato must sound for three-quarters of their length
[in fast movements]. In slow movements, however, they must sound for nearly
their full length, so that there is a “very slight break” between them.93 The expla-
nation of portato as a type of perceptible pressure or accent was still being upheld
in singing as late as 1880. In his primer entitled Singing, Alberto Randegger—
professor at the Royal Academy of Music in London—advises that notes marked
portato “should be sung in a marked but somewhat less detached manner” than
other types of staccato indicated by dots or wedges (Fig. 2.20 ).94 Today, portato
is generally interpreted as an articulation somewhere between legato and staccato.
For example, Robert Taub describes it as “semidetached; detached but weightier
than staccato.”95
But early in the nineteenth century, certain pianists, evidently aware of the
effect of portato on the clavichord, assigned an alternative though no less expressive
effect for the portato sign in piano playing. In his Méthode du piano du conservatoire
(c. 1804), Louis Adam (1758–1848) advises the player that to render a successful
portato (Fig. 2.21), “one must not jab at the key, but only lift the finger.” He
adds that this technique enhances greatly the expression of the melody line and
that sometimes portato is “made with a little retard on the note which one

Figure 2.21 Adam, pictorial explanation of the portato.

Friederich Starke, Wiener Pianoforte-Schule (Vienna: 1819), vol. 1, 13. Der tragende Stoss
(Appoggiato) “ist am wenigsten staccato, und heisst der tragende Stos, weil man hier die Töne gleich-
sam tragen, d.h. jeden Ton mit einem gelinden Drucke angeben muss, man nimmt der Note nur ein
Viertheil ihrer Geltung.”
Johann N. Hummel, A Complete Theoretical and Practical Course of Instructions on the Art of Playing
the Piano Forte (London: Boosey, 1828), 66.
Ignaz Moscheles, Studies for the Pianoforte Op. 70, ed. Ernst Pauer (London: Augener, 1886), 7.
Alberto Randegger, Singing (London: Novello and Ewer, 1880), 175.
Richard Taub, “Glossary,” Playing the Beethoven Piano Sonatas (Portland, Ore.: Amadeus Press,
2002), 250.
Play ing One Hand after the Other 71

expresses thus.”96 Here, the portato is not merely a type of articulation. The pres-
sure achievable on the clavichord appears to be translated into dislocation between
the hands on the piano. What is notable about Adam’s example is that in extended
passages marked portato the effect is of repeated syncopation.
Adam was not alone in giving this explanation for the portato. In his Metodo
per clavicembalo (1811), Francesco Pollini (1762–1846) notes that at the appear-
ance of portato passages in music of a cantabile character, a little delay of the
melody note “contributes not a little to the expression.”97 Fifteen years later,
Pollini’s musical explanation was repeated almost verbatim along with his musical
example in Pietro Lichtenthal’s Dizionario e bibliografia della musica (Milan, 1826).98
Like Adam, Pollini and Lichtenthal’s example (Fig. 2.22 ) shows the effect of
repeated syncopation caused by dislocation.99
The question remains whether Adam, Pollini, and Lichtenthal considered this
dislocated style of playing suitable only where the portato sign was marked.
Elsewhere in his Metodo, Pollini warned that particular care should generally be
taken to play the hands well together and that as a rule, the right hand is always
struck precisely over the corresponding left-hand notes. This, he added, would
“result in a continually equal tempo, and a strong, masterful performance.”100 The
implications are that generally the two hands should be synchronized, but that
occasionally they did not play quite together. Although the circumstances in which
the rule of “togetherness” might be broken are unknown, the warning itself prob-
ably indicates that a dislocated style was sometimes, if not often, to be heard.
Without audible evidence, we cannot in any case be certain what Pollini and others
considered “well” or “precisely” together. Such descriptive terminology has been
used time and time again, but, as we will see, its meaning and significance have
clearly changed.

Louis Adam, Méthode du piano du conservatoire (Paris: Conservatoire impéral de musique,
1804/5), 156: “On ne doit nullement piquer la touché, mais seulement lever le doigt; cette manière de
détacher ajoute beaucoup à l’expression du chant, et se fait quelquefois avec un petit retard de la note
qu’on veut exprimer ainsi.”
Francesco Pollini, Metodo per clavicembalo (Milan, Italy: Ricordi, 1811), 59: “Un piccolo ritardo
della Note segnate in quest’ultima maniera contribuirà non poco all’espressione di una frase cantabile,
como per esempio.”
Pietro Lichtenthal, Dizionario e bibliografia della musica (Milan, Italy: Fontana, 1826), 216:
“Un piccolo ritardo nella Note separate in quest’ultima maniera, contribuirà non poco all’espressione
d’una frase cantabile, como si vede nella Fig. 2.22.”
Lichtenthal, Dizionario, unpaginated.
Pollini, Metodo per clavicembalo, 100: “Metta particolare attenzione, acciò le due mani vadano
bene unite, e le Note da eseguirsi colla mano dritta siano per massima sempre precisamente bat-
tute sopra quelle della sinistra cui appartengono. Una continua eguaglianza di tempo, una robusta
e maestrevole esecuzione saranno il risultato di tale avvertimento.”
72 off the record

Back to the Future

Our journey started with Brée’s description of dislocation, a technique valued and
advocated by Leschetizky. The correlation between that description and earlier
references is striking. According to Brée, particular expressive effect is achieved
by playing a melody note somewhat later than the corresponding note of the
accompaniment. Her reference provides solid contemporary written evidence of
what can be heard on early recordings, and certainly those of Leschetizky. The
recordings therefore validate the practice as general rather than idiosyncratic.
Although Brée was fairly detailed in her description, several important matters
lack clarity. For example, she does not and probably could not describe the myriad
of possible circumstances in which dislocation was applicable. Notably, she does
not mention cases in which the melody note precedes the accompaniment, even
though Merrick confirms Leschetizky’s approval of it. Nor could she indicate what
extremes of delay between the hands remained within the boundaries of good
taste. She used general expressions, leaving their interpretation to the reader who
would acquire experience of such matters by listening to the best artists of the
day. Clearly, the written text could convey only a basic impression of the practice.
In a similar way, Thalberg’s rule mentioned earlier, which relies on verbal descrip-
tions such as “almost imperceptible” to quantify the amount of delay, leaves in
doubt what was intended.
There are also some notable inconsistencies between Brée’s description of dis-
location and her annotated example. For instance, she states that dislocation
between melody and bass notes may occur only at the beginning of a phrase and
usually (but not always) on important notes and strong beats. But her illustration
shows dislocations at the beginning of every bar without exception. And although
the downbeats of each bar might be considered strong but with varying intensity
according to position in the phrase or harmonic importance, Brée makes no dis-
tinction between them. She could have notated, for example, multiple or darker
dotted lines to differentiate the degrees of delay. On the other hand, too exact an
indication might have been felt to hinder individual inspiration. Also missing
are details about the other “important notes” that might be considered worthy
of dislocation. We know from Merrick, at least, that Leschetizky paid special
attention to dissonant harmonies.

Recorded Evidence
Fortunately, we can hear Leschetizky’s approach to the Chopin Nocturne anno-
tated by Brée, which is preserved on a 1906 Welte-Mignon piano roll. He clearly
uses dislocation to give both emphasis and relief to melody notes. In this respect,
the aural evidence corroborates Brée’s description. But the recording also shows
that Leschetizky made dislocations much more frequently than is suggested
Play ing One Hand after the Other 73

Figure 2.23 Chopin Nocturne Op. 27 No. 2, bars 1 to 9, Leschetizky, 1906 piano roll
(Audio Ex. 2.2 ).

by Brée (Fig. 2.23).101 In addition to most downbeats, for example, Leschetizky

employed dislocation at several other moments during bars 1 to 9. Occasionally,
as can be seen in bars 4, 6, and 8, this enhances the expression of consecutive
notes in a poignant melodic sequence. No doubt these are among the “important
notes” to which Brée alludes. And in bar 7, it seems that Leschetizky played the
melody note on the fourth quaver beat slightly before the bass.
Another apparent discrepancy arises in Brée’s advice that dislocations should
be imperceptible, giving the impression that they were hardly to be heard or
noticed. By today’s standards, many of Leschetizky’s dislocations produce fairly
marked and noticeable gaps. Clearly, it would have been very difficult for Brée to
describe such variations. Some skeptics have argued that such noticeable delay
could in some way be the fault of the roll playback. Yet similarly noticeable time
lapses between the left and right hands can also be heard in many acoustic and
electrical recordings. There is no mistake here: such dislocations were simply a
product of the current vogue. We are now slaves to a highly synchronized style of

Note that the annotated examples provided throughout this text are intended as an approxi-
mate indication of what can be heard on the recordings.
74 off the record

playing. What sounds glaring or uncomfortably obvious to us was very probably

normal, imperceptible, or hardly noticeable a century earlier.
Leschetizky was an artist of the highest order, and he created colors by making
dislocations of varying intensity. In the end, Brée’s written text corresponds
approximately with Leschetizky’s own practices but significantly fails to convey
much of what his piano roll reveals. Without audible evidence to qualify Brée’s
description, so many important features of Leschetizky’s dislocations (delay and
frequency) could not be appreciated.
Leschetizky’s own edition of Chopin’s Nocturne Op. 27 No. 2 was published in
1880 by Rahter,102 in a series that boasted of the repertoire being presented
“exactly as played” by Leschetizky and his wife Annette Essipoff “with the greatest
effect” in their concerts.103 As might be expected, there is no indication of disloca-
tion. No doubt this and other expressive nuances were considered to be as normal
as, say, vibrato or portamento in string playing and singing, and left to individual
taste. These would vary from performance to performance according to the play-
er’s mood, the sonority of the piano, and the acoustic of the performance space
among other things. Their notation was impractical and opposed the spirit of
Leschetizky was not alone in making such noticeable and frequent use of
dislocation in Chopin’s Nocturne Op. 27 No. 2. Several pianists from within and
outside his circle also used it, sometimes to a greater extent. In his 1912 recording
of the work for the Victor Company, La Forge makes similar dislocations to those
of his teacher Leschetizky (Fig. 2.24 ; Audio Ex. 2.3 ).104 Powell also makes
similar dislocations to his teacher Leschetizky, evident in his 1929 Duo-Art piano
roll of the work (Fig. 2.25 ; Audio Ex. 2.4 ).105
Several important pianists not associated with Leschetizky also made disloca-
tions in the work. They can be heard in Louis Diémer’s (1843–1919) recording of it
in circa 1903–4 (Fig 2.26 ; Audio Ex. 2.5 ). Diémer was particularly successful
during his studies at the Paris Conservatoire, winning many prestigious prizes in
diverse subjects. He concertized regularly in Paris and other regions, winning over

Frédéric Chopin, “Nocturne Op. 27 No. 2,” Repertoire Leschetizky: 14 ausgewählte Stücke für
Pianoforte, ed. Theodor Leschetizky (Leipzig, Germany: Rahter, 1880).
These prefatory remarks appear in John Field, “Nocturne in B Flat,” Stücke aus dem Repertoire
Essipoff-Leschetizky, ed. Theodor Leschetizky (Hamburg and Leipzig, Germany: Rahter, 1883), 2.
La Forge made several recordings for the Victor Company and was accompanist for singers such as
Marcella Sembrich (1858–1935) and Emma Eames (1865–1952). He studied with Leschetizky for four
years. See Geoffrey Howard, [Sleeve Notes], Pupils of Theodore Leschetizky, Pearl Opal CD 9839 (1988),
unpaginated 5; Wilson Lyle, “La Forge, Frank,” A Dictionary of Pianists, London: Hale, 1985, 161.
Powell studied with Leschetizky from 1902 to 1907, making his Berlin debut in 1908. This was
followed by concerts tours in Europe and eventually America. He is best known for his composition
Rhapsodie Nègre (1919) for piano and orchestra in which he appeared as soloist with the New York
Philharmonic under Damrosch during a European tour in 1920. See Lyle, “Powell, John,” A Dictionary
of Pianists, 223.
Play ing One Hand after the Other 75

the public with his virtuoso style. His interest in historical recitals led him to form
the Société des anciens instruments. Through his teaching at the Paris Conservatoire
from 1887 onward, he influenced many students who went on to successful careers
including Alfred Cortot (1877–1962) and Robert Casadesus (1899–1972).106
Pachmann’s two recordings of the work, the first in 1916 (Fig. 2.27 ; Audio
Ex. 2.6 ) and the second in 1925 (Fig. 2.28 ; Audio Ex. 2.7 ), both exhibit
his use of dislocation. He was among three students who received the silver medal
(the highest award) from the Vienna Conservatory in 1869, and went on to estab-
lish a highly successful, if somewhat controversial, career as one of the foremost
interpreters of Chopin’s music. In a career that spanned almost sixty years, he
achieved celebrity status throughout Europe and America.107
In his 1936 recording of the work, Rosenthal, too, makes dislocations
(Fig. 2.29 ; Audio Ex. 2.8 ). From 1872 to 1874 he studied with Karol Mikuli
(1819–97), Chopin’s pupil and assistant. And from 1875 he settled in Vienna to
study with Rafael Joseffy (1852–1915), a former pupil of both Liszt and Tausig.
After 1884, he made many successful concert tours in Europe and America and
was awarded the title of “Imperial and Royal Court Pianist” by Emperor Franz
Josef in 1912.108
The dislocations of Leschetizky, La Forge, Powell, Diémer, Pachmann (1925
recording) and Rosenthal are presented in Table. 2.3. The numbers listed refer to

Table 2.3 Dislocations Made by Leschetizky La Forge, Powell, Diémer,

Pachmann, and Rosenthal, Chopin Nocturne Op. 27 No. 2, Bars 2 to 9
Bar 2 Bar 3 Bar 4 Bar 5 Bar 6 Bar 7 Bar 8 Bar 9

Leschetizky 1 9, 11 1, 11, 12 1 1, 9, 10, 11, 12 1, 7 1, 11, 12 1

La Forge 1,9 1 12 1 12 1 1, 7, 9, 10, 1, 7
11, 12
Powell 1 1 1, 9, 12 1 9, 11, 12 1 1, 8, 9, 10, 1
11, 12
Diémer 1, 7 1 1 1 1, 9, 10, 11, 12 1
Pachmann 1 1 11, 12 1
Rosenthal 1, 9 9 1 1 1, 12 1, 7 1 1

Elizabeth Bernard and Charles Timbrell, “Louis Diémer,” The New Grove Dictionary of Music and
Musicians, 2nd ed., ed. S. Sadie (London: Macmillan, 2001), vol. 7, 328; see also Lyle, “Diémer, Louis,”
A Dictionary of Pianists, 77–78.
See Mark Mitchell, Vladimir de Pachmann: A Piano Virtuoso’s Life and Art (Bloomington: Indiana
University Press, 2002).
Mark Mitchell and Allan Evans eds., Moriz Rosenthal in Words and Music (Bloomington: Indiana
University Press, 2006), xviii–xix.
76 off the record

Table 2.4 Dislocations Made by Pachmann, Chopin Nocturne Op. 27 No. 2,

Bars 26 to 31
Bar 26 Bar 27 Bar 28 Bar 29 Bar 30 Bar 31

Pachmann (1916) 1 8, 9 11, 12 1, 5, 11, 12

Pachmann (1925) 1 1 1

the semiquaver beats in each bar at which point dislocation can be heard.
Pachmann’s 1916 recording commences at bar 26 of the work, where the opening
material is repeated. For this reason, I have provided numbers for the dislocations
he makes from bars 26 to 31 in both his 1916 and 1925 recordings (Table. 2.4)
because this permits, at least, a certain degree of comparison.
Although these pianists made dislocations in subtly different ways (in terms of
both degree and frequency of delay), the underlying principle was common to all.
Clearly, the practice of dislocation was not an idiosyncrasy of a few, but a general
performing practice. It is highly likely that many nineteenth-century pianists
used it as a matter of course. For instance, Brahms can be heard dislocating melody
from accompaniment quite regularly on his 1889 wax cylinder recording of a frag-
ment from his Hungarian Dance No. 1.109 As Crutchfield confirms, where it is pos-
sible to tell this occurs on practically all “accented first beats where the texture is
melody/accompaniment—never on big accented chords.”110 The Brahms cylinder
provides an inkling about the way in which Brahms actually played. Without it, we
might never have known that he used dislocation. That he did is quite significant
from another point of view. His practice may have impacted on a later generation
of pianists who heard him play or whom he taught and nurtured, and who left a
legacy of recordings. Their playing (discussion to follow) appears to preserve the
remnants of a Brahmsian tradition, one that has all but disappeared.

Truth and Contradiction: Written Texts versus

Audible Evidence
Brée’s advice about dislocation corresponds fairly closely to Leschetizky’s practice
and that of many other pianists. Other written texts, however, provide examples

To sample “denoised” versions of Brahms’s 1889 cylinder visit
groups/edison/brahms/brahms.html. See also Jonathan Berger and Charles Nicols, “Brahms at the
Piano: An Analysis of Data from the Brahms Cylinder,” Leonardo Music Journal vol. 4 (1994): 23–30.
Crutchfield, “Brahms, by Those Who Knew Him,” 14.
Play ing One Hand after the Other 77

in which, perplexingly, advice appears to contradict practice. This is particularly

noticeable when one compares the references to dislocation by both Saint-Saëns
and Pugno with their own recordings. Saint-Saëns enjoyed an unusually long
career both as a formidable concert pianist and a prodigious composer. In his
youth he was compared to Mozart and Mendelssohn. Bülow rated him an “all-
round musician greater than Liszt.”111 His powerful technique was a product of
the Kalkbrenner method taught by his teacher Camille-Marie Stamaty (1811–70).
His pianism, which combined “technical brilliance and emotional reserve,” elicited
a wide range of reaction.112 Saint-Saëns took particular interest in Chopin’s
manner of giving flexibility to melodic material—the “bel canto” tempo rubato.
In an edition of Le Courier musical (1910), he describes this style as told to him by
two of Chopin’s students, the Vicomtesse de Grandval (1830–1907) and the
famous singer Pauline Viardot-Garcia (1821–1910).113 To effect a proper tempo
rubato, “the accompaniment remains undisturbed while the melody floats capri-
ciously, rushes or retards, sooner or later to find again the support of the accom-
paniment.” He acknowledges the difficulty in achieving the requisite “independence
of the two hands.” Nevertheless, he is highly critical of those who, unable to
achieve the effect, delude both themselves and others “by playing the melody in
time and dislocating the accompaniment in order to make it fall at the wrong
time.” Worst of all in his eyes, some simply play “the two hands one after the
other.” He considers it the lesser of two evils “to play everything evenly in time
and the two hands together,” but adds sarcastically that they would then not
appear to carry the “artistic air.”114 Quite obviously, a particular practice, which
shared the characteristics of dislocation, had become a substitute for the true

Harold C. Schonberg, The Great Pianists (London: Gollancz, 1964), 265.
Stephen Studd, Saint-Saëns: A Critical Biography (London: Cygnus Arts, 1999), 59–61.
Saint-Saëns tells us that “she [Pauline Viardot] was a great friend of Chopin and she remembered
his playing almost exactly and could give the most valuable directions about the way he interpreted
his works. I learned from her that the great pianist’s (great musician’s, rather) execution was much
simpler than has been generally supposed. It was as far removed from any manifestation of bad taste
as it was from cold correctness.” See Camille Saint-Saëns, “Pauline Viardot,” École Buissonnière: notes et
souvenirs (Paris: 1913), 222; trans. Edwin G. Rich in Musical Memories (Boston: Small, Maynard, 1919),
151–52: “Grand amie de Chopin, elle avait conservé de son jeu un souvenir très précis et donnait
les plus précieuses indications sur la manière d’interpréter ses ouvres. Par elle, j’ai su que l’exécution
du grand pianiste (du grand musicien plûtot) était beaucoup plus simple qu’on ne se maniérisme de
mauvais gout que d’une froide correction.”
Camille Saint-Saëns, “Quelques mots sur l’exécution des oeuvres de Chopin,” Le Courier
musical vol. 13, no. 10 (1910), 386–67: “Dans le vrai, l’accompagnement reste imperturbable, alors que
la mélodie flotte capricieusement, avance ou retarde, pour retrouver tôt ou tard son support. Ce genre
d’exécution est fort difficile, demandant une indépendance complete des deux mains; et quand on ne
peut y parvenir, on en donne à soi-même et aux autres l’illusion, en jouant la mélodie en mesure et en
disloquant l’accompagnement pour le faire tomber à faux; ou bien encore,—c’est le dernier degré,—
on se contente de faire arriver les deux mains l’une après l’autre. Mieux vaudrait cent fois jouer tout
uniment en mesure et les deux mains ensemble; mais alors on n’aurait pas ‘l’air artiste.’”
78 off the record

tempo rubato. The anomaly here is clear. The practices by which Saint-Saëns’s
could not abide appear to correspond with the types of dislocation heard on early
Elsewhere, Saint-Saëns conveys a similarly disparaging message. He opines
that a general disregard for the true style of tempo rubato often led to the
disfiguring of Chopin’s music. He is adamant that this true style “in no way resem-
bles the dislocations by which it is so often caricatured.”115 Further to this, in a
lecture on early music given in San Francisco in 1915, Saint-Saëns again criticizes
the use of dislocation. He makes reference to the extreme profusion of grace notes
(ornament signs) in French Baroque keyboard music for which an explanation or
realization can generally be found in the composer’s published ornament tables.
And he singles out the suspension, explaining that it “indicates that the right hand
should arrive upon the keys a little after the left,” and concluding “that there was
not then that frightful habit of playing one hand after the other as is often done
nowadays.”116 Reading between the lines, Saint-Saëns seems to have developed
the notion that the French harpsichordists made dislocations only at places where
the suspension sign was notated and therefore that it was used only sparingly
during that era. But it is also clear that it was the habitual, unconscious, and incor-
rect use of dislocation that Saint-Saëns despised in his own era, not necessarily
the device per se.
Given the force of Saint-Saëns’s comments, it is shocking to find him making
frequent dislocations (sometimes on every quaver beat of the bar) in his perfor-
mance of Chopin’s Nocturne Op. 15 No. 2 preserved on a 1905 Welte-Mignon
piano roll (Fig. 2.30). At times, the aural impression is that the melody note is
aligned with the pulse, anticipating the bass note; at others, it is the bass note
that appears to be aligned with the pulse. These types of dislocation can be heard
throughout the first section and its recapitulation.
Eugene D’Albert (1864–1932), a student of Franz Liszt (1811–86), can be
heard making dislocations occasionally on his 1916 recording of the same work
(Audio Ex. 2.10 ). This leads me to wonder whether Liszt, too, made use of
Saint-Saëns practices precisely what he forbids, or so it seems. How can this
glaring inconsistency be explained? It is possible that during the period between
making his 1905 piano roll and his 1915 lecture, he changed his mind about the
use of dislocation. But this seems rather unlikely. Perhaps the practices he was
railing against were of a nature not preserved in recordings. These might include
even more frequent dislocations with much wider gaps between the left and right

Saint-Saëns, “Pauline Viardot,” 151–52: “qui ne ressemble en rien aux dislocations au moyen
desquelles on en donne trop souvent la caricature.”
Ronald Stevenson, “Saint-Saëns’s Views on the Performance of Early Music,” Performance
Practice Review vol. 2, no. 2 (1989): 130.
Play ing One Hand after the Other 79

Figure 2.30 Chopin Nocturne Op. 15 No. 2, bars 1 to 8, Saint-Saëns, 1905 piano roll
(Audio Ex. 2.9 ).

hands. Another explanation might be that when he asked for the two hands to be
played together, he meant almost together or more closely together. Or it might
be that dislocation, which had existed alongside the “true” tempo rubato, had
become too dominant a feature. Although its use was not altogether inappropri-
ate, some pianists’ playing exhibited an imbalance between the two. Whatever the
reason for such an anomaly, the fact remains that in this case, written texts and
audible evidence do not accord.
Indeed, Saint-Saëns’s criticisms may, like those of Polish pianist Jan Kleczyński
(1837–95), have been directed only at certain types of pianists who used various
expressive techniques incorrectly and to detrimental effect. In 1879, Kleczyński
was scathing of the effeminate performances of Chopin’s music by “numbers
of school-girls” who believed they were playing with expression and feeling.
He felt that their performances were “diametrically opposed to the sense and
meaning” of Chopin’s music and led to a diminishing of his reputation. Among
the abuses, Kleczyński included “striking the chords with the left hand just before
80 off the record

the corresponding notes of the melody.”117 It cannot be absolutely certain, how-

ever, that Kleczyński was not simply criticizing the type of dislocation made by
Saint-Saëns, Leschetizky, and others. He may have been one of several who were
actively trying to eradicate dislocation practices from piano playing—a matter
that we will come to later.
Another fascinating anomaly presents itself when one compares the written
advice of the French pianist, teacher, and composer Raoul Pugno with his own
playing. Pugno studied at the Paris Conservatoire from 1866 to 1869 with Chopin’s
eminent pupil Georges Mathias (1826–1910), winning several prestigious prizes.
From 1896, he was a professor at the Conservatoire, establishing an active con-
cert career as virtuoso soloist and chamber music player and forming a celebrated
duo with violinist Eugène Ysaÿe (1858–1931).118 He was recognized as a leading
pianist of the time who excelled in the music of Mozart, Chopin, and Franck.119 In
1910, Pugno published a pedagogical work entitled Les leçons écrites de Raoul
Pugno, translated the following year as The Lessons of Raoul Pugno. In this, he tack-
les many aspects of performance style with particular reference to a few of
Chopin’s piano pieces, including the Nocturne Op. 15 No. 2. Fortunately, this
work is one that he recorded, providing an ideal opportunity for direct compari-
son. Like Saint-Saëns, Pugno gives the impression of being completely opposed to
the use of dislocation, making it abundantly clear that in the opening bars of this
Nocturne, it was not to be tolerated. He explains that the first part of this Nocturne
(Fig. 2.31 ) is imbued with “a mood of peacefulness and resignation,” requiring
“absolute tranquillity” in performance. But he remonstrates vehemently: “I repeat,
and shall repeat again and again: Keep the two hands well together [sic]. To hear the
C sharps and F sharps of each bar in the left hand preceding the note in the right
hand is a thing to make the hair stand on end, and it is wholly anti-musical.”120
According to Pugno, therefore, absolutely no separation between the right and
left hands should occur at the downbeat of each bar. Curiously, however, in his
1903 acoustic recordings of the work, he unabashedly dislocates each downbeat as
well as various other beats in the bars (Fig. 2.32).
Continuing on, Pugno instructs that during bar 6, the listener should be made
to “wish for the F sharp.” To achieve this, he advises, somewhat paradoxically
(remembering what he just said about the opening), that the F sharp may even be

Jan Kleczyński, O wykonywaniu dziel Szopena (Warsaw: 1879); trans. Alfred Whittingham
as How to Play Chopin: The Works of Frederic Chopin and Their Proper Interpretation, 4th ed. (London:
Reeves, 1882), 18–19.
Jerrold N. Moore, [Sleeve Notes], Raoul Pugno: His Complete Published Piano Solos, Pearl, Opal
CD 9836 (1989), unpaginated. 2–3.
Guy Bourligueux, “Raoul Pugno,” The New Grove Dictionary of Music and Musicians, 2nd ed.,
ed. S. Sadie (London: Macmillan, 2001), vol. 20, 592.
Raoul Pugno, Les leçons écrites de Raoul Pugno (Paris: 1910); trans. Ethel Colburn Mayne as
The Lessons of Raoul Pugno (London: Boosey, 1911), 66.
Play ing One Hand after the Other 81

Figure 2.32 Chopin Nocturne Op. 15 No. 2, bars 1 to 8, Pugno, recorded 1903 (Audio
Ex. 2.11 ).

somewhat isolated by playing it after the chord in the left hand, warning however
that this is “an exceptional thing.”121 Clearly, however, this way of isolating a note
was not exceptional and occurs at many other points in Pugno’s rendition. At the
moment in question, Pugno does indeed delay the F sharp in the melody but he
also arpeggiates the left-hand chord—a practice he does not mention. Although
Pugno’s performance of the Nocturne Op. 15 No. 2 is much slower and more
languid than Saint-Saëns’s, the frequency of their dislocations is unmistakably
Pugno also employs dislocation fairly frequently in Chopin’s Berceuse Op. 57
and Chopin’s Valse Op. 34 No. 1, both also recorded in 1903. In the latter, he
instructs the player to give the first note of bar 274 “a sentiment of regret and
remoteness.”122 He achieves this effect very successfully in his performance by
making a single dislocation (Fig. 2.33 ; Audio Ex. 2.12 ), thereby marking the
sudden soft poignant moment. Yet this could never have been imagined from his
verbal advice alone.
The reasons for the glaring discrepancy between Pugno’s written advice regard-
ing the Nocturne Op. 15 No. 2 and the way he played it are unclear. Like Saint-
Saëns, it is possible (though unlikely) that Pugno changed his mind about the use
of dislocation between the time of the recording and the time of writing The
Lessons. The answer may simply lie in the fact that his advice was primarily for

Pugno, The Lessons, 67.
Pugno, The Lessons, 16.
82 off the record

the benefit of students who were not yet skilled enough to render dislocations
artistically. In this case, he may have considered it a lesser evil to hear the hands
played absolutely together than with gaping or inartistic silences. Such comments
underscore a culture of abuse of this technique that was heavily frowned upon.
Perhaps in writing, he was wary of being seen to support the practice. Yet another
possibility might be that, as with Saint-Saëns, there is hidden meaning in Pugno’s
advice to “keep the hands well together.” Today, we take these words to mean
absolute synchrony between the hands; in truth, Pugno may simply have meant
that the hands should not be played so apart as to cause ridiculously wide
gaps. Hence, the expression “well together” may mean “fairly closely” instead of
“absolutely together.”
Pugno’s and Saint-Saëns’s own dislocations were obviously acceptable to them.
Without hearing what it was they found unacceptable, it is difficult to appreciate
the underlying meaning of their advice in this matter. Their words give the impres-
sion that dislocation was to be employed very sparingly, if at all. However, they
use it frequently and in a way that sounds highly exaggerated compared with the
synchronous style of today. Whatever the reasons for such discrepancies, it is an
undeniable fact that a face-value interpretation of their advice would produce an
effect completely divergent from their own playing. Here, written texts fail to
convey what happened in reality.
Other examples also demonstrate the obscurity that sometimes plagues the
true or underlying meaning of written texts. In 1882, the German composer, pia-
nist, pedagogue, and music editor Moritz Moszkowski (1854–1925) commented
on “the difficulty of bringing out clearly with one hand two themes moving inde-
pendently.” With reference to Schumann’s Warum? Op. 12 No. 3, he berated the
use of continual arpeggio playing to circumvent this difficulty and considered
such means to be “one of the most perfunctory styles of which a pianist can be
guilty.” Moszkowski warns that for all intervals apart from the ninth in bar 7,
a tasteful interpretation demands “the simultaneous striking of intervals.”123
But Moszkowski’s warning is intriguing because all but one interval in Warum?
can easily be played together.
Warum? is certainly complex in its voice exchange and syncopated rhythms.
But Schumann often helps to delineate one melody from another, perhaps strate-
gically so, by decorating particular notes with ornaments, thus emphasizing these
in the texture (Fig. 2.34 ).124 An example of this is the second crotchet beat in
the right hand in bar 10. Sometimes he notates an anticipatory grace note at the

Moritz Moszkowski, [Notes] to Robert Schumann, Warum? Op. 12 No. 3 (London: Augener,
1882), 10.
Robert Schumann, “Warum? Op. 12 No. 3,” Klavierwerke. Erste mit Fingersatz und
Vortragsbezeichnung versehene instructive Ausgabe nach den Handschriften u. persönlicher Überlieferung,
ed. Clara Schumann, vol. 2 (Leipzig, Germany: Breitkopf & Härtel, 1879), 86.
Play ing One Hand after the Other 83

point where one voice momentarily finishes and another commences—for example,
in the right hand between bars 8 and 9. Other places—for example, in the right
hand in bars 5 and 6—are naturally delineated because one voice is written in
syncopation with another in the same hand. That leaves very few instances in
which Schumann makes no delineation—for example, at bars 35, 38, and 39
(Fig. 2.35 ).125 Clearly, even if all these moments are arpeggiated, the resulting
effect does not amount to anything approaching continual arpeggiation. Indeed,
the effects are momentary: hardly enough, it seems, to cause Moszkowski’s level
of concern. It is difficult to appreciate exactly what he was complaining about, but
the answer may lie in some manner of playing Warum? that was commonly
Reinecke’s 1905 Welte-Mignon piano roll of Warum? may shed light on the
issue. It reveals, among other things, a manner of dislocation that may have been
the stimulus for Moszkowski’s censure. Reinecke, born while Beethoven and
Schubert were still alive, is probably the oldest pianist to have made piano rolls.
During the 1840s, Reinecke’s skills as a composer and pianist were highly regarded
in Leipzig. He roused Mendelssohn’s interest after playing at the Gewandhaus in
1843, where he remained for three years. Schumann bestowed his esteem upon
Reinecke, saying, “you understand me like few others” (presumably referring to
Schumann’s music). And Liszt admired Reinecke’s “beautiful, soft, legato and
singing touch,” and employed him as piano teacher to his daughter.126 His link
to Leipzig was firmly established around 1860 when he became a professor at
the Leipzig Conservatory and eventually its director in 1897. Many important
musicians—including Grieg, Riemann, and Felix Weingartner (1863–1942)—
studied there during Reinecke’s successful tenure. Reinecke apparently regarded
it as his role to “perpetuate the example of the Classical composers” and to be a
“representative and guardian of tradition.”127 Many saw him as the last important
representative of the Mendelssohn/Schumann era. He was hailed as “an artist of
truly aristocratic and fine feeling” who subordinated his own personality in the
interpretation of a musical work.128
Reinecke, of all players, was most likely to have known the style appropriate to
Schumann’s works. He certainly used arpeggiation to delineate between the mul-
tiple melodies played by one hand, and he also arpeggiated several syncopated
chord figures (examined in chapter 3). But perhaps the most striking element is
his very frequent separation of melody from accompaniment by dislocation of the

Schumann, “Warum? Op. 12 No. 3,” ed. Clara Schumann, 86.
Reinhold Sietz, “Carl Reinecke,” Die Musik in Geschichte und Gegenwart: Allgemeine Enzyklopädie
der Musik, ed. Friedrich Blume (Kassell, Germany: Bärenreiter, 1963), vol. 11, 187–88.
Reinhold Sietz, “Carl Reinecke,” The New Grove Dictionary of Music and Musicians, 2nd ed.,
ed. S. Sadie (London: Macmillan, 2001), vol. 21, 158.
Fritz von Bose, “Carl Reinecke: An Appreciation,” The Musical Times, vol. 51 (May 1910): 302.
84 off the record

Figure 2.36 Schumann Warum? Op. 12 No. 3, bars 1 to 12, Reinecke, 1905 piano roll
(Audio Ex. 2.13 ).

hands, producing an effect of almost continuous arpeggio (Fig. 2.36). The effect is
obvious when compared with the much less frequently dislocated performance of
Warum? by Gabrilowitsch, who recorded the work in 1924 (Audio Ex. 2.14 ).
Taking into account Reinecke’s interpretation, it is likely that for Moszkowski,
the term arpeggiation incorporated within its meaning dislocation of the hands.
It must surely have been this frequency of separation of notes written in strict
vertical alignment against which Moszkowski was railing. Yet could the highly
respected Reinecke have been considered one of the main sinners? Significantly,
Reinecke can also be heard making frequent dislocations in Schumann’s Kreisleriana
Op. 16 No. 6, preserved on a Hupfeld piano roll recorded in circa 1905 (Audio
Ex. 2.15 ).
Reinecke had a predilection for arranging movements from Mozart’s piano
concertos as piano solos. One such arrangement is the second movement—
Larghetto—from Mozart’s Piano Concerto K 537, another work he recorded for
the Welte-Mignon in 1905.129 Here we have an invaluable opportunity to compare
Reinecke’s notation with his performance. One might understandably expect
some similarity between the two, but the extent to which this is not the case
is quite a shock. Reinecke is quite specific, for example, in his placement of
arpeggio signs, giving the impression that arpeggios are to be used only at the
moments indicated. In the opening section of his arrangement (Figs. 2.37 and
2.38), the only apparent separations of notes to be made are between bars 9 and 12,

To view a catalogue of other rolls by Reinecke, visit
Play ing One Hand after the Other 85

Figure 2.37 Mozart Larghetto arr. by Reinecke, bars 1 to 4.

Figure 2.38 Mozart Larghetto arr. by Reinecke, bars 9 to 12.

in the form of arpeggiations of widely spaced and richly textured chords in both
However, Reinecke’s recording of the Larghetto reveals that throughout his
performance, he added arpeggios freely and introduced an astonishing number of
dislocations of the hands (Figs. 2.39 and 2.40).131 Clearly, Reinecke had a very
flexible approach toward the dislocation of melody notes and accompaniment, as
well as the addition of notes to, or the subtraction of notes from, his original nota-
tion (see bars 1, 9, and 12). Sometimes, as exemplified in bars 2, 3, and 4, the
right-hand melody note is delayed until after the left-hand accompanying chord
(played unarpeggiated) is sounded. At other times, as exemplified in bar 1, there
is a significant gap between the right-hand melody note and the final note of the
arpeggiated left-hand accompaniment chord. And in a similar way, the section
from bar 9 to bar 12 preserves examples in which a gap is made between the unar-
peggiated right-hand chord (embellishing the melody) and the final note of the
arpeggiated left-hand accompanying chord. Therefore, in addition to frequent
arpeggiation, Reinecke made around fifteen dislocations of the hands between
bars 1 and 4, and a further nineteen between bars 9 and 12. Notable, too, is

Wolfgang A. Mozart, “Larghetto aus dem Krönungs—Concerte (D dur No. 20) für Pianoforte
solo zum Concertvortrage,” arr. Reinecke (Leipzig, Germany: Breitkopf & Härtel, 1874), 2.
In fact, Reinecke reordered certain sections in his recording of the Larghetto. Thus instead
of playing bars 9–16 where they appear in his arrangement, he inserts bars 17–27 after bars 1–8,
followed by bars 9–16. For the purpose of this discussion however, bars 9–12 refer to the musical
material as it occurs in Reinecke’s arrangement.
86 off the record

Figure 2.39 Mozart Larghetto arr. by Reinecke, bars 1 to 4, Reinecke, 1905 piano roll
recording (Audio Ex. 2.16 ).

Figure 2.40 Mozart Larghetto arr. by Reinecke, bars 9 to 12, Reinecke, 1905 piano roll
recording (Audio Ex. 2.17 ).

Reinecke’s practice of delaying melody notes—particularly when marked, as they

are so often in his arrangement—with portato articulations. He appears to have
had the same regard as Adam, Pollini, and Lichtenthal (cited earlier) for such
notational marks. Such a flexible approach in Reinecke’s playing is also noticeable
in his circa 1905 piano roll recording for Hupfeld of the same arrangement (Audio
Ex. 2.18 ), as well as of his solo piano arrangement of the Adagio from Mozart’s
Piano Concerto in A K 488 (Audio Ex. 2.19 ).
Play ing One Hand after the Other 87

It is really fascinating to compare Reinecke’s performance of Mozart’s Larghetto

with the verbal advice in his book about the revival of Mozart’s piano concertos
published in 1891. Despite his bar-by-bar discussion, one would never envisage
the style that can be heard in his performance. He mentions nothing about dislo-
cation, arpeggiation, or rhythmic flexibility, and he even seems to imply a degree
of tempo regularity that is at odds with his playing (he condemns the conductors
of his day for their obtrusive tempo manipulation).132
The implications here are significant. Most important, Reinecke did not regard
his notation as binding and certainly added much more than was indicated.
Therefore, a strict adherence to Reinecke’s notation would lead to a performance
completely different from his own. Today, such adherence to the score is generally
regarded as the hallmark of successful musical interpretation. But the results are
often a musical blandness compared with, for example, a Reinecke performance.
By early-twenty-first-century standards, Reinecke’s interpretations of both
Mozart’s Larghetto and Schumann’s Warum? are absolutely extraordinary. Yet, in
his era, such interpretations were far more usual. Reinecke was not alone in
employing dislocation in Warum? for expressive purposes. Paderewski made simi-
lar ones in his 1912 recording of the work (Fig. 2.41). At the age of twenty-four,
Paderewski had lessons with Leschetizky, under whose masterful guidance he

Figure 2.41 Schumann Warum? Op. 12 No. 3, bars 1 to 12, Paderewski, recorded 1912
(Audio Ex. 2.20 ).

Carl Reinecke, Zur Wiederbelebung der Mozart’schen Clavier-Concerte. Ein Wort der Anregung
an die clavierspielende Welt von Prof. Dr Carl Reinecke (Leipzig, Germany: Eigenthum und Verlag von
Gebrüder Reinecke [1891]), 25–49.
88 off the record

developed a successful approach to piano playing. He went on to enjoy a highly

successful concert career in Europe and America and was hailed as “an outstand-
ingly imaginative performer.”133 He also enjoyed an active political career, which
included a stint as prime minister of Poland, and became a film celebrity through
his appearance in the 1937 production Moonlight Sonata. For several generations
his name was “a household word.”134
In terms of frequency of dislocation, the similarities between Reinecke’s and
Paderewski’s recordings of Warum? are clear. Both applied dislocation not only to
the most poignant moments in the music but also to almost any part of the bar
where it was possible. The aural effect of such dislocations is one of continual syn-
copation, over and above that which is already notated by Schumann.
Like other examples, comparison between Paderewski’s advice and his practice
reveal apparent inconsistencies. In “The Best Way to Study the Piano” (c. 1895),
he censures the abuses of rhythm and tempo—“frequent ritardandi and long
pauses on single notes”—that lead to “over-sentimentalism.” He is also critical of
a type of exaggerated rubato that is so often used to add “feeling and character”
to Chopin’s music. To remedy this, he advises the pianist “to keep strictly as pos-
sible to the rhythm and the tempo.”135 Like Kleczyński cited earlier, Paderewski
apparently felt that such alterations weakened the music. One gets the strong
impression that he upheld a philosophy of strict adherence to the score.
Further to this, in his essay “Tempo Rubato” (1909), Paderewski advises that
“real knowledge of different styles, a cultured musical taste, and a well-balanced
sense of vivid rhythm” should stave off any abuse. For him, excessive freedom is
“often more pernicious than the severity of the law.”136 Such advice, however, gives
no tangible indication of the boundaries governing such excesses of freedom
around the turn of the twentieth century. The fact that style and taste are so radi-
cally different now makes it difficult, or almost impossible to understand that
which constituted “a cultured musical taste” or “a well-balanced sense of vivid
rhythm” 100 years ago. Indeed, the rhythmic nuances and freedom that result
from Paderewski’s employment of dislocation and other expressive devices are
striking. His playing comes across as thoroughly excessive by modern standards.
Paderewski’s use of dislocation is also particularly noticeable in his playing of
Chopin’s works. In the Nocturne Op. 9 No. 2 recorded in 1930, dislocations occur
frequently throughout each bar (Fig. 2.42 ; Audio Ex. 2.21 ). For us now,
the aural effect is a fairly major distortion of the original rhythms. Paderewski

Jim Samson, “Ignacy Jan Paderewski,” The New Grove Dictionary of Music and Musicians, 2nd ed.,
ed. S. Sadie (London: Macmillan, 2001), vol. 18, 870–73.
Adam Zamoyski, Paderewski (New York: Collins, Atheneum, 1982), v.
Ignacy J. Paderewski, “The Best Way to Study the Piano,” The Strand Musical Magazine (c. 1895),
reprint in The Musical Educator vol. 2 (c. 1900): vii.
Ignacy J. Paderewski, “Tempo Rubato” (1909); reprint in Ronald Stevenson, The Paderewski
Paradox (Lincoln, U.K.: Klavar Music Foundation, 1992), 32.
Play ing One Hand after the Other 89

certainly did not adhere exactly, or even closely, to Chopin’s notation. Paderewski’s
seemingly conservative advice may be attributed to a desire to halt practices he
considered lacking in artistry and skill, practices that are not fully described in
written texts or preserved in aural evidence. Clearly, however, a literal interpreta-
tion of Paderewski’s written advice, according to the current understanding of
words such as strict, slight, distortion, and exaggeration, leads to a significantly
false impression of the way he played.
Pachmann made similar dislocations to Paderewski in his 1915 recording for
Columbia of the Nocturne Op. 9 No. 2 (Fig. 2.43 ; Audio Ex. 2.22 ). And Olga
Samaroff (1880–1948) also made dislocations, though significantly less fre-
quently, in her 1923 recording of the same work (Audio Ex. 2.23 ). Though their
dislocations are subtly different in terms of time lapse between melody and
accompaniment, the similarities between Paderewski’s and Pachmann’s and, to a
lesser extent, Samaroff ’s placement and frequency of dislocation are striking.
Clearly such practices were a norm of the era.
Paderewski also used dislocation prolifically in compositions of earlier compos-
ers such as Haydn, Mozart, and Beethoven. Indeed, during the first movement of
Beethoven’s Moonlight Sonata Op. 27 No. 2 recorded in 1937, he dislocates the
beginning of almost every bar and every change of harmony (Fig. 2.44 ; Audio
Ex. 2.24 ).
In complete contrast to Paderewski, his compatriot and fellow student of
Leschetizky, Ignaz Friedman (1882–1948) makes extremely few dislocations in
his 1926 recording for Columbia of the same work. His playing sounds markedly
synchronized (Audio Ex. 2.25 ).

Changing Tastes
Some written references document a move away from or against the use of expres-
sive devices such as dislocation in certain musical circles. Take, for example, the
Klavierschule (first published in 1868), by the pianist, pedagogue, and piano man-
ufacturer Theodor Steingräber (1830–1904)—alias Gustav Damm. Among the
bad pianistic habits, Steingräber counted “the absurd manner of touching the
keys” when playing with both hands, using two successive motions instead of
one united motion and thereby producing the effect of syncopation. Notably,
the anonymous English translator of the 1879 edition further denigrated the
practice, describing it as one of the “vicious habits” in piano playing.137
Another good example is found in the prefatory notes to Materials to Illustrate
Friedrich Wieck’s Pianoforte-Method (1875) by Alwin Wieck. This provides various

Gustav Damm, Klavierschule und Melodienschtaz für die Jugend (Leipzig, Germany: 1868); 21st
ed., rev. with English trans. (Leipzig, Germany: Steingräber, 1879), vol. 1, 83.
90 off the record

precepts, which, according to Alwin, his father Friederich upheld in the training of
many successful students, among them his daughter Clara Schumann née Wieck.
Alwin explains that “in playing with both hands together, both hands must always
strike the keys precisely at the same moment, and not the left hand before the
right.”138 We do not know whether Clara remained steadfast to such a rule, but, as
we will see later, some of her famous students certainly did not.
Other publications were similarly scathing of the use of dislocation. In the
Magazine of Music: Pictorial Pianoforte Tutor (1891), the anonymous editor
describes dislocation as a “common and very amateurish fault.” Referring to
Fairy Revels by Marian Saunders—a work in popular genre (Fig. 2.45)139—
the editor advises that the quavers in sixths should be played “very smoothly,
with a full soft, singing tone and very exactly with each other.” The editor stresses
that dislocating the first or any other accented notes in the bar “must be
Yet at the same time, many professional pianists were practicing this amateur-
ish fault. By 1900, the move to eradicate the practice of dislocation seems to
have accelerated. According to Ronald Stevenson (b. 1928), Busoni actively
banished from his own playing the improvised style that was so much a feature
of nineteenth-century recitals. This style “smacked too much of the circus” for
him. According to Busoni’s student Egon Petri (1881–1962), “Busoni was the only
pianist in Berlin around 1900 who studiously expunged from his pianism the
gratuitous arpeggio and bass anticipation of melody.” This in itself is highly
significant: the implication is that there were many pianists in Berlin who were
still freely employing both techniques. For Busoni, a more synchronous approach
with the chords “attacked from the scapula” was better suited to his weighty
conception of music by Bach and Beethoven. The result was an austere style

Figure 2.45 Marian Saunders Fairy Revels, bars 1 to 6.

Alwin Wieck, Materialien zu Friederich Wieck’s Pianoforte-Methodik, trans. as Materials to Illustrate
Friedrich Wieck’s Pianoforte-Method (Berlin: Simrock 1875), 3.
Marian Saunders, “Fairy Revels,” Magazine of Music: Pictorial Pianoforte Tutor (1891),
part 2, 213.
Saunders, “Fairy Revels,” 213, italics added.
Play ing One Hand after the Other 91

that had many supporters and is now widely applied to Romantic piano literature.
However, as Stevenson quite rightly explains, this “is a misconception of the
Romantic style.”141
Busoni’s negativity toward dislocation and arpeggiation is evident in the notes
on “Interpretation” in an accompanying volume to his edition of J. S. Bach’s Das
wohltemperierte Klavier (c. 1894). With reference to piano transcriptions of works
originally written for the organ by Bach, he warns that “arpeggios, or the hasty
anticipation of the bass, are of very doubtful taste” because they are contrary to
the character of the organ. In his opinion, they produce the effect of overexertion
and they lack the requisite weight. He blames such practices on the transcriptions
themselves and advises editors to make sure that unidiomatic writing is avoided.142
For instructive purposes, Busoni gave examples of how to space chords success-
fully in order to avoid such problems (Fig. 2.46).143 Despite his best efforts,
Busoni did continue to use dislocation, albeit on far fewer occasions than his
contemporaries. Dislocation can be heard at the beginning of every bar in his
1922 recording of Chopin’s Prelude Op. 28 No.7, which he played twice (Audio
Ex. 2.26 ). He very probably thought this entirely suitable to enhance Chopin’s
marking p dolce.
While Busoni was going through the process of expunging dislocation from his
style, other pianists—some among the oldest generation to record—were already
using it very infrequently. Grieg was one of these. His 1903 recordings raise
many questions about his apparent avoidance of dislocation. Compared with
Reinecke or Leschetizky, Grieg’s playing sounds starkly synchronous. This is
even more interesting when one considers that he studied at the Leipzig Conser-
vatory when Reinecke was its director. Among his teachers were Ernst Ferdinand
Wenzel (1808–80) and Moscheles, who are likely to have used dislocation in their
Unfortunately, Grieg recorded only his own compositions. It would have been
illuminating to hear him play works by earlier Romantic composers like Chopin
and Schumann to which dislocation was most readily applied. It is possible that
in his own music, as well as those of more contemporary composers, he made
dislocations sparingly, but applied them more freely in specific Classical and

Stevenson, The Paderewski Paradox, 13.
Ferrucio Busoni, “Vortrag,” Johann S. Bach, Das wohltemperierte Klavier, erster Teil, bearbe-
itet und erläutert, mit daran anknüpfenden Beispielen und Anweisungen für das Studium der modernen
Klavierspieltechnik von Ferrucio Busoni, vol. 1 (Leipzig, Germany: c. 1894); trans. as “Interpretation,” in
“Corrections and Augmentations to Books I and II” of The Well Tempered Clavichord Revised, Annotated,
and Provided with Parallel Examples and Suggestions for the Study of Modern Pianoforte-Technique by
Ferrucio B. Busoni (New York: Schirmer, 1899), 87.
Busoni, “Interpretation,” 87. Note that Example 91 is not credited, but the better version of
Example 92 is by Liszt.
92 off the record

Figure 2.46 J. S. Bach keyboard works, arr. and annotated by Busoni.

Romantic repertoire. Many other players appear to have differentiated between

the type of repertoire for which it was or was not appropriate (Table 2.2). Without
audible evidence, however, this is impossible to determine.
Grieg can be heard making dislocations in his recordings very occasionally. In
the very nostalgic Remembrances Op. 71 No. 7 he makes a noticeable dislocation
of the melody note F sharp from the accompanying chord on the first beat of bar
25. This was seemingly to mark or enhance the beginning of the new section
marked pp dolce (Fig. 2.47). He also makes dislocations on the second beats in bars
30 and 38. In the Finale of his Sonata Op. 7, Grieg arpeggiates the chord on the

Figure 2.47 Grieg Remembrances Op. 71 No. 7, bars 24 and 25, Grieg, recorded 1903
(Audio Ex. 2.27 ).
Play ing One Hand after the Other 93

Figure 2.48 Grieg Sonata Op. 7, Finale, bars 75 and 76, Grieg, recorded 1903 (Audio
Ex. 2.28 ).

second dotted crotchet beat in bar 75 and then dislocates the left and right at
the beginning of bar 76, probably to mark the change of texture and figuration
(Fig. 2.48). There may well be a few other examples in other works recorded by
Grieg, but unfortunately the sound quality makes it hard for me to be certain.
Based on this, the conclusion is that Grieg made dislocations much less frequently
than many of his contemporaries. The reason for his avoidance of it in some of the
highly expressive and lyrical pieces that he recorded is unclear. However, a clue
may lie in his apparent desire to safeguard against too much artistic license.
Reportedly, he did not like pianists to overinterpret and particularly despised
what he termed the “rubato influenza.” Perhaps unsurprisingly then, nothing
sounds overexaggerated in his playing.144 In this light, Grieg may have found the
current prolific use of dislocation, among other things, undesirable in his music or
perhaps for all repertoires. On the other hand, he was known to be broad-minded
and generous when it came to artistic matters.145 Grainger recalled that Grieg was
very flexible even with the performance of his own works. In 1906, Grieg accused
Grainger of not playing his folk song according to Grieg’s intentions. Despite this,
Grieg told Grainger not to alter his performance in any way, claiming that he loved
Given this apparent flexibility, Grieg would probably have approved of the
playing of Landon Ronald, who made the earliest acoustic recording of his Bridal
Procession Op. 19 No. 2 in 1900. Ronald plays the first part in a synchronized
manner. However, in the very soft section commencing at bar 40 he broadens the
tempo considerably and makes expressive dislocations at the beginning of bars 44
and 45, where the dynamic is marked ppp (Fig. 2.49). Notably, Grieg does not
make any dislocations at this point in either his acoustic recording or his two
1906 Welte-Mignon piano rolls of the work. Ronald also makes dislocations in his

Per Dahl, “Contemporary Evaluations of Grieg as a Pianist,” Edvard Grieg: The Piano Music in
Historic Interpretations, trans. William H. Halverson, SIMAX PSC 1809 (1992), 61.
Percy Grainger, “Personal Recollections of Grieg,” The Musical Times vol. 48, no. 777 (1907): 720.
John Bird, “Grainger’s Anecdotes,” Percy Grainger (Oxford: Oxford University Press, 1999), 134.
94 off the record

Figure 2.49 Grieg Bridal Procession Op. 19 No. 2, bars 43 to 45, Ronald, recorded 1900
(Audio Ex. 2.29 ).

1900 recording of Grieg’s Dance Caprice Op. 28 No. 3 (Audio Ex. 2.30 ). These
enhance the characteristic dance rhythm effects. Dislocations are also heard in his
1900 G & T recording of a fragment of Chopin’s Polonaise Op. 40 No. 1. Ronald’s use
of dislocation is not reserved for solo works: they can be heard to very expressive
effect in the piano accompaniments of songs in recordings made in 1905 with Patti
for the Gramophone Company. His dislocations in the introduction to Mozart’s
Voi che sapete from Le nozze di Figaro are a good case in point (Audio Ex. 2.31 ).
Quite clearly, dislocation was so much part of Ronald’s style that he used it
wherever he considered it appropriate, regardless of composer or repertoire.
There is little evidence to suggest that Grieg would have disapproved of such
practices, providing they were sparingly applied. On the other hand, what Grieg
thought of the dislocation practices of pianists such as Reinecke, Leschetizky, and
others cited earlier, with whom he must have had a certain amount of contact, can
only be imagined.
In Grieg’s case, the recorded evidence is not wide-ranging enough to reach firm
conclusions. He may, like other pianists, have been selective in his employment
of dislocation, saving it for certain repertoire. This type of judicious use is
also exemplified in recordings made by Grünfeld. He enhances the beauty of
Schumann’s Träumerei Op. 15 No. 7 recorded in 1913, with many dislocations
(Audio Ex. 2.32 ). He also makes them in his 1909 recording of the arrange-
ment of Wagner’s Liebestod. And in Chopin’s Nocturne Op. 32 No. 2 recorded in
1911, dislocation is made with the frequency exhibited in some of Paderewski’s
and Pachmann’s recordings. Yet in repertoire by Bach and the faster movements
of other composers, he makes none at all (Table 2.2).
Planté makes dislocations only at a few moments in his 1928 recordings. In
Chopin’s Etude Op. 25 No. 1, dislocations mark the points of climax. In Chopin’s
Etude Op. 25 No. 2, he makes a dislocation on the first note of the piece, heighten-
ing its expression and giving poise to the opening of the movement (Audio
Ex. 2.33 ). He also makes dislocations in Brahms’s arrangement of Gluck’s
Gavotte. But in other repertoire, particularly of a faster and more rhythmically
active nature, dislocation is noticeably absent (Table 2.2).
Play ing One Hand after the Other 95

Figure 2.50a Rachmaninoff Prelude in C sharp minor Op. 3 No. 2, bars 1 to 3.

In the early years of the twentieth century, some pianists tried to eradicate
completely the practice of dislocation. In 1909, Hofmann referred to it scathingly
as “limping” and firmly advised against its use branding it as “the worst habit” in
piano playing.147 And in 1922, Hambourg denigrated the practice, calling it
“another blunder” and likening it to drawling in speech or stuttering. Providing
the opening sequence (octaves in the left hand, single note in the right hand) of
Rachmaninoff ’s Prelude in C sharp minor Op. 3 No. 2 (Fig. 2.50a), as well as an
annotated version (Fig. 2.50b), he bemoaned the “loss of simultaneousness of
sound” resulting in a corresponding loss of grandeur.148 One is reminded of
Busoni’s sentiments in this regard. Significantly, there is a direct correlation
between the notated dislocations for the passages in bars 1 and 2 in Hambourg’s
annotated example, and Leschetizky’s dislocation of similar figures in bars 1, 3,
and 5 in Mozart’s Fantasia K 475 (Fig. 2.51) preserved on a 1905 Welte-Mignon

Figure 2.50b Rachmaninoff Prelude in C sharp minor Op. 3 No. 2, bars 1 to 3,

annotated by Hambourg.

Joseph Hofmann, Piano Questions Answered (facs. reprint of 1920 ed., New York: Dover, 1976),
Mark Hambourg, How to Become a Pianist (London: Pearson, 1922), 57.
96 off the record

Figure 2.51 Mozart Fantasia K 475, bar 1, Leschetizky, 1906 piano roll (Audio
Ex. 2.34 ).

piano roll. Indeed, Leschetizky makes dislocations throughout his performance.

Hambourg seems, at least in verbal advice, to have been directly opposed to the
style in which his teacher Leschetizky excelled. Carl Reinecke’s circa 1905 piano
roll recording for Hupfeld of the same work reveals that he, too, employed disloca-
tion in the opening sequence and throughout the performance (Audio Ex. 2.35 ).
In 1930, Gieseking labeled dislocation as a “grievous offence against all musical
feeling,” which was often committed by famous concert pianists. He emphasizes
that for the sake of successful expression, “both hands must strike the keys pre-
cisely at the same moment.”149 But he acknowledges that this may not be easy to
achieve. It stands to reason that some pianists would not have found it easy to
strike the keys simultaneously because this went against the grain, particularly
in highly expressive music. The difficulty may be more easily understood by con-
sidering current practices. Pianists today find it abnormal to make separations
between the hands because they are accustomed to an entirely different expressive
practice that is based on absolute synchrony.
Clearly, Hofmann, Hambourg, and Gieseking actively tried to stem the practice
of dislocation, and their own playing can be seen to match their verbal advice. Yet
force of habit remained strong. Despite their advice, they still occasionally
employed dislocation. In his 1912 recording of Schumann’s Warum? Hofmann
makes a dislocation in bar 6 that enhances the expression of the interval between
C flat and F. Noticeable are his dislocations in bars 1 and 5 of Chopin’s Valse
Op. 64 No. 2 recorded in 1916 (Audio Ex. 2.36 ). Hambourg makes prolific use
of dislocation in his 1921 recording of Chopin’s Nocturne Op. 55 No. 1 and
his 1927 recording of Chopin’s Nocturne Op. 9 No. 2. And during bars 1 to 12 of
the second movement, Largo, from Beethoven’s Piano Concerto No. 3 Op. 37
recorded in 1929, Hambourg’s use of dislocation is particularly noticeable
(Audio Ex. 2.37 ). Like other cases discussed earlier, it is impossible to know
why there is such a discrepancy between his verbal advice and his practice.

Walter Gieseking and Carl Leimer, Modernes Klavierspiel nach Leimer-Gieseking (1930); trans. as
The Shortest Way to Pianistic Perfection (1932); repr. in Piano Technique (New York: Dover 1972), 56.
Play ing One Hand after the Other 97

Gieseking also occasionally uses dislocation, though not in his circa 1939 record-
ings of Beethoven’s Andante from Sonata Op. 109 or Brahms’s Intermezzo
Op. 117 No. 2, works in which it might be expected. But he could not resist it in
the opening section of Mendelssohn’s Andante and Rondo Capriccioso recorded live
in 1956. Here, dislocation enhances the beauty and poignancy of the music, giving
an importance to certain melody notes (Audio Ex. 2.38 ).
The more sparing use of dislocation evident in the playing of Hofmann,
Hambourg, and Gieseking is not entirely representative of early-twentieth-
century style: despite the strength of their warning, their influence was not widely
felt until the second half of the twentieth century when the practice all but disap-
peared. The legacy of recordings of several other pianists shows clearly that it
survived, in some cases, fairly healthily until the 1950s. The legendary pianist
Cortot was still using dislocation in his 1934 recordings of Chopin’s Preludes Nos.
7, 13, 15, and 17 Op. 28 and his 1949 recording of Chopin’s Berceuse Op. 57. And
Vladimir Horowitz (1903–89) can be heard making dislocations to poignant effect
in his 1935 performance of the first movement of Brahms’s Piano Concerto in
D Minor Op. 15 with the New York Philharmonic conducted by Toscanini.150
In his The Well-Tempered Accompanist (1949), the Dutch accompanist Coenraad
Valentyn Bos (1875–1955) was critical of “the unforgivable musical sin of antici-
pating the right hand with the left” which he calls a “faulty mannerism.” He
admits, however, to having used the device in his youth. As Crutchfield notes, Bos
seems readily to have accepted the change in style.151 Yet other pianists continued
to use it beyond the mid-century. These include Ilona Eibenschütz, Adelina de
Lara, and Fanny Davies, all of whom studied with Clara Schumann (1819–96) and
Brahms, as well as Carl Friedberg (1872–1955) and Etelka Freund, both of whom
were admired by Brahms, Freund particularly so.152
Eibenschütz, the pianist entrusted with the premieres of many of Brahms’s
piano pieces, makes dislocations in his Waltz Op. 39 No. 2 recorded in 1903, and
at the beginning of the Intermezzo Op. 119 No. 2 recorded in a live broadcast in
1952 (Audio Ex. 2.39 ). Though the taste for such practices had already changed
during the fifty-year lapse between her recordings, Eibenschütz did not appar-
ently change her ways. This is also evident in comparing her 1903 recording of
Brahms’s Waltz Op. 39 No. 15 (Audio Ex. 2.40 ) with her 1962 recording of the
same work made at her home (Audio Ex. 2.41 ).
De Lara employs dislocation at the beginning of Brahms’s Rhapsody Op. 79
No. 2 (Audio Ex. 2.42 ) and throughout his Intermezzo Op. 117 No. 1 (Audio
Ex. 2.43 ), as well as Schumann’s Fantasiestücke Op. 12 No. 1 and No. 2,
Kinderszenen Op. 15 No. 1, and Arabeske Op. 18, all recorded in 1951. Although

This can be heard on YouTube:
Crutchfield, “Brahms, by Those Who Knew Him,” 20.
For a fuller discussion, see Crutchfield, “Brahms, by Those Who Knew Him.”
98 off the record

she was very old when the recordings were made and there are no earlier examples
with which to compare her playing, we can assume that she continued playing in
much the same style as her younger days. Of particular interest is her 1951 record-
ing of Schumann’s Warum? Op. 12 No. 3, which exhibits some similar traits of
dislocation to the discredited style of Reinecke and Paderewski (Audio Ex. 2.44 ).
Davies’s use of expressive dislocation is particularly evident in her 1929 record-
ing of the first movement—“Von fremden Ländern und Menschen”—from
Schumann’s Kinderszenen Op. 15 (Audio Ex. 2.45 ). And she certainly used it
to great effect in the opening sequence of the first movement of Schumann’s
Piano Concerto Op. 54 with the Royal Philharmonic Society Orchestra conducted
by Ernest Ansermet, recorded in 1928 (Audio Ex. 2.46 ).
Friedberg made dislocations in his 1953 recordings for Zodiac of Schumann’s
Kinderszenen Op. 15. In No. 1—“Von fremden Ländern und Menschen”—
dislocation is fairly frequent between bars 1 and 8. In No. 3—“Haschemann”—
Schumann’s marking p for the left hand against sfp for the right hand is brought
out very effectively by making a dislocation. And in No. 7—“Träumerei”—
dislocation is heard in bars 1, 2, and 6. Interestingly, Friedberg shared with
Eibenschütz a particular way of interpreting Brahms’s notation in his Ballade
Op. 118 No. 3. Crutchfield gives a great account of this:

Twice in the “B” section . . . the upbeat is written as F sharp—the same

F sharp—in both hands. It comes where the harmony dictates a ral-
lentando (of course none is marked, but it was clear to the players), and
both Eibenschütz and Friedberg play first an accompanimental F sharp
with the left hand and then a fuller-toned melodic one with the right.
In every modern recording I’ve sampled, this doubling is treated as an
unplayable curio of notation.153

Also interesting is Friedberg’s treatment of Chopin’s Nocturne Op. 15 No. 2 in

a private recording of the work (date unknown). He uses dislocation almost as
much, if not in equal amounts, as Saint-Saëns and Pugno (Audio Ex. 2.47 ).
Freund’s recordings provide a plethora of examples of dislocation. She makes
them throughout her 1953 recording of Brahms’s Sonata Op. 5. In the first move-
ment, dislocations greatly enhance the beauty and poignancy of the lyrical chords
in the right hand at the beginning of bars 7, 8, and 12. And she makes continuous
dislocations to beautiful and extraordinary effect in the opening of the second
movement, Andante espressivo, of the same Sonata, greatly enhancing the feeling
of espressivo (Audio Ex. 2.48 ). Interestingly, Freund studied with Busoni in
1900—as attested in letters from him to her brother, Robert Freund (1852–
1936)—when Busoni was actively expunging dislocation and arpeggiation from

Crutchfield, “Brahms, by Those Who Knew Him,” 18.
Play ing One Hand after the Other 99

his own playing. So it is somewhat surprising that Freund’s playing retains so

much of this style. Evidently, she preferred the type of expression possible with the
use of dislocation, despite the practice of others around her. From 1910 to 1936,
Freund stopped giving concerts in order to raise a family, but during the time lapse
she did not apparently change her mind and adopt the newer synchronized style.
Dislocation can also be heard to very expressive effect in her 1950 recording of
Brahms’s Capriccio Op. 76 No. 1,154 as well as Brahms’s Intermezzo Op. 116 No. 2
and Intermezzo Op. 117 No. 2, both recorded in 1953. As Crutchfield enthuses,
“It becomes quite difficult to imagine that Brahms did any otherwise himself.”155
She also uses it in the opening of J. S. Bach’s Prelude in E flat minor BWV 853 from
the Well-Tempered Clavier Book 1 recorded in 1957.156 There are also many pieces
in which she does not make dislocation. These include movements that are fast or
require a more synchronized style because of their character.
As late as 1961, another student of Leschetizky, Moiseiwitsch, can be heard
occasionally playing the right hand very slightly before the left—a Leschetizky
technique—in his recording of the third movement from Chopin’s Sonata Op. 58.
But he also makes dislocations in the reverse order (Audio Ex. 2.49 ).
Contrary to current notions, the practice of dislocation heard in early piano
recordings was not simply an idiosyncrasy or the habit of a few players but a gen-
eral performing practice that can be traced back to an earlier era, and which con-
tinued for a significant time during the twentieth century. Dislocation was not, as
some have suggested, a special characteristic of the early twentieth century. It had
already been a characteristic—not necessarily special—for a long period before
the recording era. Considering their age, Reinecke, Brahms, Leschetizky, Saint-
Saëns, and others must be considered, of all those who recorded, true representa-
tives of pianism during the second half of the nineteenth century. Their recordings
in conjunction with those of a younger generation show that the practice of dislo-
cation was part of an ongoing nineteenth-century tradition. As Philip notes, the
frequency of dislocation preserved in Leschetizky’s piano rolls corresponds closely
with that of his student Paderewski at his most extreme. I find this incredibly
thought-provoking, because if Paderewski used dislocation as much as his teacher
Leschetizky, then it is very likely that Leschetizky’s teacher (Czerny) and his
teacher’s teacher (Beethoven), used it just as much, or perhaps even more.
Dislocation has sometimes been classed as the illegitimate offspring of tempo
rubato—a manner in which the rhythm of melody notes is played more flexibly
than notated, against a steady accompaniment.157 But despite the fact that both

Published for the first time by Pearl in 1996.
Crutchfield, “Brahms, by Those Who Knew Him,” 18.
Published for the first time by Pearl in 1996.
See Rosenblum, “The Uses of Rubato in Music, Eighteenth to Twentieth Centuries,” 52; and
Charles Rosen, The Romantic Generation (London: HarperCollins, 1996), 413.
100 off the record

share the common feature of separation between notes of the melody and accom-
paniment, their function can be seen to be quite separate. Dislocation was usually
a localized event, applied not necessarily to bend the rhythm of the melody, but to
give to a particular melody note or notes dynamic emphasis or de-emphasis. As
Crutchfield explains, it “made possible many varieties of nuance. Melodic voices
could be (indeed, almost always were) set in relief without the slightly harder
touch needed to bring them out over simultaneously struck accompaniment.”158
As we will see in chapter 4, tempo rubato produced noticeable bending of the
rhythm of the melody, often in an extended fashion.
Although some written texts from around the turn of the twentieth century
validate the use of dislocation, many fail to mention it at all. Others are disparag-
ing and call for its studious avoidance. However, as early recordings show, the
effect of what is promoted in these texts was not fully felt until the second half of
the twentieth century.
For me, dislocation is a thoroughly effective way of giving expression to indi-
vidual melody notes in a range of repertoire spanning several centuries up until
the early twentieth century. I am so seduced by the beauty of expression that
results from its application that in its absence, the music often sounds strangely
naked and/or straightjacketed. In a recent live performance, dislocation helped
me achieve, on the fortepiano, the singing quality I wanted for the opening
sequence of the second movement of Mozart’s Piano Quartet in G Minor K478
(Audio Ex. 2.50 ).159 In another recent live performance, I used dislocation to
give poignant expression to the high notes of the piano part in bars 23 and 24 of
the first movement of Brahms’s Sonata Op. 78 for violin and piano (Audio
Ex. 2.51 ).160 I also used it in many other places throughout the Sonata, when-
ever the urge took me. When it comes to Brahms, Crutchfield’s humorous but sage
words spring immediately to my mind: “Playing Brahms with the hands together
is just like playing Bach with lots of pedal and octave doublings.” He made this
comment as an illustration of how “the concepts and principles underlying the
Baroque revival”—in other words, the HIP movement—might alter the way we
play later music.161
Although many will continue to frown upon the use of dislocation and other
expressive devices, one thing is certain: if we are to be more faithful to the sound
world and intentions of many pre-twentieth-century composers, then we must
consider using these in performances of their keyboard compositions.

Crutchfield, “Brahms, by Those Who Knew Him,” 18.
Live performance with Ironwood, recorded on November 7, 2010, in Verbrugghen Hall, Sydney
Conservatorium of Music.
Live performance with violinist Robin Wilson, recorded October 19, 2010, in Recital Hall East,
Sydney Conservatorium of Music.
Crutchfield, “Brahms, by Those Who Knew Him,” 60.

Unnotated Arpeggiation

On one occasion we were rehearsing for a “Pop” concert . . . Lady

Hallé, Leonard Borwick and myself were preparing a Brahms’ Trio,
Sir Charles [Hallé] kindly turning over for Borwick. After playing
the slow movement Sir Charles said, “Mr. Borwick, do you mind if
I say something?” and of course Borwick said “Certainly not, Sir
Charles.” “Well,” he said, “there is a very prevalent habit among
pianists of ‘spreading’ the notes of a chord with the idea of giving
expression to a passage. It is a habit much to be deplored and
should be discouraged.” When he had quite finished his little
“lecture,” Lady Hallé said, “Yes dear, but you do it!”
—William E. Whitehouse (1859–1935)1

At the beginning of the twentieth century arpeggiation was as intrinsic to piano

playing as dislocation. Early recordings show that many pianists frequently made
unnotated arpeggiations, playing the notes of chords separately where not indi-
cated in the musical text. These arpeggiations cause the separation of vertically
aligned material comprising two or more notes, variously described as double
notes, octaves, and chords. The speed of these arpeggiations varies according to
function, mood, and context, creating different effects. Early recordings also show
that certain pianists made unnotated arpeggiations far less frequently, or not at
all. This more synchronized style of playing, however, did not become the rule
until the second half of the twentieth century.
Unnotated arpeggiation seems to have been most appropriate in slower expres-
sive movements of Classical and Romantic repertoire, less so in late-nineteenth-
century and contemporary repertoire or in music requiring a clean attack and
rhythmic incisiveness. On early recordings, pianists arpeggiate chords in either
hand, or in both together. And most commonly, the notes are played from the
lowest to the highest note. In some cases, the notes in both hands are spread
simultaneously; in others, the spread commences with the lowest note in the left
hand and proceeds continuously to the highest note in the right hand. The aural

William E. Whitehouse, Recollections of a Violoncellist (London: Strad Office, 1930), 27.


Peres da Costa-Ch03.indd 101 3/14/2012 6:55:03 PM

102 off the record

effect is that sometimes the highest note in the chord (the melody note) is aligned
with the pulse; the accompanying note or notes anticipate it. At other times, the
lowest note in the chord is aligned with the pulse, delaying the arrival of the high-
est note. When arpeggiation occurs in conjunction with dislocation of the hands
or tempo modification, it is not always easy to discern where any of the notes lie
in relation to the pulse. In these cases, the practice contributes to a sense of
ambiguity, softening the edges of the rhythm and texture. Arpeggiation differs
from dislocation in that separation of the hands—playing one hand after the
other—is not the underlying principle. And as the historical documentation pre-
sented shortly reveals, the practice of arpeggiation can be seen to stem from a
different root.
Unnotated arpeggiation is preserved on recordings until at least the 1950s. It
can be heard in the playing of the oldest generation to record, including Reinecke,
Leschetizky, and Saint-Saëns, as well the younger generation of Pachmann, Pugno,
Paderewski, Rosenthal, and many others. Generally, those who used dislocation
also made use of unnotated arpeggiation.2 In early piano recordings pianists
arpeggiated chords to achieve the following:

• Emphasize melody notes by delaying and setting them apart from the
harmonic accompaniment
• Provide a cushion of sound supporting the melody note
• Enhance the effect of poignant harmonies by strengthening or softening
• Give particular effect to special accents such as sforzando
• Enliven the momentum of the music, propelling it forward
• Enrich the sound and or texture of the musical material
• Delineate the boundaries of phrases
• Give separation to overlapping melody lines played in one hand

In spite of its widespread use around the turn of the twentieth century, detailed
contemporaneous written advice about unnotated arpeggiation is somewhat
scant. Many pedagogical texts fail to discuss it all. Some consider it indispensable
but describe it only in general terms or very briefly. Others advise its extremely
judicious employment or absolute avoidance, branding it as a perfunctory device
resulting in oversentimentality. Nevertheless, such warnings did not prevent
many leading pianists from continuing to use it well into the twentieth century.
Here, as with dislocation, there is a significant gulf between written advice and
Nowadays, unnotated arpeggiation is generally limited to performances on
plucked or tangent-struck keyboard instruments such as harpsichords, virginals,

See Table 2.2, pages 47–50.
Unnotated A r peg g iation 103

spinets, and clavichords. It is also being used more and more in early piano (forte-
piano) playing. On these instruments arpeggiation is accepted as a historically
appropriate practice. It helps to fill out and sustain the sound, as well as to enhance
color, texture, dynamic, and accent in a range of repertoire and particularly in
figured bass realizations. Its use is infrequently indicated by the composer but
may be implied in the character of the composition or (when they exist) the words.
These and other considerations, including the resonance and tone quality of the
instrument and the acoustic and size of the performing space, influence arpeggio
speed, length, and shape.
Unnotated arpeggiation has been used much less frequently, or not at all, in
live and recorded modern piano playing during the past half century. It has become
discredited to the extent that most modern pianists are extremely hesitant about,
or would hardly ever consider, arpeggiating unless it is expressly indicated in the
score. It is in the context of this present-day obsession with synchronized chord
playing that the abundant chord spreading on some early recordings sounds
foreign and wayward.
In recent times the use of unnotated arpeggiation has occasionally met with
fierce objection. This is best exemplified in the criticisms received by Melvyn Tan
and Lars Vogt for arpeggiating the first chord of Beethoven’s Piano Concerto Op.
58, cited in chapter 2.3 Fairly blatant opposition to its use is found in Rosen’s
popular text The Romantic Generation (1995). Rosen explains that, according to
earwitnesses, Brahms “arpeggiated most chords when he played.” However, in the
same breath and for reasons unstated, he swiftly denounces the evidence as a
guide to performing Brahms’s works.4 Surely, though, it is reasonable to accept
unnotated arpeggiation as a legitimate Brahmsian performing practice if Brahms
himself made use of it?

Written Sources and Historical Precedents

In 1902, Brée—whose description of Leschetizky’s dislocation practice was noted
in chapter 2—documented his approval of unnotated arpeggiation citing a few of
presumably many instances in which it was suitable. In addition to chords that are
too wide for the notes to be played together, arpeggios could be used to create “a
tender or delicate effect.” Giving an example with chords in both hands (Fig. 3.1),
she explains that to achieve this “the right hand plays arpeggio, while the left
strikes its chord flat.”
Arpeggiation could also help in making a chordal texture sound energetic yet
not harsh. This is achieved “when the right hand strikes its tones simultaneously

See chapter 2, pages 42–43.
Rosen, The Romantic Generation, 413.
104 off the record

Figure 3.1 Brée, arpeggiation to create a tender or delicate effect, excerpt from
Paderewski Légende.

Figure 3.2 Brée, arpeggiation to create an energetic effect without harshness, excerpt
from Chopin Scherzo.

and the left arpeggiates.” In this case (Fig. 3.2), Brée adds that the arpeggio must
be very swift. Arpeggiation could also be used to delineate between voices in a
polyphonic texture (Fig. 3.3) or in a canon (Fig. 3.4). In these cases, Brée reserved
its use for important moments when, for example, “one part ends and the other
begins at the same time.”5 Elsewhere in The Groundwork, Brée recommends the
use of arpeggio to create a special type of emphatic nuance (Fig. 3.5): “The octave
marked * is arpeggio’d, and so played that the lower bass tone exactly coincides
with the first beat, while the upper bass tone is struck together with the right-
hand chord, producing an extremely slight retardation.”6 Notably, this arpeggia-
tion is not in the original text of Schumann’s Grillen Op. 12 No. 4. Here, the
“extremely slight retardation” of musical material in the right hand was clearly
intended to enhance the effect of sforzando.
Though useful, Brée’s advice leaves a few important factors unclarified. The
position of the chords (when one or other hand arpeggiates) compared with the
beat is unspecified. Obviously, it is possible to align either hand with the beat
(as in the example from Schumann’s Grillen) or to make arpeggiations across it.

Brée, The Groundwork, 72–73.
Brée, The Groundwork, 70.
Unnotated A r peg g iation 105

Figure 3.3 Brée, arpeggiation to delineate voices in a polyphonic texture, excerpt from
Schumann Romanze.

Figure 3.4 Brée, arpeggiation to delineate voices in a canon, excerpt from Paderewski
Thème varié.

Figure 3.5 Brée, arpeggiation to create emphasis, excerpt from Schumann Grillen.

Also unspecified is the pattern of distribution of notes: that is, lowest to highest
or some alternative. It is also unclear whether Brée intended arpeggiations in all
situations to be very swift (as per her advice for Fig. 3.2).
Like Brée, Merrick affirmed the positive attitude (at least in Leschetizky’s
circle) toward the practice of unnotated arpeggiation. Leschetizky advised Merrick
that sometimes chords in one hand should be spread out, or that the interval of
the seventh should be broken in order to intensify the expression.7

Merrick, “Memories of Leschetizky,” 336.
106 off the record

The So-Called Modern Manner of Arpeggiating in the

Second Half of the Nineteenth Century
Written texts document the widespread use of unnotated arpeggiation during the
second half of the nineteenth century. In their “Instructive Edition” (1891) of
Beethoven’s Piano Sonata Op. 27 No. 2, for example, Immanuel von Faisst and
Sigmund Lebert sternly warn “against applying universally to this piece the modern
manner of arpeggiating resulting in delay of the melody notes throughout.” But
whether this rejected manner of continuously arpeggiating was indeed modern is
questionable. Evidence that follows reveals that, whereas each generation of pia-
nists thought of continuous arpeggiation as a modern practice (perhaps the result
of limited historical perspective), it was really nothing new and was practiced liber-
ally in preceding eras. Faisst and Lebert were not opposed to arpeggiation when
applied judiciously. For the last beat of bar 8 and elsewhere in the first movement,
they notate a vertical wavy line, recommending that a “quick arpeggio be played,
even by those whose hands can otherwise reach all the notes.” In such places, the
chords produce unusual harmonies and so their arpeggiation “enables the melody
notes to emerge more clearly from those of the accompaniment.”8
Other late-nineteenth-century texts point to the liberal use of arpeggiation.
The Viennese-trained pianist and pedagogue Ernst Pauer (1826–1905) considered
the main requisite for effective chord playing to be “the possession of sufficient
and equal strength in all the fingers,” something that in his opinion was a rarity in
the playing of students. In The Art of Pianoforte Playing (1877), he too was critical
of a so-called modern tendency in which “the broken or arpeggio manner
has become so generally diffused, that some performers seem to consider firm
chords altogether obsolete.” For him, chords played firmly express “determina-
tion, strength, and earnestness,” whereas arpeggiated chords express “softness,
langour, despondency, and irresolution.”9 Elsewhere, Pauer vehemently prohibits

Ludwig van Beethoven, “Sonata Op. 27 No. 2,” Sonatas and Other Works for the Pianoforte,
Instructive Edition of Classical Pianoforte Works, ed. Immanuel von Faist and Sigmund Lebert (Stuttgart,
Germany: Cotta, 1891), 60: “So dringend wir davor warnen müssen, auf dieses Stück durchgängig die
moderne Manier harpeggirenden Nachschlagens der Melodietöne anzuwenden, so empfehlen wir
doch an dieser und den wenigen anderen Stellen, welche wir mit [vertical wavy line] bezeichnet haben,
auch für solche Hände, denen die Griffe nicht zu weit sind, ein raches Harpeggieren, um bei der
Eigentümlichkeit der betreffenden Zusammenklänge den Melodieton klarer von dem Begleitungston
Pauer, The Art of Pianoforte Playing (London: Novelleo, Ewer and Co.), 46. Further to this Pauer
explained, “The one may be likened to the man, the other to the women, in Milton’s great epic: ‘For con-
templation he, and valour formed; / For softness she, and sweet attractive grace.’” Pauer was directly
linked with the famous Viennese piano firm Streicher. He studied piano with W. A. Mozart’s son
F. X. W. Mozart. In 1851 he moved to London, where his piano playing was much admired, and eventu-
ally became a professor of piano at the newly established Royal College of Music.
Unnotated A r peg g iation 107

the use of arpeggios unless expressly notated in the text, adding that one of the
most frequent faults is “playing chords in the arpeggiando manner where firm
chords are indicated.”10 In fact, there may have been nothing modern about this
style. As we will see later, Czerny describes the arpeggio style as a modern
tendency in 1846.
Pauer was undoubtedly railing against a style of playing in which unnotated
arpeggiation and arpeggiation in general was an expressive tool. For him this
often resulted in oversentimental and weak expression. Such a style might well be
represented in Starlight Op. 55 No. 4 from Sea Pieces (1898), by the American
composer and pianist Edward MacDowell (1860–1908).11 In a footnote, MacDowell
instructs that chords marked with a bracket (“[”) “are not to be rolled.”12 But as
brackets signs seldom appear in “Starlight” (Fig. 3.6 ), the implication—
surprising as it might seem—is that the majority of chords were meant to be
arpeggiated. Presumably, fairly slow arpeggiations are appropriate here to enhance
the work’s tender character. The question remains, however, whether MacDowell
envisaged differences between the speed of arpeggiation of those chords marked
with an arpeggio sign (bars 35, 37, and 43) and those without.
Pauer, like many before and after him, encouraged the development of equality
of finger strength in chord playing to render all notes—particularly melody
notes—distinctly without resorting to arpeggiation. Nevertheless, all types of
arpeggiation may not have been considered faulty. The correlation between a slow
or drawn out arpeggio and a weakened effect is clear enough. But it is possible
that a slight or very tight arpeggiation producing the type of energized effect
described by Brée, rather than a weakened effect, may not have been considered
arpeggiation at all.
Indeed, some mid-nineteenth-century texts imply that firm chord playing and
arpeggiation were viable alternatives. In 1858, Lebert and Stark, who commented
on the usefulness of dislocation, advise on two ways of emphasizing the melody
note in a chord. In the first, “one can, and in most cases, should . . . release the
chords in the hand which contains the melody sooner [than the melody], which
naturally emphasizes the melody.” In the second, one can arpeggiate the chord
“while playing the melody note more strongly.”13 For them, arpeggiation was
clearly a suitable means of enhancing the expression of the melody note.

Pauer, Pianoforte Playing, 70.
Edward MacDowell, “Starlight Op. 55 No. 4,” Sea Pieces (Jung, 1898; Schmidt, 1899; Marian
MacDowell, 1926), 10.
Macdowell, Starlight, 11: MacDowell studied piano with Teresa Carreño and later at the Paris
Conservatoire and at Dr. Hoch’s Conservatory in Frankfurt. Note that Thalberg used this bracket for
the very same purpose fifty years earlier. See page 108.
Lebert and Stark, Klavierschule, part 3, 3: “man darf also und soll sogar in den meisten Fällen . . .
den der Melodie in der nämlichen Hand beigegeben Accord schneller auslassen, wodurch jene von
selbst hervortritt, oder harpeggiren, wobei der Gesang natürlich stärker angeschlagen wird.”
108 off the record

Around the same time in Paris, Thalberg also supported the practice of unno-
tated arpeggiation. In L’Art du chant, he advises, “The chords which carry a song or
melody to the higher note should always be played in arpeggio fashion, but very
tight and almost together, and the note of the melody more expressively than the
other notes of the chord.”14 Tantalizingly, he does not explain how the melody
note could be played more expressively. He may have meant that it should be
played more loudly or perhaps with some extra delay, or both.
Elsewhere in L’Art du chant, Thalberg presents solo piano transcriptions of pop-
ular vocal works indicating with signs where chords are either to be arpeggiated or
struck firm. His transcription Op. 70 of the Lacrymosa from Mozart’s Requiem K
626 is one of many examples showing where arpeggiated and unarpeggiated
chords are to be applied separately and together (Fig. 3.7).15 He states that to play
these transcriptions effectively, “all the chords which have the sign [[should be
played] rigorously together. Those [chords] which have the sign [ ? must be arpeg-
giated in a very tight [or dry] manner and almost together.”16 One is reminded
here of MacDowell’s markings in “Starlight.” Significantly, in bars 3 and 4 and
other places where the main melody is to be played softly, the supporting chords
are marked with arpeggio signs. But where the melody appears with a fortissimo
dynamic, Thalberg’s notation indicates strictly unarpeggiated chords. In general,
however, here and in other arrangements, Thalberg encourages a highly arpeg-
giated style of playing. In fact, throughout L’Art du chant, the sign “[” rarely
appears; the Mozart example is one of the few that uses it extensively. It is reason-
able to assume that this practice—once firmly part of a pianist’s expressive
palette—was applied freely at appropriate passages in other repertoire. Notably,
in bars 5, 6, and 7 and similar instances in which there is a break in the main
melodic line, arpeggiated and unarpeggiated chords appear in combination. Yet
neither the criteria underlying this application nor the resulting effects are clari-
fied. The correlation between such combinations and those encouraged by Brée
are obvious. Her explanations might, therefore, bear some resemblance to the
effect intended by Thalberg. In the foregoing arrangement, he is obviously
particular about their combination and must have intended them to create
specific effects.

Thalberg, L’Art du chant, unpaginated [2]: “Les accords qui porteront un chant à la note supérieure
devront toujours s’arpéger, mais TRÈS SERRÉS, Presque PLAQUÉS, et la note de chant plus appuyée
que les autres notes de l’accord”; trans. in Thalberg and Vieuxtemp’s Grand Concert Book, 5, as, “The
chords which carry a song or melody to the higher note should always be played in arpeggio fashion,
but very close and even, and the note of the melody more expressively than the other notes of the
Thalberg, L’Art du chant, 1–2.
Thalberg, L’Art du chant, 1: “Tous les accords portant ce signe [seront rigoureusement plaqués.
Ceux portant celui-ci?; devront être arpégés d’une manière tres serrée, presque plaqués.”
Unnotated A r peg g iation 109

Figure 3.7 Mozart Requiem arr. Thalberg, bars 1 to 7.

The many arrangements in L’Art du chant show that, for Thalberg, arpeggiation
was an intrinsic expressive device. But certain details remain unclear. For exam-
ple, he did not verbally state where arpeggiation was inappropriate. Furthermore,
in the foregoing references, he advises that the spreading of notes be very swift,
giving the impression that no variation of speed is permissible. This seems improb-
able for sophisticated piano playing. Yet the seemingly blanket rule may have been
propounded to prevent students and amateurs from making inappropriate arpeg-
giations resulting in exaggerated syncopations, clashes of harmonies, and large
gaps between successive melody notes. But in the absence of audible evidence, we
will never know whether Thalberg varied the speed of his arpeggiations, or not.
Thalberg’s contemporary—the violinist Charles de Bériot (1802–70)—also
promoted the swift arpeggiation of notes in a chord, which produced a desirably
brilliant effect. In his Méthode de violon Op. 102 (Paris, 1858), Bériot puts forward
a principle that he believed was universally recognized by instrumentalists.
Referring to chords that serve as “energetic articulations,” like those appearing at
the end of a piece, he advises that these “must be arpeggiated, in order to obtain
110 off the record

clarity and the desired force.” With reference to piano playing, he explains that
“many notes played together do not produce, overall, an effect as brilliant as when
a small interval is put between them, however small the interval.” This swift
arpeggiation of chords, the only successful way in de Bériot’s opinion, “must be
above all applied to the violin.”17
Interestingly, in their Violinschule (1905), Joachim and Moser reject Bériot’s
views about arpeggiating chords on the violin. They explain that it is quite possi-
ble with appropriate dexterity to play “three-part chords of short duration” with
all notes sounding together. No doubt the taste for such practices had changed
during the fifty year leading to the turn of the twentieth century. This change in
attitude also affected piano playing. Joachim and Moser assert, “Now [1905] it is
generally considered an extremely bad habit, even on the piano, to constantly
arpeggio chords which were meant to be struck as an unbroken whole.”18 Such a
statement implies, however, that many pianists were still heard making frequent
Thalberg’s arpeggiation practices no doubt evolved from earlier French
pianism. In his Méthode pour apprendre le piano-forte à l’aide du guide-mains Op. 108
(c. 1831), Frédéric Kalkbrenner (1785–1849) recommends unnotated arpeggia-
tion for particular moments in a phrase. Providing an annotated musical example
(Fig. 3.8), he insists that “in passages of double notes, octaves, or chords, the long
notes must be arpeggiated; those that precede must not be. All the notes that
have a o placed above, must be played together.”19 For Kalkbrenner, double notes,
octaves, and chords were suitable for arpeggiation if these coincided with the
highest or most harmonically dissonant points in the phrase. In these cases, the
chords leading to and from such points were to remain unarpeggiated. Notably, in
his illustration, the chords to be arpeggiated are also accented. This further sup-
ports the idea that arpeggiation could enhance such emphasis. Also noticeable is
the arpeggio sign next to the first chord in bar 2 and the final chord of the extract

Charles de Bériot, Méthode de violon Op. 102 3 parts (Paris, 1858), part 2, 86; facs. reprint
Méthodes et Traités: Violon, Vol. 6 (Courlay, France: Fuzeau, 2001): “Les accords dont nous voulons
parler ici sont pas de notes simultanées pour faire de l’harmonie soutenue, mais des articulations
énergiques, comme celles qui servent de terminaison à un morceau. Il est un principe reconnu pour
tous les instruments: C’est que les accords doivent être quelque peu arpégés, pour en obtenir la clarté
et la force voulues. En effet, il est à remarquer que sur le piano, par example, plusieurs notes frappées
ensemble ne produisent pas à beaucoup près un effet aussi brillant qu’en mettant entr’elles un petit
intervalle, quelque minime qu’il soit. Cette manière de produire les accords, la seule bonne à notre avis,
doit être surtout appliquée au Violon.”
Joseph Joachim and Andreas Moser, Violinschule, trans. Alfred Moffat (Berlin: Simrock, 1905),
vol. 2, 20a.
Frédéric Kalkbrenner, Méthode pour apprendre le piano-forte à l’aide du guide-mains, Op. 108
(2nd ed., Paris: Pleyel, 1831), 12: “Dans les passages en doubles notes, en octaves, ou en accords, les
notes longues doivent être arpégées; celles qui les précèdent ne doivent pas être. Toutes les notes sur
lesquelles il y a un o doivent être jouées ensembles.”
Unnotated A r peg g iation 111

Figure 3.8 Kalkbrenner, annotations of arpeggiated and unarpeggiated chords.

in the left hand. Presumably, here and in other cases arpeggiation enhances the
ending of a phrase or piece by producing a softened effect. In 1810 Phillip Corri
(cited later) shows a similar arpeggiation of a final chord. Elsewhere, Kalkbrenner
advises that “when playing compositions [originally] written for orchestra it is
necessary above all to suppress the arpeggios, the greatest merit of an orchestra
consisting in the ensemble.”20 Presumably, arpeggiation was not appropriate in
piano transcriptions of orchestral compositions because the resulting effect would
not reflect the synchrony inherent in orchestral style. In his Méthode complète de
piano Op. 100 (c. 1837), Henri Herz made a more direct comparison between the
orchestra and the piano: “Ensemble is the principal merit of an orchestra: the
piano, which imitates it [the orchestra], imposes on the player the same law, above
all in those passages that are dominated by a complicated harmony.”21 Nevertheless,
where the harmony was less complicated or the composition less orchestral in
style, some degree of arpeggiation was very probably permitted.
Kalkbrenner’s brief but noteworthy description must account for only a few
of the presumably innumerable situations in which arpeggiation was considered
necessary for expressive effect. Unfortunately, too, no mention is made of impor-
tant matters such as the speed and shape (note order) of arpeggios and their
placement on or before the pulse. Like so many other things, these were no doubt
left to the judgment of the player.

The Hidden Meaning in Portato Playing

For several generations of nineteenth-century pianists, arpeggio playing appears to
have been implicit when chords were marked with slurred staccato, or portato. In his

Kalkbrenner, Méthode, 12: “en jouant des choses composées pour l’orchestre il faut tout-à-fait
supprimer les arpèges, le plus grand mérite d’un orchestre consistant dans l’ensemble.”
Henri Herz, Méthode complète de piano Op. 100 (Mainz, Germany: Schott, [1838]), 86: “L’ensemble
fait le principal mérite d’un orchestre: le piano, qui en est l’imitation, impose à l’exécutant la même loi,
surtout dans les passages où domine une harmonie compliquée.”
112 off the record

Figure 3.9a Moscheles, portato chords.

Figure 3.9b Moscheles, arpeggio interpretation of portato chords.

Studies for the Pianoforte Op. 70 (London, 1827), the virtuoso pianist, pedagogue,
and composer Ignaz Moscheles (1794–1870) provides an annotated example (Fig.
3.9a to be performed as Fig. 3.9b) explaining that “when dots are used with slurs
over double notes and chords, these should be struck very slightly in the Arpeggio
manner, giving them the same length of time as a dot under a slur requires.”22 It is
probable that this practice had been going on for some time before Moscheles codi-
fied it. And hypothetically, for Moscheles and others, the arpeggiation of chords
marked portato may have been a substitute on the piano for the type of finger pres-
sure that could be achieved in clavichord playing.23
Moscheles’s Studies and accompanying remarks were widely disseminated.
They appeared in a later English edition in 1844,24 and a French edition in circa
1845.25 And it is of no little significance that Pauer’s revised edition, published in
London in 1886,26 still includes Moscheles’s instructions. Had Pauer been unequiv-
ocally opposed to the arpeggio manner—as implied in his foregoing advice—it
stands to reason that he might have removed or modified this part of the text or
at least registered his dissent in a footnote. Even as late as 1899, Gordon Saunders’s
edition reproduces Moscheles’s remarks.27 Perhaps unsurprisingly, Franklin
Taylor’s edition in 191528 suppresses them, in line with a growing trend that led to
the eradication of such practices. For most pianists today, portato chords do not
signify the arpeggio manner.
In particular circumstances Moscheles required the striking of notes in chords
absolutely together such as for compositions in the brilliant style. Accordingly,
his Study No. 13 (Allegro Brilliant) Op. 70 is intended to develop precision and

Ignaz Moscheles, Studies for the Piano Forte Op. 70, Bk. 1 (London: Cramer & Beale, 1827), 6.
See chapter 2, pages 69–71.
Moscheles, Studies for the Pianoforte Op. 70 (London: Cramer, Addison & Beale, 1844).
Moscheles, Etudes ou leçons de perfectionnement, avec notes explicatives sur la manière de les étudier
et de les executer, pour piano . . . Op. 70, 3rd ed. (Paris: Meisonnier, c. 1845).
Moscheles, Studies for the Pianoforte Op. 70, ed. Ernst Pauer.
Moscheles, Studies Op. 70, ed. Gordon Saunders (London: Hammond, 1899).
Moscheles, Etudes, ed. Franklin Taylor (London: Novello, 1915).
Unnotated A r peg g iation 113

facility for the “perfect performance of double notes and especially thirds.” He
warns the performer against yielding to weakness in the fingers that prevents the
notes in chords from “being struck with equal force and precisely at the same
time.”29 Moscheles’s comments may have been an attempt to stem the invari-
able use of unnotated arpeggiation, or simply provided for the benefit of technical

The Modern Manner of Arpeggiating in the First Half

of the Nineteenth Century
The supposed “modern” tendency of playing in the broken or arpeggio manner
mentioned by Pauer was seemingly just as pervasive during the first half of the
nineteenth century. The importance of Czerny’s notation of an arpeggio sign for
first chord of Beethoven’s Piano Concerto Op. 58 has already been noted.30 Other
written texts show that Czerny certainly approved of unnotated arpeggiation
when employed with propriety. But like Pauer, he was scathing of what he too
described as the “modern” practice of arpeggiating, particularly in compositions
of a contrapuntal nature. Giving an example (Fig. 3.10 ), Czerny states the
following in his Supplement (1846):

In the modern style, all passages in many parts are now invariably played
in arpeggio; and so greatly is this the case, that many pianists have almost
forgotten how to strike chords firmly. Many, otherwise really good play-
ers, would not be able to perform the following passage quite firm; that is,
to strike all the notes of each chord exactly together.31

Clearly, Czerny’s description implies a highly arpeggiated style. No doubt some of

the best players considered arpeggiation absolutely necessary to play this chordal
texture softly and legato.
Elsewhere, Czerny remonstrates that in fugue playing “every note must be sus-
tained precisely according to its value; and performing in arpeggio, or striking the
notes of the different parts one after another, is by no means permitted. In this
respect the pianoforte must be treated exactly like the organ, where all chords are
struck exactly together.”32 In reality, however, piano playing of this era probably
reflected anything but unarpeggiated organ style. As Clive Brown has noted, a
letter written in 1829 by the famous organist Samuel Wesley (1766–1837)

Moscheles, Studies Op. 70, Bk. 2, 1.
See chapter 2, pages 43–44.
Czerny, Art of Playing, 157.
Czerny, Art of Playing, 126.
114 off the record

strongly suggests “the ubiquity of arpeggiation in piano playing in England during

the early decades of the nineteenth century.”33 Wesley observes that in playing
any chord, pianists “do not put down the Keys simultaneously which on the Organ
should always be done, but one after another, beginning at the lowest note of the
Base: so that (to use a harsh military Metaphor) the Effect on the Ear is not that
of a general instantaneous Explosion but rather of a running Fire.”34 The over-
whelming impression here is that most chords were arpeggiated. And this was
apparently the case particularly in slow movements. As William Sheppard states
in A New Pianoforte Preceptor (London, 1824), “In slow movements it is better to
spread the Chords whether they are marked or not.”35
Other writers infer that only certain chords should be arpeggiated to give them
special emphasis. In 1848, the German composer, critic, and musicologist Adolf
Bernhard Marx (1795–1866) discussed the “veracity of an oral tradition” that
promoted the arpeggiation of chords, sometimes in an extended manner. With
reference to a passage from J. S. Bach’s Chromatic Fantasie BWV 903 (Fig. 3.11), he
explains that “when we wish to emphasise particular chords even on our full-
sounding instruments we do not play the notes exactly together, but rather in a
quick arpeggio, whilst holding down all the keys such as here at a.” Further to this,
Marx hypothesizes that “on the weaker sounding instruments of Bach’s time, this
method of playing must have been even more necessary—perhaps with an even
slower arpeggiation, possibly also descending again to freshen those notes which
had faded.” He suggests that such practices represent the limits of what was
appropriate in Bach’s tradition and that they cannot, therefore, be allowed to
influence modern players into “crumpling every wonderful chord (as . . . at b or
even with two arpeggios per chord) and so to spoil all moments of stillness and
the decisive pulses of performance.”36 The implication is that some mid-nine-
teenth-century pianists were making these types of arpeggiations in Bach’s
Chromatic Fantasie and perhaps also in other repertoire.

Brown, Classical and Romantic Performing Practice, 612.
Philip Olleson, ed., The Letters of Samuel Wesley: Professional and Social Correspondence,
1797–1837 (Oxford: Oxford University Press, 2001), 439.
William Sheppard, A New Pianoforte Preceptor (London: c. 1824), 55. See also Gerhard,
“Willkürliches Arpeggieren,” 128–29.
Adolf B. Marx, “Seb. Bach’s chromatische Fantasie—Einige Bemerkungen,” Allgemeine musikalis-
che Zeitung vol. 50, no. 3 (Jan. 19, 1848): 36–37fn; See also Gerhard, “Willkürliches Arpeggieren,” 125:
“Schon auf unseren klangvollen Instrumenten geben wir Akkorde, die mächtig hervortreten sollen,
nicht mit genau gleichzeitigem Auschlag an, sondern in reissend schneller Brechung, unter Festhalten
aller Töne . . .; bei den klangarmen Instrumenten der bach’schen Zeit muss diese Spielweise—und viel-
leicht langsamere Brechung, vielleicht selbst ein theilweises Zurückgehen, um die verklungenen Töne
wieder anzufrischen—noch viel nothwendiger gewesen sein. Hierauf wird sich wohl die überlieferte
Anweisung Bach’s bezogen und beschränkt haben; sie kann mithin uns nicht bestimmen, jene her-
rlichen Tonmassen zu verkrümeln (wie oben bei b, oder gar mit zweimaligen Arpeggio jedes Akkords)
und damit alle Ruhepunkte und die entscheidenden Schläge des Spiels zu verscherzen.”
Unnotated A r peg g iation 115

Figure 3.11 Bach Chromatic Fantasie BWV 903, annotations by Marx.

In this respect, it is interesting to consider Beethoven’s indication of the word

arpeggio in the final bar of the second movement of his Appassionata Sonata
Op. 57. In the autograph he indicated a secco chord in the right hand and inserted
an arpeggio sign next to the chord in the left hand. Additionally, he wrote the
word arpeggio underneath the chord in the left hand. In the first edition, however,
the word secco was removed and an arpeggio sign inserted next to the chord in the
right hand. The word arpeggio was retained under the chord in the left hand.
Subsequent editions followed either the autograph or the first edition or simply
marked both right and left hands with separate arpeggio signs. In his 1886 edi-
tion, Reinecke did the latter but also marked the word arpeggio between the staves
(Fig. 3.12).37 A few editors have maintained that Beethoven fully intended a secco
chord against an arpeggiated chord and that he wrote the word arpeggio to make
his meaning absolutely clear. But another plausible explanation is that he intended
the pregnant fortissimo chords—at least in the left hand—to be arpeggiated
(perhaps unusually) up and down several times to create a dramatic and/or sus-
tained effect. Interestingly, in his instructive comments of 1871, Hans von Bülow
explains that the penultimate chord may be arpeggiated “very slowly and dream-
ily,” but that the final chord must, above all things, “sound energetic.” Additionally,
he advises that the length of the pause on the final chord is dependent on the
sonority of the instrument.38 It stands to reason that in this circumstance, he
might have envisaged a final chord that was arpeggiated up and down several
times in order to make it both energetic and sustained. In 1918, Walter Niemann’s
Klavier Lexicon—a work intended as a guide to contemporary performance style—
prescribed something along these lines, calling it an older ornament (Fig. 3.13a
and Fig. 3.13b).39

Ludwig van Beethoven, Sonaten für Pianoforte, ed. Carl Reinecke (Leipzig, Germany: Breitkopf &
Härtel, 1886), vol. 2, 67.
Beethoven, Sonatas and Other Works, ed. Bülow, part 3, vol. 4, 77.
Walter Niemann, Klavier Lexicon (Leipzig, Germany: Kahnt, 1918), 46.
Figure 3.12 Beethoven Sonata Appassionata Op. 57, second movement, penultimate
and final bars, ed. Reinecke.

Figure 3.13a Arpeggiation of chords in both hands, old ornament, annotated by


Figure 3.13b Arpeggiation of longer chord, old ornament, annotated by Niemann.

Unnotated A r peg g iation 117

Figure 3.14 C. P. E. Bach Sonata VI Wq 63/6, third movement, last six chords.

This type of arpeggiation certainly had historical precedents. In the mid-

eighteenth century, C. P. E. Bach advised that the word arpeggio over a long note
“calls for a broken chord upward and downward several times.”40 Let’s not forget
Beethoven’s admiration and knowledge of Bach’s teachings. A clear example of
this appears at the end of the third movement of his Sonata VI Wq 63/6
(Fig. 3.14). A half-century earlier, in circa 1700, North advised that arpeggia-
tions should be “as swift as possible” and that when an arpeggio sign is notated
directly in front of three-note chords, the notes “must be broken in going up and
down.”41 It is conceivable that such a practice continued on during the nineteenth
With little doubt, many pianists made habitual use of arpeggiation during the
first half of the nineteenth century. In 1839, Czerny registered a stern reaction to
this in his Theoretical and Practical Pianoforte School Op. 500. He complains that
“many players accustom themselves so much to Arpeggio chords that they at last
become quite unable to strike full chords or even double notes firmly and at once;
though this latter way is the general rule, while the former constitutes the
exception.”42 Clearly, Czerny did not favour arpeggiations made continuously;
however, he certainly approved of them in specific situations. Noting that arpeg-
gios could “frequently be employed with effect,” he described situations in which
arpeggios, as a general rule, were to be avoided:

1. All chords consisting of very short notes, should be struck firmly and at once,
when the Composer has not expressly indicated the contrary [Fig. 3.15a].

Bach, Versuch, part 1, 159.
Roger North, Capt. Prencourts rules (c. 1700), in Graham Strahle, An Early Music Dictionary
(Cambridge: Cambridge University Press, 1995), 18.
Carl Czerny, Vollständige theoretische-practische Pianoforte-Schule Op. 500 (Vienna: 1839); trans.
as Theoretical and Practical Pianoforte School Op. 500 (London: Cocks, [1839]), vol. 3, 55.
118 off the record

Figure 3.15a Czerny, chords consisting of short notes.

2. Such chords as require to be played with very great power, particularly when
they form the commencement or the close of the piece, or of any considerable
portion of one, almost always produce the best effect when they are struck
plain; as arpeggioing always diminishes and destroys some part of the Forte.
The same rule applies when two or more chords follow one after another very
quickly [Fig. 3.15b]. The Composer should always indicate where he desires to
have these chords played in Arpeggio.
3. Passages in several parts, which form a connected melody, or which are
written in the syncopated or strict style, must always be played with firmness
and exactly as written; and it is only occasionally, that a single, slow, and full
chord, on which a particular emphasis is required, may be played in Arpeggio
[Fig. 3.15c]. Only the 3 chords distinguished by a + (the last one in every case)
will admit of a moderate arpeggio, which, however, must not interrupt the

Czerny, Pianoforte School Op. 500, vol. 3, 55–56.
Figure 3.15b Czerny, chords to be played with great power.

Figure 3.15c Czerny, passages in several parts or in the syncopated or strict style.

Figure 3.15d Czerny, slow and sustained chords.

120 off the record

Figure 3.15e Czerny, long chords interspersed with shorter chords.

Figure 3.15f Czerny, long chords interspersed with shorter chords played staccato.

Figure 3.15g Czerny, gradations of arpeggio speed.

It is somewhat surprising to find that no arpeggiation is permitted for the par-

ticularly long chords at the beginning of the Czerny’s second example (Fig. 3.15b).
Although he felt that arpeggiation detracted from the effect of forte, and other
writers advised that it caused a weakening effect, Bériot, as we have seen, advised
that quick arpeggiation could enhance brilliancy. And as we will see later, Corri,
like Bériot, equated rapid arpeggiation with a brilliant effect earlier in the
century. Perhaps more surprising is that in strict compositions of several parts
(Fig. 3.15c), certain arpeggiations were considered permissible, particularly where
chords formed especially poignant harmonies or consisted of an unusually large
Unnotated A r peg g iation 121

number of notes. Even in contrapuntal music, therefore, unnotated arpeggiations

were appropriate for giving certain chords particular emphasis or color. Czerny
went on to explain the situations in which unnotated arpeggiations could effec-
tively be applied:

1. In all slow and sustained chords which do not form any melody [Fig. 3.15d].
The last chord in the 4th bar must not be sprinkled, as it closes a section of
the melody; while all the other chords must be arpeggioed with moderate
quickness, yet that the upper or melodial note shall never come in out of its
2. When after a long and smoothly connected chord, several others occur which
are quicker, only the first one must be arpeggioed [Fig. 3.15e]. Here only the
chords distinguished by + are to be arpeggioed. It is still more necessary to
observe this rule, when the quicker chords are at the same time to be played
staccato [Fig. 3.15f ].
Here too only the 3 chords marked + can be played in Arpeggio.
3. In arpeggioing, the single notes may not only be played so extremely fast, that
the arpeggioed chord shall almost resemble a chord struck plain; but they may
also be played slower and slower, in every possible gradation, down to that
degree in which each single note will be equal in duration to a crotchet in slow
time; we must measure and apply these different degrees, exactly according as
the chord is to be held down long or quickly detached, and struck either piano
and smorzando, or forte and hard [Fig. 3.15g]. Here the single notes of the
arpeggioed chords must follow one another extremely slow, and we only begin
to count the time prescribed from the last and highest note. To this extension of
the time we are entitled, as the passage forms a sort of pause. If, however, this
passage were marked Fortissimo, the Arpeggio should not by any means be so
slow, but rather very quick; or, still better, not be employed at all, unless actu-
ally prescribed by the Author himself.45

There is an obvious similarity between Czerny’s first rule (Fig. 3.15d) and that
of Thalberg cited earlier. Notably, he advises that the final chord in this example
should be struck firmly, presumably to make a contrast with the arpeggiated
chord that precedes it. Therefore, in a standard “feminine” cadential formula, the
six-four chord receives color and emphasis by arpeggiation, and resolves to an
unarpeggiated and unaccented five-three chord. Czerny also advises that the
speed of arpeggiation be of moderate quickness but so that the melody note
sounds in time. The only successful way of achieving this is to commence the

Czerny, Pianoforte School Op. 500, vol. 3, 56. Note that the figured bass symbols in Fig. 3.15d
are added by N. Peres Da Costa.
Czerny, Pianoforte School Op. 500, vol. 3, 55–56.
122 off the record

arpeggiation before the beat; arpeggiating from the beat would result in a delay of
the melody note. It is difficult to appreciate what speed or range of speeds Czerny
equated with the term “moderate quickness.” It is possible that this expression
had a similar meaning to Thalberg’s “very tight” or “almost together.” Unfortunately,
however, such descriptive language lacks clarity.
There is also a clear correlation between Czerny’s examples (Figs. 3.15e and
3.15f ), in which equal-valued chords remain unarpeggiated, and Kalkbrenner’s
foregoing example showing a similar practice. Notably, Czerny’s need to remark
that staccato chords should not be arpeggiated perhaps hints at the fact that some
pianists made frequent unnotated arpeggiations regardless of context or situa-
tion. Most important, Czerny advocates the playing of arpeggios in varying speeds
depending on the character or features of the composition and its intended effect.
At least in theory, there is confusion between this and the apparent inflexibility of
arpeggio speed promoted by Thalberg. In the end, and although Czerny’s advice is
more detailed than many other references, several issues remain unclear.
Unnotated arpeggiation was certainly a feature of early-nineteenth-century
pianism as evidenced in two separate sources by the Englishman Philip Corri. In
his Original System of Preluding (London, 1813), Corri discusses chord playing
with reference to an exemplar prelude (presumably the first one in C, Fig. 3.16a).
He explains that the notes in the long chords that conclude the prelude “should
not be struck together,” but should sound with “a long extended appoggiando.”
On the other hand, chords at the beginning of any run or passage—for example,
the chord marked with a sforzando in the prelude—“should have emphasis and
should be played more together, and with more firmness.” Finally, several con-
secutive chords, like those at the beginning of the prelude (notably these are
marked with arpeggio signs), “should be played almost together and not
appoggiando.”46 As Hamilton has pointed out, Corri states that the performance
style for chords in preludes “ranges from ‘not together’ to ‘almost together,’ totally
omitting the modern default option—completely together except where other-
wise indicated.”47
It is possible that this predominantly arpeggiated style was reserved for improvi-
satory-style pieces such as preludes because in L’anima di musica (London, 1810)
Corri specifies the use of chords with notes “struck” or “played together” for works in
other genres. But whether even these expressions call for absolute synchrony of
notes is questionable. They may simply describe an effect that is neat but not abso-
lutely synchronized. With reference to various illustrations, Corri asks the reader to

observe that in the . . . Example [Fig. 3.16b], the longer notes only, are to
be played appogiando; those that are equal are to be struck together, tho’

Philip A. Corri, Original System of Preluding (London: Chappel, 1813), 4.
Hamilton, After the Golden Age, 155.
Unnotated A r peg g iation 123

Figure 3.16a Corri, exemplar prelude.

Figure 3.16b Corri, example played monotonous without expression.

not staccato; and the end of the tie must have the cadence or fall, that is;
to be touched lightly.
But if on the contrary, all the chords are played appogiando, without
distinction, the Time and Metre would be so confused and disguised that
no air or melody could be discoverable, and therefore, it should be
remember’d that where notes or chords are of equal length, in succes-
sion, they should all be played together.*
To prove what I have just asserted play the foregoing Example with all
the notes appogiando and without emphasis—Judge then which is the
most pleasing style; the 1st [Fig. 3.16b]—monotonous without expres-
sion, the 2nd [Fig. 3.16c] with proper expression—or the 3rd as just
directed, with an excess of expression.
The latter style is two [sic] often adopted by those who affect to play
with Taste and who from ignorance of its effects, distort and disfigure
the melody so hideously that no one can make it out; I therefore recom-
mend the appogiando to be used cautiously and sparingly.
There are occasions where the appogiando may be used, altho’ it be
not for emphasis, for instance;—in a slow strain, the long chords are to
be sustained, tho’ there are many of the same quality, yet their harmony
is better heard, and produces more effect by being touch’d appogiando,
(As the Minims in the following [Fig. 3.16d]) but then observe that the
124 off the record

Figure 3.16c Corri, example played with proper expression.

Figure 3.16d Corri, arpeggiation of long chords.

Crotchets that follow, being shorter, ought to be played together as a

relief to the other style.—
Further Examples [Fig. 3.16e], shewing that the appogiando should
be used on the long chords; and also on shorter ones, where brilliancy is
required to be given, touching them as nearly as possible together.
When the words “con espressione, con Anima, or Dolce etc.” are mark’d
at a passage, it signifies that the appogiando must be particularly and
often used, and made as long as possible.
* [Corri’s footnote] There is an exception which I shall next explain.48

Philip A. Corri, L’anima di musica (London: 1810), 76–77.
Unnotated A r peg g iation 125

Figure 3.16e Corri, arpeggiation to give brilliancy.

The similarities between Corri’s and Czerny’s advice are clear. Like Czerny,
Corri admonished the practice in which all chords are arpeggiated, explaining
that some pianists had adopted precisely this approach. Of paramount impor-
tance is Corri’s suggestion that certain common descriptive terms such as
con espressione, con anima, and dolce were unequivocal indications that arpeggia-
tion must be applied frequently. Where such expressions appeared there was no
need for the composer to insert arpeggio signs: their skillful addition was
taken for granted. This is particularly significant because such expressive terms
no longer carry this meaning. To make the arpeggios sound as long as possible in
these instances, various means may well have been adopted: (a) by varying
the speed of arpeggiation (from quite slow to quite fast) in accord with the char-
acter or affect of the music; (b) by prolonging the arpeggio—playing the
notes upward and downward in the way prescribed by Bach and North (cited
earlier); (c) by modifying the speed and shape of arpeggios—accelerating and/or
decelerating successive notes; (d) and possibly by adding nonharmonic tones,
which help fill out the texture of the chord and give it particular color, more of
which later.

Earlier Keyboard Practices

Without doubt, the practice of unnotated arpeggiation in piano playing stemmed
from earlier keyboard practices. In his Instructions for the Pianoforte (London,
1812), Johann Baptist Cramer (1771–1858) described the art of accompanying
singers and instrumentalists by improvising a chordal texture according to a
figured bass. He explains that chords may be played either in “an abrupt manner
striking all the Notes at once,” which is chiefly reserved for the end of a piece or a
sentence. Alternatively, the notes are arpeggiated and held down for the full value
of the chord. Further to this, he explains that various signs are generally used to
denoted arpeggios [Fig. 3.17a and 3.17b], implying however that on an undefined
126 off the record

Figure 3.17a Cramer, arpeggio sign.

Figure 3.17b Cramer, arpeggio sign.

number of occasions, arpeggios were left to the whim of the performer.49 Notably
in Figure 3.17a, the arpeggiations commence with the beat, not before it.
Cramer’s descriptions correspond closely to earlier keyboard practices. Notably,
he considered unarpeggiated chords to have an abrupt effect. This can be related
to the sound of instruments such as the harpsichord, in which the simultaneous
striking of the notes of chords (and therefore the simultaneous plucking of
strings) naturally produces a strong and accented sound. In the mid-eighteenth
century, C. P. E. Bach suggested that unarpeggiated chords were suitable for more
lively sections in recitatives, stating that “as soon as the accompaniment shifts
from sustained to short, detached notes, the accompanist must play detached,
resolute chords, un-arpeggiated, and fully grasped by both hands.”50 In 1751,
Bach’s contemporary Foucquet implied much the same thing when he explained
that arpeggiation of chords (from lowest note to highest note) in the accompani-
ment “renders the touch mellow and graceful—indispensable for pieces of
sentiment.”51 Later in the eighteenth century, the unarpeggiated style was also
recommended by Türk, for example, to make a loud effect or emphasis particu-
larly for dissonant harmonies so “that the passions should be especially aroused.”52
The synonymity of unarpeggiated chords with loud, accented, and abrupt effects

Johann B. Cramer, Instructions for the Pianoforte (London: Chappell, 1812), appendix, part 4, 42.
Bach, Versuch, 422.
Foucquet, Pièces de clavecin, preface to Second livre: “S’il se rencontre plusieurs notes dans la
basse, il faut les harpeger, c’est à dire commencer par la plus basse et ainsi de suite, observant dans le
dessus de faire entendre la plus haute, la dernière, ce qui rend le toucher moëlleux, gracieux et indis-
pensable pour les pièces de sentiments.”
Türk, Klavierschule, 340.
Unnotated A r peg g iation 127

signifies at least some relationship between harpsichord and early piano

Cramer reserves the abrupt style mainly for the ends of phrases or composi-
tions. This implies that other moments require a different effect, presumably the
arpeggio style. And like Czerny, Cramer advises that chords may be arpeggiated
“with more or less velocity, as the character of the piece requires.”53 The Dresden
court musician Johann Peter Milchmeyer gave similar advice to pianists at the
very end of the eighteenth century. In Die Wahre Art Das Pianoforte Zu Spielen
(1797), perhaps the first treatise devoted exclusively to piano playing, Milchmeyer
explains that arpeggiated chords (usually marked with a symbol) can be broken
upward or downward and “express certain passions very well.” He warns against
impairing the character of slow pieces by arpeggiating the chords too quickly.54
This again relates closely to harpsichord playing in which the arpeggiation of
chords with a variety of speeds was considered a type of ornament and left to the
judgment of the player. C. P. E. Bach advises that in recitative accompaniment, the
pace of chordal arpeggiation varies according to the tempo and content of a recita-
tive: “The slower and more affetuoso the latter is, the slower the arpeggiation.”55
Here, the accompanist becomes the composer, embellishing the music by varying
the speed and shape of arpeggios. In his “Prendcourt Tracts” (c. 1710–c. 1716),
North alludes to this in explaining that a player can learn to play well according to
a figured bass but “may not pretend to be a master of his part without being a
master of composition in generall [sic].” The improvisation of this type of accom-
paniment requires a great deal of management in the manner of playing, “some-
times striking onely [sic] the accords, sometimes arpeggiando.”56
Continuo players often enhanced both the texture and the color of their impro-
vised chords by introducing passing or nonharmonic tones. This is described by
Michel de Saint Lambert (1610–96) as “arpégés figurés” in Les Principes du clavecin
(Paris, 1702),57 and by Francesco Gasparini (1661–1727) in Larmonico Pratico al
Cimbalo (Venice, 1708).58 North mentions similar practices with particular regard
to “the ornamentall [sic] (or figurate) composition of harmony”—almost certainly

Cramer, Instructions, appendix, part 4, 42.
Johann P. Milchmeyer, Die Wahre Art Das Pianoforte Zu Spielen (Dresden, Germany: 1797),
trans. as The True Art of Playing the Pianoforte by Robert Rhein (DMA diss., University of Nebraska,
1993), 78.
Bach, Versuch, 422.
Roger North’s “Of Sounds” and “Prendcourt Tracts,” Digests and Editions, by Mary Chan and
Jamie C. Kassler with an analytical index by Janet D. Hine (Kensington: University of New South
Wales, 2000), 52.
Michel de Saint Lambert, Les Principes du clavecin (Paris: 1702); trans. as Principles of the
Harpsichord, by Rebecca Harris-Warrick (Cambridge: Cambridge University Press, 1984), 94.
Francesco Gasparini, Larmonico Pratico al Cimbalo (Venice: 1708); trans. as The Practical
Harmonist at the Harpsichord by Frank S. Stillings, ed. David L. Burrows (New York: Da Capo Press,
1980), 78–84.
128 off the record

the instrumental act of flourishing. Referring to violin playing, but relating it to

harpsichord, lute, and harp playing, he explains that through the divisions (arpeg-
gios) caused by such flourishes “may be heard all the concords, commixtures, and
passing notes, as if all were in full action.”59
Unnotated arpeggiation was certainly practiced during the seventeenth century.
As noted in chapter 2, Frescobaldi advised in 1615 that the beginnings of his toc-
catas, which are often chordal in texture, should be arpeggiated (Fig. 3.18 ).60
Influenced by Frescobaldi, Johann-Jakob Froberger’s (1616–67) toccatas (pub-
lished in 1649 and 1656) contain many similar places where, depending on the
choice of keyboard instrument, arpeggiation is desirable though not indicated
(Fig. 3.19 ). And the préludes non mesurés of French harpsichordists give clear
indications of arpeggio shapes but leave elements of rhythm and speed to the
Today, arpeggiation is widely accepted as a historically appropriate and expres-
sive technique in harpsichord and clavichord playing. As Troeger explains, on
these instruments “the breaking of chords, ranging from barely perceptible
spreading to elaborate patterns, both free and rhythmic, is important for tone
quality, for stress and timing and as an embellishment to playing.”62
From the foregoing evidence, it is safe to conclude that the practice of unno-
tated arpeggiation preserved on many early-twentieth-century recordings was
widely cultivated throughout the nineteenth century and represents a continua-
tion of earlier keyboard practices. In this light, it is impossible to appreciate what
Corri, Czerny, or Pauer would really have considered an overuse of unnotated
arpeggiation. Perhaps all they were truly criticizing was the inartistic application
of arpeggios causing distortion of the phrase or the melodic line, their rules for-
mulated to set some boundaries for students and amateurs that did not necessar-
ily apply to trained artists. It is inappropriate, therefore, to judge their idea of
the sparing use of unnotated arpeggiations by today’s much more text faithful
In any case, it is clear that during the nineteenth century, musical notation
simply did not preserve many such practices like arpeggiation as were considered
intrinsic to musical expression, any more than composers normally indicated, say,
vibrato or portamento in violin music. In this respect, Brown’s observation about
the ornamental nature of arpeggios is of particular interest: “As with all such
ornaments in this period, there is no reason to think that composers troubled to
mark every place where they might have expected, or been happy to have heard

Wilson, Roger North, 193–94.
See chapter 2, note 84, page 68; Frescobaldi, Toccate e partite: “Li cominciamenti delle toccate
sieno fatte adagio, et arpeggiando.”
See Figures 2.11 and 2.12, page 61.
Troeger, Technique and Interpretation, 138.
Unnotated A r peg g iation 129

arpeggiation, or that they specified every aspect of its performance.”63 It was the
keyboard players’ responsibility to be creative by embellishing the score according
to current notions of good taste.

The Evidence of Recordings

With little doubt, the practice of unnotated arpeggiation, scorned by Pauer in the
second half of the nineteenth century and encouraged by Brée in the early twen-
tieth century, had significant historical precedents. As we have seen, Brée cata-
logued some arpeggiation practices promoted by Leschetizky, which fortunately
can be compared with his own performance style. Unnotated arpeggiation is
clearly evident in Leschetizky’s performances of Mozart’s Fantasia K 475 and
Chopin’s Nocturne Op. 27 No. 2, preserved on his 1906 piano rolls.64 Although he
makes them more sparingly in Mozart’s Fantasia than in Chopin’s Nocturne,
examples from both reveal how his unnotated arpeggiations contribute to a
variety of expressive effects.
On beats 1 and 3 in bars 16 and 17 of the Fantasia (coinciding with the enhar-
monic shift from G flat to F sharp in the bass), Leschetizky arpeggiates noticeably
the chords that accompany the expressive sighing figures (Fig. 3.20). Here, as in
most of the following examples, the separation of notes sounds fairly continuous
from the lowest note in the left hand to the highest note in the right hand. To my
ears, the first of the sighing figures in bar 16 is spread more slowly than those that
follow. These arpeggiations contribute substantially to the character of longing,
delineating the two bars from the material that precedes and follows. In bars 26,
28, and 32, containing similar thematic material, Leschetizky makes arpeggia-
tions to the chords marked sforzando (Fig. 3.21). This was presumably so that the
sforzando effect itself was at an appropriately soft dynamic level, giving emphasis
without harshness. Elsewhere, Leschetizky’s added arpeggios enhance the myste-
rious atmosphere of the music. This is particularly evident at the interrupted
cadence in bar 33 (Fig. 3.22) and the passage from bar 34 leading to the unex-
pected dramatic section marked Allegro (Fig. 3.23). For the final figure—also the
softest and the most questioning—in bar 35, Leschetizky’s arpeggiation is notice-
ably slower than the previous figures. In certain instances, Leschetizky added
arpeggios to one hand only; the notes in the other are struck together. This is
noticeable of the chord in the left hand at the beginning of bar 84 in the cadenza
section, which is arpeggiated quickly, whereas the notes of the octave in the
right hand are played together. The effect is of renewed energy without harshness

Brown, Classical and Romantic Performing Practice, 610.
Note that the aural effect of arpeggiations, unless otherwise stated, commences before the
notional beat, with the highest note aligned with the beat.
Figure 3.20 Mozart Fantasia, bars 15 to 18, Leschetizky, piano roll recording, 1906
(Audio Ex. 3.1 ).

Figure 3.21 Mozart Fantasia, bar 26, Leschetizky, piano roll recording, 1906 (Audio
Ex. 3.2 ).

Figure 3.22 Mozart Fantasia, bar 33, Leschetizky, piano roll recording, 1906 (Audio
Ex. 3.3 ).

Unnotated A r peg g iation 131

Figure 3.23 Mozart Fantasia, bars 34 and 35, Leschetizky, piano roll recording, 1906
(Audio Ex. 3.4 ).

Figure 3.24 Mozart Fantasia, bars 83 and 84, Leschetizky, piano roll recording, 1906
(Audio Ex. 3.5 ).

Figure 3.25 Mozart Fantasia, bar 95, Leschetizky, piano roll recording, 1906 (Audio
Ex. 3.6 ).

following the descending scale in the previous bar (Fig. 3.24) and correlates closely
with Brée’s description of this technique. In bar 95, the lowest note of the octave
on the third beat in the left hand is aligned with the pulse, and the upper note is
struck with the unarpeggiated chord in the right hand (Fig. 3.25). This, too, cor-
relates with Brée’s description in which the technique causes a slight delay of the
melody note, giving it special emphasis. Between bars 86 and 93, Leschetizky
132 off the record

Figure 3.26 Mozart Fantasia, bars 86 to 93, Leschetizky, piano roll recording, 1906
(Audio Ex. 3.7 ).

makes several arpeggiations that create differing effects (Fig. 3.26).65 A moder-
ately slow arpeggio softens and thereby gives a gentle expression to the opening
chord of the phrase commencing at bar 86, and the feminine cadence with which
it ends at bar 89. During the phrase commencing at bar 90, arpeggiation enhances
each successive thematic fragment, with the most poignant and the slowest spread
saved for the chord at the beginning of bar 91. The chord at the beginning of bar
92 is given a gentle expression by arpeggiating it, and the feminine cadence at bar
93 is treated as at bar 89, with an arpeggiated chord resolving to a firmly struck
chord. At such cadence points, the pattern of arpeggiated followed by unarpeg-
giated chords enhances the effect of strong and weak, or tension and release.66
Leschetizky’s performance of Chopin’s Nocturne Op. 27 No. 2 provides yet
more fascinating examples of his use of unnotated arpeggiation. Here, the notes
of chords comprising thirds, sixths, and octaves in the right hand are separated in
a variety of ways. Sometimes, as exemplified in bars 10, 12, and so on, the lower
note of the chord in the right hand anticipates the upper note that is aligned with
the corresponding note in the left hand (Fig. 3.27). At such moments, the separa-
tion itself causes the melody note to be emphasized. In bars 13 and 40, this type
of arpeggiation helps to mark the poignant syncopation caused by the accented

Note that only chords that sound noticeably arpeggiated are marked. Others appear to be very
tightly spread.
Carl Reinecke’s circa 1905 roll recording for Hupfeld of the same work reveals remarkably similar
arpeggiations, which is noticeable, for example, in bars 16 and 17 (Audio Ex. 3.8).
Unnotated A r peg g iation 133

chord on the third quaver beat. Again, the upper melody note sounds emphasized
because of the pregnant delay (Fig. 3.28). At other times, such as bars 14, 18, 21,
and 33, the lower note of the chord is aligned with the corresponding note in the
left hand; the upper melody note is again emphasized by the delay. In bar 14, the
arpeggiation of the first chord in the right hand is coupled with a dislocation of
the hands, thus further delaying the upper melody note (Fig 3.29). In bar 33, the
arpeggiation of the first chord in the right hand has the effect of continuing the
triplet figurations of the previous bar (Fig. 3.30). In bar 37, both types of arpeg-
giation with lower note anticipating (third quaver chord in the right hand) and
aligned with the corresponding bass note (fifth and sixth quaver chord in the right
hand) are noticeable (Fig. 3.31). And between bars 71 and 74, the two voices in
the right-hand polyphony are delineated by arpeggiations at every possible

Figure 3.27 Chopin Nocturne Op. 27 No. 2, bar 10, Leschetizky, piano roll recording,
1906 (Audio Ex. 3.9 ).

Figure 3.28 Chopin Nocturne Op. 27 No. 2, bar 13, Leschetizky, piano roll recording,
1906 (Audio Ex. 3.10 ).
134 off the record

Figure 3.29 Chopin Nocturne Op. 27 No. 2, bar 14, Leschetizky, piano roll recording,
1906 Audio Ex. 3.11 ).

Figure 3.30 Chopin Nocturne Op. 27 No. 2, bars 32 and 33, Leschetizky, piano roll
recording, 1906 (Audio Ex. 3.12 ).

moment and with varying speeds. Here, the lower note of the chord is generally
aligned with the pulse, though sometimes there is also a slight dislocation between
the hands. The chords at the beginning of bars 72 and 74 are spread more slowly
than the other chords (Fig. 3.32). Brée recommended this technique to bring out
the polyphony distinctly, but her description fails to convey both the frequency
and the nature of Leschetizky’s way of applying it.
In summary, it is clear that Leschetizky made unnotated arpeggiations in many
varied ways. In the piano rolls examined earlier, these help create the following:

The effect of longing

The differentiation among chords of varying characters, thus effecting
dramatic contrast
Figure 3.31 Chopin Nocturne Op. 27 No. 2, bar 37, Leschetizky, piano roll recording,
1906 (Audio Ex. 3.13 ).

Figure 3.32 Chopin Nocturne Op. 27 No. 2, bars 71 to 74, Leschetizky, piano roll
recording, 1906 (Audio Ex. 3.14 ).

136 off the record

A sense of softening and ending

A particular emphasis lacking harshness for chords requiring accentuation
The mysterious nature of an interrupted cadence or the increase of tension in
the transition to a pregnant pause
The energetic effect achieved by the combination of an arpeggiated chord in
the left hand with a chord struck firmly in the right hand
A sense of tension and release at feminine cadence points
A varied expression for a sequence of thematic fragments, in which the slowest
arpeggiation is saved for the most important moment
A gentle expression for the beginning of a phrase
The delineation of different voices in a polyphonic texture

Brée’s descriptions provide a broad indication of Leschetizky’s unnotated arpeg-

giation practices. However, Leschetizky’s piano rolls reveal a greater complexity
and subtlety of arpeggiation. Evidently, her text did not have the scope to cover
these, nor was it necessarily her intention.
Interestingly, Leschetizky’s own edition of the same Nocturne gives no indica-
tion of the unnotated arpeggios preserved on his piano roll. Clearly, such prac-
tices were not considered in the least special or extraordinary and so needed no
mention. It is noticeable, however, that some of Leschetizky’s unnotated arpeg-
gios correspond with specific performance instructions he added to Chopin’s text.
For example, where he marked espressivo for the double-note sequence commenc-
ing at bar 10 (Fig. 3.33),67 arpeggios can clearly be heard. And for the molto espres-
sivo cantando section commencing in the middle of bar 70 (Fig. 3.34),68 arpeggios
help to delineate the overlapping compound melodies in the right hand.
Like Corri, Leschetizky’s use of such expressions appears to have implied the

Figure 3.33 Chopin Nocturne Op. 27 No. 2, bar 10, ed. Leschetizky.

Chopin, “Nocturne Op. 27 No. 2,” ed. Leschetizky, 19.
Chopin, “Nocturne Op. 27 No. 2,” ed. Leschetizky, 24.
Unnotated A r peg g iation 137

Figure 3.34 Chopin Nocturne Op. 27 No. 2, bars 70 to 74, ed. Leschetizky.

application of unnotated arpeggiations, yet without his recording this could not
have been appreciated. In these cases, a literal interpretation of Leschetizky’s
notation would produce a style significantly different from his own.
Similar unnotated arpeggiations can be heard in the performances of
Leschetizky’s students Powell and La Forge. On his 1929 piano roll of the same
Nocturne, Powell arpeggiates various double-note chords in the right hand in bars
10 and 11. He does this in such a way that the lower notes anticipate the upper
notes; the latter are aligned with the corresponding notes in the left hand
(Fig. 3.35). And on the third quaver beat in bar 13, Powell makes exactly the same
arpeggiation as Leschetizky (Fig. 3.36). In a similar way to Leschetizky, Powell
arpeggiates the first double-note chord in the right hand in bar 33. This has the
effect of continuing the broken chord triplet figure that precedes it (Fig. 3.37 ;
Audio Ex. 3.17 ), with the lower note aligned with the corresponding note in
the left hand. Between bars 71 and 74, Powell delineates between the multiple

Figure 3.35 Chopin Nocturne Op. 27 No. 2, bars 11 and 12, Powell, piano roll
recording, 1929 (Audio Ex. 3.15 ).
138 off the record

Figure 3.36 Chopin Nocturne Op. 27 No. 2, bar 13, Powell, piano roll recording, 1929
(Audio Ex. 3.16 ).

melody lines in the right hand by arpeggiating at the point where they overlap.
This is similar to Leschetizky, but at the beginning of bars 71 and 73 spreading
occurs, contrary to the norm, from the upper note downward to the lower note
(Fig. 3.38 ; Audio Ex. 3.18 ).
Unnotated arpeggiations are heard less frequently in La Forge’s 1912 record-
ing of the same Nocturne, but where he does arpeggiate, the expressive quality is
clear. Like Leschetizky and Powell, he arpeggiates the first chord in the right hand
of bar 33 (Fig. 3.37 ). He also arpeggiates the first chord in the right hand of bar
56 (Fig. 3.39 ; Audio Ex. 3.19 ), and the chord in the right hand at the begin-
ning of the second half of bar 61 (Fig. 3.40 ; Audio Ex. 3.20 ). And for the
compound melodic sequence in the right hand between bars 71 and 74, he arpeg-
giates the first chord of bars 71 and 73 (Fig. 3.41 ; Audio Ex. 3.21 ).
Rosenthal made unnotated arpeggiations much less frequently in his 1936
electrical recording of the same Nocturne, relying more on dislocation as an
expressive device. He does, however, arpeggiate the first chord in the right hand
of bar 33 and the third chord in the right hand of bar 37, in a similar manner to
Leschetizky. And during the section commencing at bar 70, he makes one arpeg-
giation in which the upper voice repeats in the middle of bar 72.
Other pianists such as Pachmann and Paderewski also make unnotated arpeg-
giations. In his 1915 recording of Chopin’s Nocturne Op. 9 No. 2, Pachmann
arpeggiates particular chords in the left hand at poignant moments, or to fill out
the texture when there is a significant broadening of the tempo. This is clearly
the case at the beginning of bars 2 and 4 and other similar places (Fig. 3.42 ;
Audio Ex. 3.22 ).
In his 1930 electrical recording of the same work, Paderewski also makes unno-
tated arpeggiations in the left hand in a similar way to Pachmann. During the
chordal passage at bar 12, he intersperses dislocation with very tight arpeggia-
tion, giving variety to the chromatic chordal progression (Fig. 3.43 ; Audio Ex.
3.23 ). And in his 1911 recording of Chopin’s Nocturne Op. 15 No. 1, Paderewski
Unnotated A r peg g iation 139

makes continuous unnotated arpeggiations in the left hand, which helps to bring
out the tenor-voice melody (Fig. 3.44 ; Audio Ex. 3.24 ).

Truth or Different Truth

Brée’s instructions (albeit incomplete) accord with some of Leschetizky’s arpeg-
giation practices. Written references—often extremely brief—concerning the
practices of other pianists, raise serious doubts as to what occurred in reality.
Brahms is a good case in point. Robert Pascall quotes the English pianist Florence
May (1845–1923) as having learned during her lessons with Brahms in 1871 that
“he particularly disliked chords to be spread unless marked so by the composer.”69
An examination of a larger section of May’s report, however, provides a different
slant on the matter:

Whatever the music I might be studying, however, he would never allow

any kind of “expression made easy.” He particularly disliked chords to be
spread unless marked so by the composer for the sake of special effect.
“No arpège,” he used invariably to say if I unconsciously gave way to the
habit, or yielded the temptation of softening a chord by its means.70

It seems that Brahms was attempting to curtail some careless habit in May’s play-
ing. But there is little reason to believe that his paraphrased words necessarily
reflected his personal practice. To judge from contemporary accounts, Brahms
certainly made frequent unnotated arpeggiations. Rosenthal recalled that he
“arpeggiated all chords.”71 And Brahms was severely criticized for the “incessant
spreading of chords in the slower tempos” after a performance of his Piano
Concerto No. 1 in 1865.72 In 1929, the revered musicologist and pianist Sir Donald
Tovey (1875–1940) shed some light on Brahms’s attitude to unnotated arpeggia-
tion, though from the point of view of string playing. Tovey was a close friend and
musical associate of Joachim and played the piano part of Brahms’s Piano Quintet
in F Minor Op. 34 with the Joachim Quartet in 1905. According to him, Brahms
took it for granted that string chords marked pizzicato would be arpeggiated, so
there was no need for him to give a special indication of this. Referring to the piz-
zicato violin chords at the beginning of the development of the first movement of

Robert Pascall, “Playing Brahms: A Study in 19th-Century Performance Practice,” Papers in
Musicology vol. 1 (1991): 18.
May, The Life of Johannes Brahms, vol. 1, 18.
Hudson, Stolen Time, 333.
[H. Kroenlein], “Konzertbericht” Karlsruher Zeitung (Nov. 9, 1865); in Gerhard, “Willkürliches
Arpeggieren,” 123: “das unablässige Brechen der Akkorde bei langsameren Tempi.”
140 off the record

Brahms’s Sonata for Violin and Piano in G major Op. 78, Tovey explains that
Brahms “did not anticipate a time” when violinists, who would normally have
arpeggiated these chords, felt compelled to play them tightly so that they sound
like “dry clicks.”73
Due to poor sound quality, it is difficult to be certain whether Brahms arpeg-
giated any of the chords in his cylinder recording of his Hungarian Dance No. 1.
To my ears, it sounds like he is making very rapid arpeggiations of the dotted
crotchet chords in bars 13 and 14 (the bars at which the recording commences),
and also possibly the dotted crotchet in bar 17. On the other hand, he may have
chosen to arpeggiate less frequently than usual considering the strongly accented
character of the Hungarian Dance.
The seeming contradiction between Brahms’s advice to May and his practice
highlights the difficulty that often presents itself when interpreting the written
word. We do not know the circumstances that led him to censure May. Presumably
he felt that she had developed a bad habit—of unconsciously arpeggiating—that
contributed overly to a feeling of “expression made easy.” In the end, the apparent
discrepancy might arise simply from too literal an interpretation of May’s account,
taken out of context. Brahms may have made such comments in passing for May’s
benefit alone, and did not expect them to be taken literally or applied universally.
As we will see later, several pianists whom he influenced and of whose playing he
approved certainly made unnotated arpeggiations.
A parallel anomaly arises when considering Chopin’s apparent dislike of unno-
tated arpeggiation. His student Karol Mikuli (1819–97) explains that “for playing
double notes and chords, Chopin demanded that the notes be struck simultane-
ously; breaking was allowed only where the composer himself had specified it.”74
In Jean-Jacques Eigeldinger’s opinion, Chopin was reacting “to the sentimental-
izing fashion of spreading this or that chord or beat” an abuse that he claims
reached its peak at the turn of the twentieth century.75 This may or may not be
true. As we have seen, in Chopin’s era and before, the sources are critical of these

Donald F. Tovey, “Brahms,” Cobbett’s Cyclopedic Survey of Chamber Music, ed. Walter. W. Cobbett
with supplementary material ed. Colin Mason, 2nd ed. (London: Oxford University Press, 1963), vol.
1, 165–66. Tovey further explains that it never occurred “to an experienced composer to provide such
signs” until the composer Max Reger “discovered that string players had forgotten their own instincts”
in this regard. Reger apparently marked arpeggio signs in front of “every three-part and four-part [piz-
zicato] chord to prevent them from choking it.”
Karol Mikuli, “Vorwort” to Frederic Chopin’s Pianoforte-Werke, ed. Karol Mikuli (Leipzig,
Germany: Kistner, 1880), vol. 1, 4. The following translation appears in Karol Mikuli, “Introductory
Notes” to Frederic Chopin’s Complete Works for the Piano (New York: Schirmer, 1895), unpaginated [2]:
“For paired notes and chords he exacted strictly simultaneous striking of the notes, an arpeggio being
permitted only where marked by the composer himself.”
Jean-Jacques Eigeldinger, Chopin vu par ses élèves (Neuchatel, France: 1970); trans. Naomi
Shohet, Krysia Osostowicz, and Roy Howat as Chopin: Pianist and Teacher—As Seen by His Pupils, 3rd
ed., ed. Roy Howat (Cambridge: Cambridge University Press, 1986), 108.
Unnotated A r peg g iation 141

practices when used incorrectly, but support them when used judiciously and with
taste. And there seems little justification for assuming that the practice peaked at
the turn of the twentieth century. The unnotated arpeggiations heard in the play-
ing of Leschetizky, Powell, Paderewski, and others are representative of the end of
an ongoing tradition that very probably reached a peak much earlier.
Indeed, Chopin may have disliked the tendency to arpeggiate everything to the
detriment of the musical effect. His seemingly draconian words may have been for
the benefit of students or amateurs until they developed a sophisticated sense of
how to apply arpeggios. Or he may simply have made such comments in passing,
or in the heat of a moment, not expecting that his words would be followed verba-
tim or applied without exception. It is also possible that his concern was only
about arpeggiations made inartistically, perhaps too slowly, awkwardly or notice-
ably. Like Thalberg and others, he may readily have accepted very swift or tight
spreading of notes in chords, which he did not regard as problematic at all. In this
respect, Gerhard has noted the possibility that during the nineteenth century,
arpeggiation—even when notated—sounded different than today. Perhaps it
sounded more like “a quick ‘run up’ the chord.”76
It is arguable, in any case, whether Chopin was himself meticulous in his
notation of arpeggios. Eigeldinger notes that Chopin’s annotations in a score of
one of his other students—Camille Dubois—reveals an arpeggiation sign, in
bar 7 of the Prelude Op. 28 No. 6, that did not appear in the original French
edition (Fig. 3.45 ).77 This sign bears remarkable visual resemblance to others
that appear in the Nocturne Op. 32 No. 1 at bar 9 and bar 32 (Figs. 3.46 and
3.47). Eigeldinger also states that “there are places in Chopin’s compositions
where the spreading of chords, though necessary, is not always clearly specified,
probably bearing in mind the different hand spans of pianists.”78 He gives as an
example the section commencing at bar 25 of Chopin’s Nocturne Op. 48 No. 1
(Fig. 3.48). Although hand spans may have been the criterion for the somewhat
erratic notation here, it seems more likely to me that the arpeggios (notated in
various ways) are ones that Chopin particularly wanted to hear. This may not have
been intended to exclude the addition of other arpeggios according to the taste of
the individual, his or her hand span, the acoustics of the hall, or the type of piano
being played. In the end, it is dangerous to assume from Mikuli’s comment alone
that unnotated arpeggiation was not part of Chopin’s expressive practice.
Pugno’s 1903 recordings prove interesting in this regard. He studied with
another of Chopin’s students, Georges Mathias (1826–1910), so it is reasonable
to expect that at least some of his practices derived from Chopin. Pugno makes

Gerhard, “Willkürliches Arpeggieren,” 129: “Vielleicht ging es viel öfter um ein kurzes ‘Anreisen’
der Akkorde.”
Eigeldinger, Chopin, 108.
Eigeldinger, Chopin, 108.
142 off the record

Figure 3.46 Chopin Nocturne Op. 32 No. 1, bars 8 and 9.

Figure 3.47 Chopin Nocturne Op. 32 No. 1, bar 32.

unnotated arpeggiations in Chopin’s Nocturne Op. 15 No. 2 and Valse Op. 34 No.
1. In the Nocturne, he arpeggiates several of the chords in the left hand between
bars 6 and 9, though only one of these is notated (bar 8) in his edition of the work
(Fig. 3.49).79 These sound before the beat and create harped effects that enhance
the frequent and poignant changes of harmony (Fig. 3.50). He also makes similar
arpeggiations in bars 15 and 21. At the beginning of bars 5 and 13, Pugno doubles
the bass notes at the lower octave and arpeggiates them so that the lower
note comes before the beat (Fig. 3.51a and Fig. 3.51b ; Audio Ex. 3.26 ).
These sound similar in effect to his arpeggiated octaves in the left hand at bars
20 and 21 (Fig. 3.52 ; Audio Ex. 3.27 ).
Considering this very noticeable use of unnotated arpeggiation, one might
justifiably expect some mention of it in Pugno’s remarks about the work’s perfor-
mance, but this is not the case. Not once does he recommend the addition of
arpeggios. And this in spite of the fact that for Chopin’s notated arpeggio at the
end of bar 8, he advises the player to “spread out the chord very broadly from the
first note in the bass to the A which begins the melody again.”80

Chopin, “Nocturne Op. 15 No. 2,” ed. Pugno, 67.
Pugno, The Lessons, 67.
Figure 3.48 Chopin Nocturne Op. 48 No. 1, bars 25 to 32.

Figure 3.49 Chopin Nocturne Op. 15 No. 2, bars 6 to 9, ed. Pugno.

Figure 3.50 Chopin Nocturne Op. 15 No. 2, bars 6 to 9, Pugno, recorded 1903 (Audio
Ex. 3.25 ).

144 off the record

Figure 3.53 Chopin Valse Op. 34 No. 1, bars 17 to 20, Pugno, recorded 1903 (Audio
Ex. 3.28 ).

In the Valse, Pugno’s unnotated arpeggiations enhance the expression of the

melody written in sixths in the section commencing at bar 17 (Fig. 3.53). Chopin,
in any case, notated two grace-note arpeggios at the beginning of bars 26 and 29
(Fig. 3.54 ),81 and Pugno seemingly extended this to other places. Here, arpeg-
giation helps to achieve a feeling of dolce e cantando as marked by Pugno. Although
there is no direct instruction to arpeggiate, there may be a veiled message to do so
when he advises that the theme “should convey an impression of languid grace”
and that that the “delicate shades and meanings absolutely forbid the dryness of
too precise a rhythm.” He goes on to describe the theme poetically as “a lissome
lady at a ball, whose movements in her long trained gown evoke all the charm of
the Polish dance measure.”82 At the return of the theme at bar 80, Pugno again
makes an arpeggiation. Here he notes that “we come back to the charm, to the
delicate shading, to the wayward delays, the tender coquetry.”83 Clearly, unno-
tated arpeggiation, among other things, must have been intended to help achieve
some of these expressive feelings.
Quite obviously, Pugno’s verbal annotations do not convey to us what he actu-
ally did. But he surely expected musicians of his era to understand the hidden
implications in his words, in the same way that for Corri (100 years earlier), cer-
tain terms such as dolce and espressivo implied the use of arpeggiation. But such
terminology no longer carries those implications. Without Pugno’s recordings, his
manner of adding arpeggios would be unknown and a reliance on Pugno’s written
text alone would result in a style quite different from his own.
Pugno seems to have adopted a different attitude to the addition of arpeggios
in his recording of Chopin’s Berceuse Op. 57. Here, he studiously avoids any chord
spreading, apart from places where the interval is too wide to play simultaneously
or where he occasionally doubles the bass note at the lower octave. In his annota-
tions to the work, there are no obvious comments about this. However, he

Pugno, The Lessons, 8
Pugno, The Lessons, 8.
Pugno, The Lessons, 10.
Unnotated A r peg g iation 145

describes the peculiar characteristic of the piece with its “insistent (volue) monot-
ony of the bass, which necessarily implies a sameness of mood in the right hand,
an almost complete neutrality in the left.” He warns, “No emotion is to be sought
for.”84 Again, for readers today, this instruction implies perhaps nothing more
than simplicity in the interpretation. But Pugno’s words probably signified much
more: another veiled message perhaps to avoid expressive devices such as unno-
tated arpeggiation.

Added Arpeggio Signs in Musical Editions

Late-nineteenth-century revised or instructive editions often reveal the addition
of arpeggio signs that augment those originally notated by the composer. But in
the cases that I have surveyed, the frequency of these additions rarely if ever
approach the frequency of unnotated arpeggiations heard in early recordings.
Here, the disparity between musical notation and actual practice is most striking.
For example, in his critical edition of Chopin’s Nocturnes,85 the pianist and editor
Karl Klindworth (1830–1916) notates several arpeggio signs not in the original.
These appear to fulfill specific expressive functions as noted in Table 3.1.86
Klindworth’s added arpeggio signs in Chopin’s Nocturne Op. 27 No. 2 are particu-
larly interesting because they afford comparison with audible examples. His only
additional signs, in the right hand in bars 11, 13, and 33 (Figs. 3.55, 3.56, and
3.57),87 certainly bear a resemblance to those made by Leschetizky, Powell, La
Forge, and Pachmann cited earlier. However, these pianists make much more fre-
quent arpeggiations. Although there is a marked disparity between Klindworth’s
edition and the early recordings, there is nothing to suggest that Klindworth did
not expect more arpeggios to be added. Here, and in other places, he may simply
have marked those that he thought absolutely necessary, leaving others to the
taste and skill of the performer.
Klindworth’s edition is valuable for the study of late-nineteenth-century per-
forming practice. It confirms his need for more arpeggiation than originally
notated by Chopin and it provides notated evidence of some of the types of arpeg-
giation preserved in some early recordings. Yet, a strict adherence to his notation
would produce significantly less arpeggiation than was made by many pianists of
the era.

Pugno, The Lessons, 26.
Frédéric Chopin, Oeuvres complètes : revues, doigtées et soigneusement corrigées d’aprés les éditions
de Paris, Londres, Bruxelles et Leipsic [sic], ed. Karl Klindworth (Berlin: Bote & Bock, c. 1890).
Chopin, Oeuvres complètes, ed. Klindworth, 57–60 and 73–75.
Chopin, Oeuvres complètes, ed. Klindworth, 29–30.
Table 3.1 Chopin Nocturnes Op. 55 No. 1 and Op. 72 No. 1, Ed. Klindworth

Work Notated Arpeggio Sign Possible Intended Effect

Chopin Chords in right and left hands and Enhancing a softened effect for
Nocturne grace-note octave doubling in left the end of the section and
Op. 55 hand at the beginning of bar 48 marking the boundary before
No. 1 (Appendix A, Fig. A1 ) proceeding with the more
impassioned più mosso

Chords in right and left hands at the Enhances the poignancy of the
beginning of bar 52 (Appendix A, grace note D natural forming a
Fig. A2 ) dissonant ninth with the C minor

Chord in the left hand at the Same effect as bar 52; note that
beginning of bar 56 (Appendix A, Chopin marked an arpeggiation
Fig. A3 ) in the right hand

Double-note interval formed by the Delineates the entry of the

compound melody on the first beat of second voice
bar 58 (Appendix A, Fig. A4 )

Double-note interval at bars 59 and Distinguishes between the

63 (Appendix A, Figs. A5a and compound voices; in the case of
A5b ) bar 63, the arpeggio may mark
the boundary between the
decrescendo that precedes and
the crescendo that follows

Chord in the left hand at bar 69; Similar to Brée’s description of

note that the chord in the right this technique, creating energy
hand remains unarpeggiated without harshness appropriate to
(Appendix A, Fig. A6 ) the character; here, the chords
form an energetic pivot point
between two sections

Chord formed on the restruck Delineates the end of the

pedal point F at bar 87 decrescendo in the right hand and
(Appendix A, Fig. A7 ) the start of a crescendo

Table 3.1 Cont’d

Work Notated Arpeggio Sign Possible Intended Effect

Chopin Octave on first beat in the right hand Distinguishes between lower
Nocturne at bar 10 (Appendix A, Fig. A8 ) voice and the newly introduced
Op. 72 No. 1 upper voice

Arpeggiation of the third last Clarifies the reentry of the upper

double-note triplet chord in the voice that has been momentarily
right hand at bar 12 (Appendix A, silent, and also propels the
Fig. A9 ) momentum forward into the
next bar

Double-note chord on the third Emphasizes the poignancy of the

crotchet beat in the right hand at bar dissonant ninth formed with the
14 (Appendix A, Fig. A10 ) bass

Chord in the right hand at the Emphasizes and enhances the

beginning of bar 26 (Appendix A, effect of strong/weak at the
Fig. A11 ) feminine cadence

Last octave of a pattern of four in the Gives heightened significance and

right hand at bar 41 (Appendix A, helps to mark the subsequent
Fig. A12 ) leap of the upper voice down an

Third last triplet-quaver chord in the Same effect as at bar 12,

right hand at bar 41 (Appendix A, propelling the movement forward
Fig. A12 )

First chord in the right hand at bar Enhances the resolving or

46; here, there is an added E as well softening effect of the chord that
as an added grace-note octave is preceded by a common cadence
doubling in the left hand formula
(Appendix A, Fig. A13 )

Final chord in the left hand at Enhances the calando effect

bar 57 (Appendix A, Fig. A14 )

148 off the record

Figure 3.55 Chopin Nocturne Op. 27 No. 2, bar 11, ed. Klindworth.

Figure 3.56 Chopin Nocturne Op. 27 No. 2, bar 13, ed. Klindworth.

From the late-eighteenth century onward, the addition of arpeggio signs to the
canon of Classical masterworks became increasingly common. Klindworth was
preceded by a host of celebrated musicians, including Muzio Clementi (1752–
1832), whose publications provide an intriguing scenario. He published many
works including sonatas, sonatinas, and studies. The first editions of these are
noticeably devoid of arpeggio signs. But, as Gerhard points out, in subsequent
editions of the same works he regularly added arpeggio signs.88 Czerny, we have
already seen, added arpeggio signs to Beethoven’s music. He notated an arpeggio
for the first chord of the third movement of the Piano Sonata Op. 7 (Fig. 3.58),89
as well as one for the first chord of the Piano Concerto Op. 58. In this regard,

Gerhard, “Willkürliches Arpeggieren,” 125–26.
Czerny, Art of Playing, 38.
Unnotated A r peg g iation 149

Figure 3.57 Chopin Nocturne Op. 27 No. 2, bars 32 and 33, ed. Klindworth.

Figure 3.58 Beethoven Piano Sonata Op. 7, third movement, ed. Czerny.

either Czerny remembered Beethoven’s practice or thought that an arpeggio was

appropriate for the enhancement of Beethoven’s indication p dolce. And, though
not mentioned, it is likely that he expected this type of arpeggiation to continue
at other appropriate moments in the music.
The addition of arpeggio signs by the pianist and editor Cipriani Potter (1792–
1871) are also very intriguing. Between 1822 and 1859, Potter was associated
with the Royal Academy of Music in London, first as a teacher and eventually as
its principal. In 1818, he studied with Beethoven in Vienna. From 1819 onward,
he gave the English premiere of several of Mozart’s and Beethoven’s piano concer-
tos. He considered the Viennese classics to be “the proper foundation of modern
music,”90 and has been credited with establishing the English School of piano play-
ing. Potter’s editions of Mozart’s piano sonatas began to be published around
1836, and were reissued by Novello in 1851. They reveal many added arpeggios,
some of which in the first and second movements of Mozart’s Sonata K 310 are
listed in Table 3.2.91 Potter’s 1854 edition of Beethoven’s Sonata Pathétique Op. 13

Nicholas Temperley, “London and the Piano 1760–1860,” The Musical Times, vol. 129 (1988): 292.
Wolfgang A. Mozart, An Entirely Honourable and Complete Edition of the Pianoforte Works with and
without Accompts. of this Celebrated Composer, ed. Cipriani Potter (London: Novello, c. 1857), 107–19.
Table 3.2 Mozart Sonata K 310, Ed. Potter

Work Notated Arpeggio Sign Possible Intended Effect

Mozart Sonata Compound melody in the right Differentiation of two voices

K 310: First hand at bar 17 (Appendix B, when one enters an octave
movement— Fig. B1 ) higher; note the similarity
Allegro between this and Brée’s advice
maestoso to use arpeggiation in
polyphonic writing

Chord in the left hand at bar 57 Enhances sf accent and fills the
(Appendix B, Fig. B2 ) sound of the bar

Minim chords in the right hand Adds to the bravura character by

during the section from bar 58 to creating a strummed effect; it is
bar 68 inclusive (Appendix B, obvious that this is to be
Fig. B2 ) continued throughout even
though the arpeggio signs do
not continue; Potter prompts
the player at bar 63

Alternating chords in the left hand Enhances the effect of strong

from bars 118 to 119 and weak
(Appendix B, Fig. B3 )

Chord in the left hand at bar 126 Enhances the sforzando effect
and the right hand at bar 127 and fills out the bar
(Appendix B, Fig. B4a and B4b )
Second Chord on the first beat marked Gives emphasis to the six-four
movement— across left and right hands at bar 2 harmony on the first beat that
Andante (Appendix B, Fig. B5 ) resolves to an unarpeggiated
cantabile con five-three harmony on the
espressione second beat; this occurs at many
other feminine cadences
throughout the movement

Chord on the third beat marked Enhances the effect of the fp

across the left and right hand in marking
bar 2 even though an anticipatory
arpeggio was already notated by
Mozart; note that Potter adds
notes to Mozart’s original chord
(Appendix B, Fig. B5 )

Unnotated A r peg g iation 151

Table 3.2 Cont’d

Work Notated Arpeggio Sign Possible Intended Effect

First chord in the right hand in bar Enhances the poignancy of the
3 (Appendix B, Fig. B5 ) seventh chord

Chord in the right hand at bar 27 Enhances the accent and fills
and corresponding material in bar out the sound in the bar
82 (Appendix B, Fig. B6 )

Chords on the first beat in the Emphasizes the strong beats

right hand at bar 40 and bar 42 and poignant dissonant
(Appendix B, Fig. B7 ) harmonies

also provides examples of added arpeggio signs (Table 3.3).92 Particularly interest-
ing is the arpeggio sign for the opening chord of the first movement, presumably
intended to enhance the dramatic effect (Fig. 3.59).93 Significantly, this bears
resemblance with Czerny’s added sign for the opening chord of Beethoven’s
Fourth Piano Concerto. Another notable example is seen in bar 9 of the second
movement, where Potter indicates arpeggio signs for the main beats of the bar
(Fig. 3.60).94 No doubt Potter intended arpeggiation here to enhance the feeling
of cantando and con molto espressivo, expressions that he added to Beethoven’s
text. And it is probable that he intended the effect to carry on during the following
bars of the phrase.
In his 1854 edition of Beethoven’s Moonlight Sonata Op. 27 No. 2, Potter marks
the octaves in the right hand of bars 6 and 7 in the first movement to be arpeg-
giated, thereby separating the melody note from the inner-voice accompaniment
(Fig. 3.61).95 Again, this type of arpeggiation was likely intended to continue
throughout the movement. Potter also marks an arpeggio for the interval of a
ninth in the right hand at bars 52 and 54 (Fig. 3.62).96 The simplest and obvious
explanation is that this was for the benefit of those for whom the interval was too
wide. On the other hand, he may have marked these specially so that those who
could strike the interval simultaneously did not do so, in order to give particular

Ludwig van Beethoven, “Sonata Pathétique for the Piano Forte,” ed. Cipriani Potter (London: Mills,
1854), 1–17.
Beethoven, “Sonata Pathétique,” ed. Potter, 1.
Beethoven, “Sonata Pathétique,” ed. Potter, 9.
Ludwig van Beethoven, “Moonlight Sonata Op. 27 No. 2,” ed. Cipriani Potter (London: Mills,
1854), 2.
Beethoven, “Moonlight Sonata Op. 27 No. 2,” ed. Potter, 4.
152 off the record

Table 3.3 Beethoven Sonata Pathétique Op. 13, Ed. Potter

Work Unnotated Arpeggiation Possible Effect

Beethoven Chord on the first beat across Creates heightened dramatic

Sonata left and right hand at bar 3 effect; note that the chord at the
Pathétique Op. (Appendix C, Fig. C1 ) beginning of bar 2 is not
13: First arpeggiated, perhaps for a
movement— contrasting effect

Chord on the first beat of bar Creates heightened dramatic

133; here the left and right effect; note that the sign is used
hands have separate arpeggio only once in this section, perhaps
signs (Appendix C, Fig. C2 ) simply to remind the player to
make arpeggios
Second Chord on the first beat across the Emphasis of the six-four chord
movement— left and right hands at bar 16 that resolves to an unarpeggiated
Adagio (Appendix C, Fig. C3 ) five-three chord
Chord on the first beat across the Enhances singing quality; it is
left and right hands at repeat of surprising that Potter did not mark
the opening material at bar 29 the same arpeggio sign at the
(Appendix C, Fig. C4 ) opening

Chords on the first beat across Enhances singing quality; again

the left and right hands at bars there is intermittent use of
52 (Appendix C, Fig. C5 ) and arpeggio signs
bar 59 (Appendix C, Fig. C6 )
Third Long chords across the left and Creates heightened dramatic
movement— right hands at bars 18 and 22 emphasis for the marking fp and
Rondo-Allegro (Appendix C, Fig. C7 ) fills out the sound of the bar
non tanto
Octave in the right hand at bar Delineates the compound voices
75 (Appendix C, Fig. C8 )

expression to the extraordinarily poignant harmony. Interestingly, later in the

century Lebert and Faisst marked an arpeggio sign next to the chord in the right
hand (forming the interval of a ninth) in bar 4 and similar places in the first
movement of Beethoven’s Sonata Op. 26. They advise that these arpeggios are
intended “for hands for which the stretch is too wide, and which therefore, should
Unnotated A r peg g iation 153

Figure 3.59 Beethoven Sonata Pathétique Op. 13, ed. Potter, first movement, bar 1.

Figure 3.60 Beethoven Sonata Pathétique Op. 13, ed. Potter, second movement, bar 9.

momentarily touch the lowest tone, but hold the highest tones to their full
value.”97 Nevertheless, this is no reason to believe that they prescribed arpeggia-
tion only to aid with these types of technical difficulty. As we have seen, for the
interval of a ninth in bar 8 and other similar places in the first movement of

Ludwig van Beethoven, “Sonata Op. 26,” Sonatas and Other Works for the Pianoforte, Instructive
Edition of Classical Pianoforte Works, ed. by Immanuel von Faisst and Sigmund Lebert (1871), part 3,
vol. 2, trans. J. H. Cornell (Stuttgart, Germany: Cotta, 1891), 27.
154 off the record

Figure 3.61 Beethoven Moonlight Sonata Op. 27 No. 1, ed. Potter, first movement,
bars 6 and 7.

Figure 3.62 Beethoven Moonlight Sonata Op. 27 No. 1, ed. Potter, first movement,
bars 52 to 54.

Beethoven’s Sonata Op. 27 No. 2, they encouraged arpeggiation so that the melody
notes in chords comprised of peculiar tone combinations could stand out from
notes of the accompaniment.98
Returning to Potter, it is not always possible to appreciate why he indicated
arpeggio signs at some moments and not at others. Sometimes they appear to
serve as prompts to remind the player to arpeggiate or simply to provide an exam-
ple of when to do so. Nevertheless, that they exist is proof of the importance of
arpeggiation at this time.
Potter was not the only editor to mark arpeggiations in the first and second
movements of Beethoven’s Sonata Pathétique. In an English edition of 1861,
William Dorrell follows the tradition, perhaps inculcated by Potter, by marking an
arpeggio for the opening chord of the first movement and the corresponding

See page 106.
Unnotated A r peg g iation 155

Figure 3.63 Beethoven Sonata Pastorale Op. 28, ed. Potter, first movement, bars 1 to 3.

chord at bar 133.99 He also marks an arpeggio for the poignant chord on the first
beat of bar 3 in Beethoven’s Sonata Pastorale Op. 28 (Fig. 3.63).100
A further example is preserved in the Magazine of Music: Pictorial Pianoforte
Tutor (Leipzig, 1891), where the anonymous editor marked an arpeggio sign for
the double-note chord (A flat–B flat) in the right hand at bar 11 of the second
movement of Beethoven’s Sonata Pathétique (Fig. 3.64 ).101 Like Potter, this
was probably to ensure that the poignant harmony was indeed arpeggiated.
Significantly, the editor also marked an arpeggio sign for the first chord—a
dissonant seventh chord—in the right hand at bar 13 (Fig. 3.64 ).
Another enlightening example preserved in the Magazine of Music reveals that
in certain cases, poignant harmonies were required to be arpeggiated. In the fol-
lowing excerpt (Fig. 3.65) from the third movement—the Marcia Funebre sulla
morte d’un eroe from Beethoven’s Sonata Op. 26—an arpeggio sign is marked for
the chord in the right hand at the beginning of bar 25. The anonymous editor

Figure 3.65 Beethoven Sonata Op. 26, third movement, bars 24 and 25, Magazine of

Ludwig van Beethoven, The Sonatas of Beethoven for the Pianoforte, ed. William Dorrell (London:
Boosey, 1861), vol. 1, 100–102.
Beethoven, “Sonata Pastorale Op. 28,” ed. Dorrell, vol. 1, 188.
Ludwig van Beethoven, “Adagio from ‘Sonata Pathétique,’” Magazine of Music (1891): part 2, 130.
156 off the record

Table 3.4 Beethoven Sonatas Op. 2 No. 1 and Op. 22, Ed. Dorrell

Work Notated Arpeggio Sign Possible Intended Effect

Beethoven’s Long chord across the left and Enhances the effect of ff and its
Sonata Op. 2 right hands at the penultimate resolution on to the unarpeggiated
No. 1: First bar of the exposition—bar 47 chord in bar 48, and fills out the
movement— (Appendix D, Fig. D1 ) sound of the bar
Long chords across the left and Similar effect to above
right hands at bars 146 and 148
(Appendix D, Fig. D2 )
Beethoven’s Chord on the first beat in the Enhances the expressive effect of the
Sonata Op. 22: right hand at bar 13 E flat major harmony; note that this
Second (Appendix D, Fig. D3 ) arpeggiation is not marked at other
movement— similar places, such as bars 15, 19,
Adagio con and 21, but a similar arpeggio may
molto well have been expected
First chord in the right hand at Similar effect to above
bar 58 (Appendix D, Fig. D4 )

recommends in a footnote “to play the C natural with the second finger and to
spread this chord, in order to make the sublime beauty of this change into major
fully heard and appreciated.”102
Other sonatas by Beethoven edited by Dorrell provide further examples of
added arpeggio signs (Table 3.4).103 Often Dorrell notated an arpeggio sign only
on the first appearance of recurring musical material. Like Potter, his marking may
have been intended as a prompt to continue arpeggiating where appropriate.
Hans von Bülow’s annotations to Beethoven’s late sonatas in his “Instructive
Edition” (1871) also provide important evidence of added arpeggio signs seem-
ingly to enhance expressivity. For example, Bülow marked an arpeggio sign next
to the pungent and widely spaced chord in the right hand on the first beat of bar
248 in the first movement of the Hammer Klavier Sonata Op. 106 (Fig. 3.66).104
Furthermore, in bars 9, 10, and 58 in the “Adagio espressivo” sections of the first

Ludwig van Beethoven, “Marcia Funebre sulla morte d’un eroe,” Magazine of Music (1891):
part 2, 131.
Beethoven, The Sonatas of Beethoven, ed. Dorrell, 1–3 and 139–42.
Ludwig van Beethoven, “Sonata Op. 106,” Sonatas and Other Works for the Pianoforte, Instructive
Edition of Classical Pianoforte Works ed. by Hans von Bülow (1871), part 3, vol. 5, trans. J. H. Cornell
(Stuttgart, Germany: Cotta, 1891), 31.
Unnotated A r peg g iation 157

Figure 3.66 Beethoven Sonata Op. 106, first movement, bar 248, ed. Bülow.

movement of the Sonata Op. 109, Bülow marked arpeggios signs next to various
chords. These can be seen to emulate Beethoven’s own grace-note arpeggiated
first chord in bars 9 and 58 and therefore to sustain and enhance the feeling of
espressivo (Fig. 3.67).105 Here the link between espressivo playing and arpeggia-
tion is clear. It is probable that Bülow expected other chords in these sections to
be likewise arpeggiated. Perhaps one of the most significant references to the
importance of arpeggiation as a means of bringing out melody notes is found in
Bülow’s annotation to bars 157 and 158 in the second movement “Arietta” of the
Sonata Op. 111 (Fig. 3.68). At the climax of the section, Bülow marks arpeggios
for various chords in the right hand, with the following explanation:

Arpeggiating, which in the delivery of the classics we otherwise exclude

on principle, appears to us here necessary even for hands of greater span-
ning-capacity, in order to assist the upper voice to attain its full right to
most emphatic prominence. The player should, of course, beware of an
anticipation disturbing the purity of the harmony. The uppermost tone
may lag a little without disadvantage.106

Beethoven, “Sonata Op. 109,” Sonatas and Other Works, part 3, vol. 5, ed. Bülow, 75.
Beethoven, “Sonata Op. 111,” Sonatas and Other Works, part 3, vol. 5, ed. Bülow, 141: “Das
Arpeggiren, das wir beim Vortrage der Klavierklassiker sonst grundsätzlich ausschliessen, erscheint
uns hier, auch für Hände grösserer Spanungsfähigkeit nothwendig, um der Oberstimme zu ihrem
158 off the record

Figure 3.67 Beethoven Sonata Op. 109, first movement, bars 9 and 10, ed. Bülow.

Figure 3.68 Beethoven Sonata Op. 111, second movement, bars 157 and 158, ed.

It is fascinating to note here that, like so many others, Bülow seemed to regard
the use of arpeggiation as a modern device, not generally suitable for earlier rep-
ertoire. In a few other cases, and with particular reference to widely spaced chords,
Bülow steadfastly opposed arpeggiation if it hindered what he considered the req-
uisite effect. For example, for the left-hand chords spanning a tenth in bars 22
and following of the third movement of the Hammer Klavier Sonata, Bülow asked
for an avoidance of arpeggio “as much as possible,” explaining that here “the tone
colour should correspond to a pianissimo of trombones.”107 Similarly, for the left-
hand chords spanning a tenth in bars 134 to 136 of the same movement, he warns
against an involuntary arpeggio because this “would produce here an undignified
effect, and these various closing chords should impart to the charm of the whole
adagio an after-effect as lasting as possible.”108

vollen Rechte auf nachdrücklichste Hervorhebung zu verhelfen. Der Spieler hüte sich natürlich vor
einem die Reinheit der Harmonie trübenden Anticipiren. Der höchste Ton darf ohne Nachtheil sich
Beethoven, “Sonata Op. 106,” Sonatas and Other Works ed. Bülow, part 3, vol. 5, 43: “Ein
Arpeggiren dieser Accorde in der Dezimenlage ist nach Kräften zu vermeiden: Die Klangfärbung
muss dem Charakter eines Pianissimo von Posaunen entsprechen.”
Beethoven, “Sonata Op. 106,” ed. Bülow, part 3, vol. 5, 55: “Ein unfreiwilliges Arpeggiren ist im
Uebrigen nach Kräften zu vermeiden: es würde hier einen würdelosen Eindruck hervorbringen und
Unnotated A r peg g iation 159

Discrepancies between Written Texts and Practice

To what extent late-nineteenth-century editors considered their annotations to
be binding is an interesting question. In this respect, it is fascinating to compare
Reinecke’s own published arrangement of the Larghetto from Mozart’s Piano
Concerto K 537109 with his performance of it. Reinecke notates arpeggio signs
specifically where intervals are too widely spaced to play simultaneously, or where
he wishes large chords in both hands to be arpeggiated. At other places, such as
bars 1 to 8, no arpeggio signs are indicated (Fig. 3.69 ).110 However, Reineceke’s
1905 Welte-Mignon piano roll of the Larghetto reveals that in these bars and
many others, he introduced a significant number of unnotated arpeggiations
(Fig. 3.70). Often, these have the aural effect of sounding before the beat: the
melody notes give a sense of the pulse. There is also a very flexible attitude to the
alteration of the notes themselves. For example, Reinecke fleshes out the chords on
the first beat of bars 1 and 5, creating forward propulsion that is dramatic in effect.
Examples of Reinecke’s dislocation practices have been cited in chapter 2. Most
striking is his close interspersion of arpeggiations and dislocations, producing an
overall effect of almost continuous syncopation. As noted in chapter 2, this effect
is also evident in his circa 1905 Hupfeld piano roll of the same arrangement.
Of significance, too, is Reinecke’s interpretation of the portato notation (which
he added to Mozart’s text), particularly noticeable in the main theme. His arpeg-
giations here (sporadic though they are) are certainly in keeping with Moscheles’s
principle (discussed earlier). Note, for example, the left-hand arpeggiations of
chords marked with portato in bars 2, 3, and 7. Bars 9 and 10 also provide exam-
ples of Reinecke’s unnotated arpeggiations (Fig. 3.71a and Fig. 3.71b ).111
Here, the effect of arpeggiation before the beat in the left-hand figurations is
clearly audible (Fig. 3.72). In the section from bars 15 to 19, Reinecke embellishes
the melody and arpeggiates almost every chord in the accompaniment contrary to
his own notation (Fig. 3.73 ).112 Here, too, the arpeggiations sound before a
notional beat punctuated by the melodic material in the right hand (Fig. 3.74).
Reinecke arpeggiates the chords across the left and right hands on the first beats
of bars 28, 29, and 30 (Fig. 3.75 ).113 Here, the arpeggiations are swift and start
before the beat (Fig. 3.76). In addition, Reinecke makes several significant changes
to his notation. Similar arpeggiations are also made between bars 36 and 41.

diese verschiedenen Schlussaccorde müssen dem Zauber des ganzen Adagio eine möglichst dauernde
Nachwirkung verleihen.”
Mozart, Larghetto arr. Reinecke, 2–7.
Mozart, Larghetto arr. Reinecke, 2.
Mozart, Larghetto arr. Reinecke, 2.
Mozart, Larghetto arr. Reinecke, 3.
Mozart, Larghetto arr. Reinecke, 3.
Figure 3.70 Mozart Larghetto arr. Reinecke, bars 1 to 8, Reinecke, piano roll, 1905
(Audio Ex. 3.29 ).

Figure 3.72 Mozart Larghetto arr. Reinecke, bars 9 and 10, Reinecke, piano roll, 1905
(Audio Ex. 3.30 ).

Figure 3.74 Mozart Larghetto arr. Reinecke, bars 15 to 19, Reinecke, piano roll, 1905
(Audio Ex. 3.31 ).

Figure 3.76 Mozart Larghetto arr. Reinecke, bars 29 to 30, Reinecke, piano roll, 1905
(Audio Ex. 3.32 ).

162 off the record

Yet, as noted in chapter 2, none of these are mentioned in his book about the
revival of Mozart’s piano concertos.
Clearly, Reinecke the pianist did not adhere to the notation of Reinecke the
editor and arranger. The implications here are significant. His notation does not
preserve the nature or frequency of arpeggiation intrinsic to his style. In fact,
a strict adherence to his notation would produce a result entirely different from
his own.
This discrepancy between notation and actual practice has strong historical
precedents. Reinecke, like Corri and earlier writers, may have employed verbal
expressions like con espressione, con anima, and dolce—and by inference larghetto—
to convey that, in addition to those marked in the score, other expressive
arpeggios should frequently be added. We might therefore consider Reinecke
a true representative of a tradition that had already been ongoing for a consider-
able period.
In 1904, Reinecke was acknowledged by the Zeitschrift für Instrumentenbau
(Leipzig) “to be the greatest and most conscientious performer of Mozart” then
living. It was hoped that his prospective roll recording project (seemingly unfin-
ished) of the complete Mozart piano sonatas for the Aeolian Company, would pre-
serve “the style of the famous Leipzig Mozart-Players” for future generations.114 It
seems therefore that Reinecke was a representative of a particular style of playing
Mozart that made use of arpeggiation, dislocation, and many other techniques.
These, of course, were not peculiar to Leipzig. But Reinecke may have been the
purveyor of a tradition going back to Mendelssohn and almost certainly influ-
enced other Leipzig-trained pianists in this style. In this respect, the violinist Carl
Flesch’s (1873–1944) recollections of his piano accompanist Julius Röntgen
(1855–1932) are elucidating. Significantly, Röntgen’s first piano teacher in Leipzig
was Reinecke. Flesch valued, above all, Röntgen’s “ability to give himself unreserv-
edly to a work’s emotional content” and to “commune with its creator and forget
the world.” But he was critical of Röntgen’s technique, which “was not flawless
when judged by the highest standards.” Particularly irritating for Flesch was
Röntgen’s “Leipzig” manner—“his arpeggio execution of chords and the delaying
of thematic notes in the right hand.”115 Apart from the apparent connection
between the styles of Röntgen and his teacher Reinecke, it is interesting to
note that Flesch (one of the founders of modern violin playing) associated
arpeggiation and dislocation with flawed playing style. In his performance of

Anon., “Altmeister Karl Reinecke und das Pianola,” Zeitschrift für Instrumentenbau (Sept. 11,
1904): 1039: “Reinecke gilt bekanntlich als der größte und gewissenhafteste Mozartspieler, der jetzt
lebt . . . Weitere Aufnahmen sind in Aussicht genommen, bis auf diese Weise sämtliche Mozartschen
Klavierwerke in der Auffassung des berühmten Leipziger Mozart-Spielers der Nachwelt überliefert
sein werden.”
Carl Flesch, The Memoirs of Carl Flesch, eds. Hans Keller and Carl Flesch, trans. Hans Keller
(London: Rockliff, 1957), 215. Flesch worked with Röntgen between 1903 and 1908.
Unnotated A r peg g iation 163

Beethoven’s Variations in F Op. 34 preserved on a Philipps Duca piano roll (1909),

Röntgen makes very infrequent use of unnotated arpeggiation and dislocation,
certainly not enough to cause Flesch’s level of concern.116 But this may not be the
best example by which to gauge Röntgen’s use of these devices. He may have used
them to a greater extent in other solo works or in his accompaniments.
By modern standards, Reinecke’s recording of the Larghetto sounds extraordi-
nary because we are unaccustomed to such a highly arpeggiated style. But for
many nineteenth-century musicians, such a style was clearly the norm. In this
respect, the Scottish composer Ronald Stevenson (b. 1928) argued, “Schumann
notated more of the Romantic style than possibly any of his contemporaries; but
it is probable that arpeggiation was not notated by even earlier masters, such as
Weber and Schubert, simply because it was a widespread practice, taken for
granted and left to the individual performer’s discretion.”117
Schumann’s notation certainly preserves an abundance of grace-note figures
that imply arpeggiation. This is noticeable, for example, in many places through-
out his Warum? Op. 12 No. 3. In spite of this, pianists such as Reinecke
added more arpeggios to the texture. During the sequence in bar 20, repeated
in bar 24 (Fig. 3.77 )118 in his 1905 piano roll of the work, he arpeggiates the
syncopated chord in the right hand that introduces a leap, playing the lower note
before the beat (Fig. 3.78). During the sequence in bar 34, repeated in bar 38
(Fig. 3.79 ),119 the last chord in the left hand is occasionally arpeggiated. This
separation of the narrowly spaced chord draws attention to the chromatic note C
flat as well as enhances its question-like nature (Fig. 3.80). And in the section
from bars 35 to 40 and its repeat, Reinecke delineates, wherever possible, the

Figure 3.78 Schumann Warum? Op. 12 No. 3, bar 20, Reinecke, piano roll, 1905
(Audio Ex. 3.33 ).

The Pianola in the Netherlands (Rotterdam: Erasmus Muziek Producties, 1995).
Ronald Stevenson, The Paderewski Paradox, 13–14.
Schumann, “Warum? Op. 12 No. 3,” ed. Clara Schumann, 86.
Schumann, “Warum? Op. 12 No. 3,” ed. Clara Schumann, 86.
164 off the record

Figure 3.80 Schumann Warum? Op. 12 No. 3, bar 38, Reinecke, piano roll, 1905
(Audio Ex. 3.34 ).

Figure 3.81 Schumann Warum? Op. 12 No. 3, bars 35 to 40, Reinecke, piano roll, 1905
(Audio Ex. 3.34 ).

multiple melodies in the right hand by arpeggiating the notes that coincide
(Fig. 3.81). In this regard, there is a noticeable similarity to Brée’s advice cited
earlier. Particularly interesting is Reinecke’s downward arpeggiation of the right-
hand chord at the beginning of bar 38. This seemingly unconventional practice is
also noticeable in his manner of playing the last left-hand chord in bars 4 and l6
(Fig. 3.82).120 Here, an arching shape is produced when the three-note chord is
arpeggiated in the order lowest–highest–middle note (Fig. 3.83).
Reinecke was not alone in applying such arpeggiations to Warum? Paderewski’s
1912 recording of it also reveals the use of several unnotated arpeggiations. In
bars 10, 11, 12, and 18 (the first time only), Paderewski delineates the multiple
melodies in the right hand by arpeggiating wherever necessary (Fig. 3.84 ;
Audio Ex. 3.36 ). At places where these melodies form the interval of a seventh,
such as in bars 27 and 29, Paderewski makes further arpeggiations (Fig. 3.85 ;
Audio Ex. 3.37 ). In both cases, the grace note is played before the lower note B
that is aligned with the beat. The A is played last. In addition, the hands are dislo-
cated, producing an even stronger effect of arpeggiation. And in the section from
bars 35 to 40 (the first time), Paderewski, similar to Reinecke, delineates the mul-
tiple melodies by adding arpeggios (see Fig. 3.81, Audio Ex. 3.38 ). On the

Schumann, “Warum? Op. 12 No. 3,” ed. Clara Schumann, 86.
Unnotated A r peg g iation 165

Figure 3.82 Schumann Warum? Op. 12 No. 3, bars 1 to 16, ed. Clara Schumann.

Figure 3.83 Schumann Warum? Op. 12 No. 3, bars 3 and 4, Reinecke, piano roll, 1905
(Audio Ex. 3.35 ).

repeat, he varies these, sometimes playing the notes in a more synchronous

manner. Clearly, his arpeggiations were calculated, not simply automatic manner-
isms. Curiously, although Paderewski commented extensively on rubato practices,
he never mentions the practice of unnotated arpeggiation.
Arpeggiatios were apparently often applied to Warum? The comments of
Moszkowski (cited in chapter 2) are particularly significant in this regard.121 In his
opinion, continual arpeggiation was “one of the most perfunctory styles of which
a pianist can be guilty.” Would Moszkowski have raised an eyebrow at the prac-
tices of Reinecke and Paderewski, who in Warum? do exactly what he forbids?
Again, the obvious disparity between written advice and actual practice is
absolutely clear.
Other editors appear to have been more relaxed than Moszkowski regarding
the addition of arpeggios in Warum? Klindworth, for example, advocated an
arpeggio not originally notated by Schumann in the right hand at the beginning

See chapter 2, page 82.
166 off the record

of bar 11 (Fig. 3.86 ) in his edition (published between 1883 and 1888).122
And he may have expected others to be added as a matter of course.
Other written references identify serious contradictions that are difficult to
comprehend. In From Grieg to Brahms (1927), Daniel Mason describes Saint-
Saëns’s caustic reaction to a particular performance by Paderewski:

Once Paderewski paid him [Saint-Saëns] the compliment of playing one

of his pieces on a state occasion signalizing the cordial political relations
of Poland and France. Alas, the Dean of French music, as Saint-Saëns was
then impressively called, could not abide the Polish romanticist’s habit of
arpeggiating or “breaking” all his chords instead of playing their notes
together—a romantic habit that must have violated the deepest instinct
of his pseudo classical soul. State occasion or no, he rose from his seat
and in his dry, nasal, insistent voice droned: “Monsieur Paderewski, il ne
faut pas jouer comme ça (gesture of arpeggiated chord) il faut jouer
comme ça (gesture of solid chord). Silence of scandalized consternation
as the skeleton at the feast resumed his seat.123

Two things stand out here. First, perhaps due to limited historical perspective,
Mason regarded the practice of unnotated arpeggiation as the habit of the
Romantic era. But, as we have seen, the practice had clearly been in existence for
several centuries before. Second, Saint-Saëns appears to have despised Paderewski’s
employment of unnotated arpeggiation in his music. But if Mason’s description
gives the impression that Saint-Saëns would never have permitted its use, this is
simply not the case: many unnotated arpeggiations are preserved on his 1905
piano rolls of his own music as well as of Chopin’s Nocturne Op. 15 No. 1. In his
Rhapsodie d’Auvergne Op. 73, he makes unnotated arpeggiations at bars 15, 17,
and 19 (Fig. 3.87), although his only indication of arpeggio is at the end of bar 12
(Fig. 3.88).124 And at bar 42, marked pp espressivo (Fig. 3.89),125 he makes arpeg-
giations not indicated in the score (Fig 3.90).
In the first section of Chopin’s Nocturne Op. 15 No. 2, Saint-Saëns frequently
arpeggiates chords in the left hand, similar to Pugno, creating rich sonorities.
These vary in speed according to character and are particularly noticeable during
bars 6 and 7 (Fig. 3.91) and bars 17, 18, and 19 (Fig. 3.92). Many other examples
can be heard in the “Doppio movimento” section of the work.

Robert Schumann, “Warum? Op. 12 No. 3,” Schumann’s Works for the Pianoforte, ed. Karl
Klindworth (London: Lucas, Weber, 1883–88), 8.
Daniel G. Mason, “Postscript after Twenty-Five Years,” From Grieg to Brahms: Studies of Some
Modern Composers and Their Art, 2nd ed. (New York: Macmillan, 1927), 236.
Camille Saint-Saëns, Rhapsodie d’Auvergne Op. 73 (Paris: Durand Schoenewerke,c. 1890), 2.
Saint-Saëns, Rhapsodie d’Auvergne, 4.
Figure 3.87 Saint-Saëns Rhapsodie d’Auvergne Op. 73, bars 15 to 19, Saint-Saëns,
piano roll, 1905 (Audio Ex. 3.39 ).

Figure 3.88 Saint-Saëns Rhapsodie d’Auvergne Op. 73, bars 12 to 20.

Figure 3.89 Saint-Saëns Rhapsodie d’Auvergne Op. 73, bars 40 to 44.

Figure 3.90 Saint-Saëns Rhapsodie d’Auvergne Op. 73, bars 42 and 43, Saint-Saëns,
piano roll, 1905 (Audio Ex. 3.40 ).

Figure 3.91 Chopin Nocturne Op. 15 No. 2, bars 6 and 7, Saint-Saëns, piano roll, 1905
(Audio Ex. 3.41 ).

Unnotated A r peg g iation 169

Figure 3.92 Chopin Nocturne Op. 15 No. 2, bars 17 to 19 Saint-Saëns, piano roll,
1905 (Audio Ex. 3.42 ).

In this light, it is impossible to appreciate why Saint-Saëns reacted as he did to

Paderewski’s performance. It may be that he found Paderewski’s arpeggiations
too frequent and too noticeable, or simply that Paderewski played them in places
where he would not have done so. Whatever the reason, clearly in this case there
is a curious discrepancy between what Saint-Saëns expected from others and what
he himself did. The written text gives a misleading impression. A similar discrep-
ancy was cited at the beginning of this chapter. In 1930, the cellist William
Whitehouse humorously recollected how Lady Hallé accused her husband Sir
Charles of doing exactly what he claimed to dislike—unnotated arpeggiation—in
the playing of her collaborative pianist Leonard Borwick.126

Manner of Execution
In addition to the many anomalies that present between written texts and
recorded evidence discussed earlier, matters concerning the speed of arpeggios
and their placement raise many questions. Writers in the first half of the nine-
teenth century, such as Corri and Czerny, indicate that the speed of arpeggios was
to be variable, depending on the character and context of the music. Hummel, on
the other hand, advised in 1828 that wherever the arpeggio sign appeared, the
notes must be arpeggiated from bottom upward “with the utmost possible
rapidity.”127 By the mid-century, this style was promoted by Thalberg, who recom-
mended an apparently unvarying fast speed resulting in the notes sounding
almost together. It appears that by the late -nineteenth century, a range of prac-
tices were prevalent. In 1891, Lebert and Faisst asked for the long arpeggio chords
in bars 229 and 236 of the third movement of Beethoven’s Sonata Op. 31 No. 1 to

See note 1, page 101.
Hummel, A Complete Theoretical and Practical Course, 66.
170 off the record

be struck “from the lowest bass-note up to the highest soprano-note successively.”128

Yet they gave no indication regarding its placement before or with the beat, or of
the speed of arpeggiation. On the other hand, for the long arpeggiated chord at
the beginning of the second movement of Beethoven’s Sonata Op. 31 No. 1, they
were more prescriptive, advising that it should be arpeggiated slowly “from below,
not fast, in such a way that the measure be counted only from the entrance of the
uppermost tone.”129 Yet as noted earlier, Bülow advised that for bars 157 and 158
in the second movement of the Arietta from Beethoven’s Sonata Op. 111, pianists
should be careful not to arpeggiate before the beat, thereby causing an anticipa-
tion that would disturb “the purity of the harmony.” In this case, there was no
harm in the uppermost tone sounding a little late. Presumably, such matters
varied according to context and taste.
Others, too, were fairly prescriptive. For example, on several occasions Reinecke
stated his preference for an unvarying fast speed of arpeggiation. For Beethoven’s
notated arpeggio sign in bar 6 of the second movement from Sonata Op. 10 No. 1
(Fig. 3.93 ),130 Reinecke warns “against the too broad separation of the notes
from the lowest bass note to the highest treble one.” He insists that the A flat at
the top of the chord must sound closely connected with the preceding D flat, which
would be prevented with a slow arpeggio. He emphasizes that in general, “the
arpeggio sign signifies that the chord should not be struck quite together.”131
Reinecke gives exactly the same advice for the opening chord of the first move-
ment of Beethoven’s Sonata Op. 31 No. 2 (Fig. 3.94 )132 as well as for the
chord in the penultimate bar of the second movement of Beethoven’s Sonata
Op. 57 (Fig. 3.95 ).133 Regarding the latter, Reinecke explains that if the chord
under the pause were to be arpeggiated slowly, the highest note, D flat, and the
A flat of the preceding chord “would be too widely separated from one another.”
Accordingly, the arpeggiation should not sound as follows (Fig. 3.96 ).134
Reinecke also explains that composers usually make it quite clear when they
require a chord to be spread broadly. To illustrate this, he points out Beethoven’s
manner of notating arpeggios in bars 93, 95, and 97 of the first movement
Beethoven’s Sonata Op. 31 No. 2 (Fig. 3.97 ).135 The notes in these chords “must

Beethoven, “Sonata Op. 31 No. 1,” Sonatas and Other Works, ed. Faisst and Lebert, part 3,
vol. 2, 120–21.
Beethoven, “Sonata Op. 31 No. 2,” Sonatas and Other Works, ed. Faisst and Lebert, part 3,
vol. 2, 129.
Ludwig van Beethoven, “Sonata Op. 10 No. 1,” Sonaten für das Pianoforte Solo, ed. Franz Liszt
(Wolfenbüttel: Halle, c. 1845), 6.
Reinecke, The Beethoven Pianoforte Sonatas, 21–22.
Reinecke, The Beethoven Pianoforte Sonatas, 59–60.
Beethoven, Sonata Op. 57, ed. Hallé, 385.
Reinecke, The Beethoven Pianoforte Sonatas, 81.
Beethoven, “Sonata Op. 31 No. 2,” ed. Liszt, 4.
Unnotated A r peg g iation 171

be more broadly separated” from one another so that a division is made between
the two hands, with the left hand always taking the minim.136
Clearly, there must have been other situations in which a slower arpeggiation
was in order to enhance the character of the composition. Clara Schumann, for
one, commented about this with regard to the grace-note arpeggios in Schumann’s
Warum? Op. 12 No. 3. She explains that “in accordance with the character of the
piece,” these must not be played quickly (Fig. 3.98).137 In her annotated example,
she clearly intended the arpeggio to start on the beat, rather than before it, and
repositioned the quaver rest to make the point clear (Fig. 3.99).138 Bülow appears to
have associated, at least on some occasions, a softened effect with slow arpeggia-
tion. In bar 89 of Beethoven’s Fantasie Op. 77, for example (Fig. 3.100 ), he
advises that the first chord marked sf “be struck together firmly, without any
breaking,” whereas the second should on the contrary “be very slowly arpeg-
giated.”139 Interestingly, Bülow’s words—“without any breaking”—would seem to
imply that on occasion, some degree of arpeggiation was permissible even for
accented chords.

Figure 3.98 Schumann, Warum? Op. 12 No. 3, bar 13, ed. Clara Schumann.

Figure 3.99 Schumann, Warum? Op. 12 No. 3, bar 13, annotated by Clara Schumann.

Reinecke, The Beethoven Pianoforte Sonatas, 59–60.
Schumann, “Warum? Op. 12 No. 3,” ed. Clara Schumann, 86.
Schumann, “Warum? Op. 12 No. 3,” ed. Clara Schumann, 86.
Beethoven, “Fantasie Op. 77,” Sonatas and Other Works, ed. Bülow, part 3, vol. 4, 104.
172 off the record

Returning to Reinecke, he, like Thalberg, gives the impression that apart from
special circumstances notated by the composer, the notes of chords to be arpeg-
giated ought to be spread quickly and almost together. But in his 1905 piano roll
of his arrangement of Mozart’s Larghetto, Reinecke’s arpeggiations often sound
broader than implied by his words. And he certainly varied their speed. Either
Reinecke advised something that he did not uphold in reality or his terminology
means something significantly different now than it did then. For him, striking
the notes “not quite together” or “not precisely together” apparently encompassed
a wider range of speed than a face-value interpretation suggests.
Others advise that arpeggios, when notated as embellishments, should be
aligned with the beat rather than starting before it. In 1871, Lebert and Faisst
explained that for the grace-note arpeggio in bar 5 of the Theme of Beethoven’s
Six Variations Op. 34 (Fig. 3.101 ), one should “strike the first note of the
arpeggio-figure at the same time with the f in the bass.”140 They made no recom-
mendation here for the speed of arpeggiation. But for the grace-note arpeggios in
Variation XV of Beethoven’s Fifteen Variations with Fugue Op. 35 (Fig. 3.102 ),
they ask the player to “begin the appoggiaturas in both hands at the same time
with the beat, and execute them rapidly.”141 In other cases, swift arpeggiation is
recommended to prevent interference with the rhythmic pattern of accompany-
ing parts. Such advice is exemplified in Bülow’s annotations to Cramer’s Study No.
18 from Studio per il Piano Forte, I, Op. 30 (Fig. 3.103 ).142 He warns of the
acoustic impurity—which damages sensitive ears—that is inevitable when the
notes of chords arpeggiated before the beat mix with the preceding harmony. To
avoid this, the chords should be arpeggiated as in bar 1, and eventually with prac-
tice as in bar 3 (Fig. 3.104).143 In 1871, Bülow insisted on this rule of alignment
with reference to the grace-note chords in bars 5, 13, and 14 of the second move-
ment of Beethoven’s Sonata Op. 109 (Fig. 3.105 ). He explains that “the written
out arpeggios are, like all other embellishments, subject to the invariable rule,
unfortunately in many ways forgotten, that their initial note should fall upon the
same metrical part to which the principal note following the embellishment in the

Ludwig van Beethoven, “Six Variations Op. 34,” Sonaten und andere Werke für das Pianoforte, 5
vols. (Stuttgart, Germany: Cotta, 1871). Reprinted with German and English text. Trans. as Sonatas
and Other Works for the Pianoforte . . . With the Cooperation of Hans von Bülow, Immanuel von Faisst, Ignaz
Lachner, Franz von Liszt, founded by Sigmund Lebert. English trans. of Instructive Text by J. H. Cornell
(Stuttgart, Germany: Cotta, 1891, part 3, vol. 3: Variations, Rondos, Bagatelles, Andante in F major
[Op. 33–51 and without number of opus]). With the cooperation of Immanuel von Faisst, elaborated
by Sigmund Lebert, 32.
Beethoven, “Fifteen Variations with Fugue Op. 35,” Sonatas and Other Works, ed. Bülow,
part 3, vol. 3, 54.
Johann B. Cramer, Sixty Selected Pianoforte-Studies, 3rd ed., ed. Hans von Bülow, trans.
Constance Bache (Munich: Aibl, 1889), 43. Note that Bülow renumbers this study as No. 192.
Cramer, Sixty Selected Pianoforte-Studies, ed. Bülow, 43.
Unnotated A r peg g iation 173

Figure 3.104 Cramer Study No. 18 Op. 30, bars 1 and 3, annotated by Bülow.

Figure 3.106 Beethoven Sonata Op. 109, second movement, annotations to bars 5,
13, and 14 by Bülow.

notation belongs.”144 To make his point clear, he provided the annotation in Figure
3.106. Bülow prescribed the same treatment for the grace-note appoggiaturas in
bars 1, 3, and 5 of Variation 1 of the same movement. In a similar vain, he observed
that for the grace-note arpeggios in bars 46 and 47 of Variation No. 33 from
Beethoven’s Diabelli Variations Op. 120 (Figure 3.107 ), “the lowest note must
every time coincide exactly with the fifth thirty-second note of the bass.”145
With reference to Cramer’s Study No. 21 from Studio per il Piano Forte, II, Op.
40 (Fig. 3.108 ),146 Bülow insists that acciaccatura grace notes (that form vari-
ous arpeggiated intervals in the bass) be aligned with the beat. He explains that
the acciaccatura itself should be more accented because it is the bass note of the
chord. Otherwise, the following note “would strike the ear more forcibly” because
it is longer.147 To make it clear, Bülow indicates the following placement for the
acciaccaturas in the left hand at bar 4 (Fig. 3.109), aligning them with the first
chord of each triplet of chords in the right hand.148

Ludwig van Beethoven, “Sonata Op. 109,” Sonatas and Other Works ed. Bülow, part 3, vol. 5, 84:
“Die ausgeschriebenen Arpeggien sind wie alle anderen Verzierungen der ausnahmslosen, leider viel-
fach in Vergessenheit gerathenen Regel unterworfen, dass ihre Anfangsnote auf den nämlich Taktheil
zu treffen hat, welchem die Verzierung folgende Hauptnote in der Notation angehört.”
Ludwig van Beethoven, “Diabelli Variations Op. 120,” Sonatas and Other Works, ed. Bülow,
part 3, vol. 5, 205: “Von den in Vorschlags—Manier notirten Accord–arpeggien muss die unterste
Note genau mit den fünften Zweiunddreissigstel des Basses jedesmal zusammentreffen.”
Cramer, Sixty Selected Pianoforte-Studies, ed. Bülow 63.
Cramer, Sixty Selected Pianoforte-Studies, ed. Bülow 63. Note that Bülow renumbers this study
as No. 29.
Cramer, Sixty Selected Pianoforte-Studies, ed. Bülow 63.
174 off the record

Figure 3.109 Cramer Study No. 21 Op. 40, bar 24, annotated by Bülow.

Figure 3.110 Beethoven Moonlight Sonata Op. 27 No. 2, bars 14 to 23, Magazine of Music.

Further evidence supporting the playing of arpeggios on the beat and quickly
is found in an annotation of the second movement of Beethoven’s Moonlight
Sonata Op. 27 No. 2, cited in the Magazine of Music (1891). Here, the arpeggio
marked in the right hand at bar 16 is to be played as shown at the end of the
excerpt (Fig. 3.110).149 Indeed, a similar execution appears to have been integral
to the Leschetizky method. Brée states that for chords with an appended grace-
note appoggiatura (Fig. 3.111), playing may be facilitated “by a swift arpeggio,
taking the first tone of the arpeggio with the fundamental.”150
The manner of execution when chords in both hands are notated to be arpeg-
giated poses questions about placement and timing. The majority of written texts
that I have examined do not clarify the matter of placement before or on the beat.
And there is varying opinion about execution. For example, with reference to the
arpeggiated chords in bar 2 of Theme of Beethoven’s Six Variations Op. 34
(Fig. 3.112 ) Lebert and Faisst advised in 1871 that “these arpeggios should be
simultaneously begun and ended with both hands and should increase in strength
from the lowest to the highest tone in each hand, so that the highest tone only
attains the power indicated through the signs sf, mf, (afterwards f).”151 And for the

Beethoven, “Moonlight Sonata,” Magazine of Music (1891): part 2, 129.
Brée, The Leschetizky Method, 58.
Beethoven, “Six Variations Op. 34,” Sonatas and Other Works, ed. Lebert and Faisst, part 3,
vol. 3, 32.
Unnotated A r peg g iation 175

Figure 3.111 Arpeggiation of chord with appended grace-note appoggiatura,

annotated by Brée.

arpeggiated chords in the right and left hand in bar 130 of the Fugue from
Beethoven’s Fifteen Variations with Fugue Op. 35, they recommend to “begin and
close the arpeggio in both hands simultaneously, and execute it rapidly.”152
Whether this was, for them, a universal rule is unclear. On the other hand, accord-
ing to Brée, Leschetizky expected the arpeggiation to start with the lowest note in
the bass and to continue through to the highest note in the treble (Fig. 3.113).
“For arpeggios in both hands, do not begin with both hands together,” she
insists.153 Brée’s annotation clearly shows the arpeggiation starting with the beat.
A few years later, in 1918, the same approach appears to have been adopted in
Niemann’s Klavier Lexicon (Fig. 3.114).154 Although it is not clear from this
example whether arpeggiation was supposed to begin before or on the beat, a
second example with shorter chords marked staccato clearly shows arpeggiation
starting on the beat (Fig. 3.115).155 No doubt such advice was aimed at inexperi-
enced performers to stem practices that would lead to confusion of harmony and
distortion of rhythm and melody. Nevertheless, early recordings reveal that many

Figure 3.113 Arpeggiation of chords in both hands, annotated by Brée.

Beethoven, “Fifteen Variations with Fugue Op. 35,” Sonatas and Other Works, ed. Lebert and
Faisst, part 3, vol. 3, 62.
Breé, The Leschetizky Method, 48.
Niemann, Klavier Lexicon, 45
Niemann, Klavier Lexicon, 46.
176 off the record

Figure 3.114 Arpeggiation of chords in both hands, annotated by Niemann.

Figure 3.115 Arpeggiation of short chords in both hands marked staccato, annotated
by Niemann.

professional pianists such as Reinecke, Leschetizky, Saint-Saëns, Pugno,

Pachmann, and Paderewski played notated and unnotated arpeggios in a variety
of speeds. And they placed them before and with the beat, depending on context
and required effect.

Changing Tastes
During the second half of the nineteenth century, several writers confirm the
widespread practice of unnotated arpeggiation, but seek its judicious use or total
Unnotated A r peg g iation 177

eradication. In The Art of Piano Playing and Teaching (c. 1895), Maria Grimaldi
notes the tendency “to spread chords which are meant to be played as a whole.”
Likening synchronous chord playing with the beautiful quality that results from
tight ensemble in orchestral playing, she decries unnotated arpeggiation as “a
grave fault” that almost always highlights “a proclivity to sentimentalism.”156
Using a similar analogy, Gieseking registered his disdain of the arpeggio
manner in 1932. He finds it remarkable that even amateurs criticize orchestras if
the notes in chords do not sound precisely together, whereas for concert pianists
“this grievous offence against all musical feeling is nearly always overlooked.”157
Bülow, too, was apparently staunchly opposed to unnotated arpeggiation.
With reference to Cramer’s Study No. 1 Op. 30, he warns that teachers “should
insist on a systematic arpeggio” whenever an arpeggio sign is notated and “should
just as strictly insist on the avoidance of the mannerism of striking notes arpeggio
where not specifically so marked.” Any concession to this from the very start
results in “ineradicable harm.”158 And for Cramer’s Study No. 61 Op. 40, he admon-
ishes “any inclination to play the sixths arpeggio,” a practice that must be checked
by the teacher.159
In his Technical Guide to Touch, Fingering, and Execution on the Pianoforte (1877),
Lindsay Sloper makes reference to the difficulty of playing the components of
double notes and chords with equal strength and absolute synchrony. In his opin-
ion, this is rarely accomplished. The result is a “broken and inarticulate perfor-
mance,” particularly by the left hand. In order to accommodate the varying
positions of white and black notes in chords, he advises scrupulous care in regu-
lating the curve of each finger, so that all make contact with the keys “precisely at
the same moment.”160 The impression here is that arpeggiating chords was so
much the norm that firm chord playing was a skill to be achieved only by serious
technical study. Nowadays, the situation is almost completely reversed. The major-
ity of pianists are taught from an early age to play the notes in chords absolutely
together. Firm chord playing has become second nature.
Sloper was not alone in insisting that the fingers make even contact with the
keys. In 1854, Gatien Marcailhou discussed the great difficulty of playing octaves
in the left hand. Due to the natural bias toward the thumb, octaves tend “to be
arpeggiated, rather than executed with equal force between thumb and fifth
finger.” According to Marcailhou, this lack of simultaneity of action causes a loss

Maria L. Grimaldi, The Art of Piano Playing and Teaching (London: Reeves, c. 1895), 22.
Gieseking and Leimer, Shortest Way to Pianistic Perfection, 56.
Cramer, Sixty Selected Pianoforte-Studies, ed. Bülow, 3.
Cramer, Sixty Selected Pianoforte-Studies, ed. Bülow, 106. Note that Bülow renumbers this study
as No. 47.
Lindsay Sloper, Technical Guide to Touch, Fingering, and Execution on the Pianoforte (London:
1877), 21.
178 off the record

of resonance, a weakening of the strong beat effect that is often associated with
the octave, as well as a disturbance in precision.161
Bülow, too, warned against arpeggiating octaves in the left hand in an annota-
tion to Cramer’s Study No. 32 Op. 30 (Fig. 3.116 ).162 Here, the difficulty of
playing the continuous semiquaver passages would undoubtedly be lessened by
arpeggiating the octaves, thus freeing the hand. In spite of this, Bülow’s advice is
that “the player must guard against” arpeggiating the octave at the beginning of
each bar.163
Further warnings against the arpeggiation of chords are found in annotations
to Henry Bertini’s Preparatory Studies Op. 32. For Study No. 6, the editor Gordon
Saunders states that the notes in the double-note theme “must be played strictly
together.”164 And for all passages of double notes in Study No. 8, great care must
be taken “to play the two notes exactly together” in the alternating thirds and
The playing of double notes strictly together was also recommended earlier in
the nineteenth century. Czerny does so for a passage in Variations Op. 11 by
Henselt (Fig. 3.117 ). He explains that it would be erroneous to separate the
double notes in the right “as if the under note were an appoggiatura.”166 Concerning
a passage in Chopin’s Piano Concerto Op. 21 (Fig. 3.118 ), Czerny warns again
that the double notes must “be struck exactly together.”167 And with reference to
the upper octaves forming the melody in one of the Etude en double notes Op. 28
(Fig. 3.119 ) by Rudolph Willmers (1821–78), Czerny warns that these “must
be struck particularly loud and firm (but not arpeggio).”168 The arpeggiation of
double notes must so often have been applied that Czerny felt a need to censure
the practice from time to time and in certain situations.
Although unnotated arpeggiation was used throughout the nineteenth cen-
tury to give heightened expression to melody notes, the bringing out of the
melody without the use of arpeggiation was also often advocated. Indeed, this is

Gatien Marcailhou, L’Art de composer et d’exécuter la musique légère (London: Schott, 1854), 7:
“Une grande difficulté existe au piano, lorsqu’on attaque des octaves à main gauche, elle consiste à
donner de la simultaneité au deux doigts qui frappent l’octave, c’est à dire au 5.e et au pouce de la main
gauche; la main est toujours entrainée du côté du pouce, et l’octave, au lieu de frapper avec la même
force dans la 5.e et le pouce, l’exécute en arpège: il en résulte de la faiblesse dans l’octave, le son obtenu,
est moins fort, l’octave est en un mot, boiteuse, le temps fort qui est souvent attaqué en octave est
faible, ce qui nuit beaucoup à la précision, et à l’entrain du rhythme.”
Cramer, Sixty Selected Pianoforte-Studies, ed. Bülow, 84. Note that Cramer renumbered this as
No. 39.
Cramer, Sixty Selected Pianoforte-Studies, ed. Bülow, 85.
Henri Bertini, Twenty Five Preparatory Studies Op. 32 for the Pianoforte, ed. Gordon Saunders
(London: Hammond, 1902), 10.
Bertini, Twenty Five Preparatory Studies, ed. Saunders, 12.
Czerny, Art of Playing, 14.
Czerny, Art of Playing, 15.
Czerny, Art of Playing, 20.
Unnotated A r peg g iation 179

strongly implied in Czerny’s annotation to a musical example of slow character

(Fig. 3.120 ) for which the performance “must be made intelligible by atten-
tively given relief to the melody.” He advises that in preference to all other means
of expression, strict legato must be observed and that all notes in each chord “must
be struck with firmness and energy; and the highest notes in the right hand, must
be brought out rather prominently, because they form the melody.”169
In addition to the use of arpeggiation, Czerny advocated the development of
finger weight for bringing out melody notes while playing firm chords. Later in
the century, many other writers adopted this practice, at least in theory. In 1861,
Adolphe Kullak (1823–62) refers to Thalberg’s rule that chords forming a melody
“should be arpeggio’d, in order to give the melody the necessary emphasis.” He
acknowledges that the melody can be more easily emphasized by this means. But
to balance this, he gives an exercise (Fig. 3.121 ), explaining that

in no case should one neglect to practise emphasizing the melody-note

when striking two or more notes simultaneously. Aside from the utility
of such practice, the monotony caused by the continuous arpeggios
during a long movement is mitigated by the change. This monotony
would be unavoidable, for instance, in the first part of Beethoven’s C
[sharp] -minor Sonata, should the player constantly strike the melody-
note after the accompaniment. Here it is best to save the arpeggio for the
most pregnant passages. The middle movement of the Sonate pathétique
would be an excellent practice-piece in the sense intended. To give the
fingers the necessary independence for the purpose in question, both
hands might play the exercise [in Figure 3.121].170

For the exercise in Figure 3.121, Kullak advises that the large notes are to be
accented more strongly than the others.
Kullak’s recommendation about how to express the melody using finger weight
rather than continual arpeggiation documents, at least in his case, a move away
from the use of unnotated arpeggiation. But Kullak was not completely opposed
to it, particularly for “the most pregnant passages.” Presumably, such places might
contain dissonant melodic or harmonic notes, syncopations, or places where there
are significant leaps in the melodic material—in short, places of extraordinary
character. Here, the contradictions between various written texts are again
apparent. It is significant, for example, that Kullak advised the avoidance of
arpeggiation in Beethoven’s Sonata Pathéthique and Moonlight Sonata, contradict-
ing the directions in Potter’s editions of these works discussed earlier. We have

Czerny, Piano Forte School, 75.
Adolphe Kullak, Die Ästhetik des Klavierspiels (Berlin: 1861); trans. of the 3rd ed. by Dr. Theodor
H. Baker as The Aesthetics of Pianoforte Playing, ed. Dr. Hans Bischoff (New York: 1893; reprint,
New York: 1972), 297–98.
180 off the record

seen that some pianists like Paderewski made unnotated arpeggiations in the
Moonlight Sonata. Kullak’s exercise for developing the digital independence
needed to accentuate an individual note in a chord is possibly one of the first of its
kind. It set a precedent that was to be adopted later in the nineteenth century.
Other late-nineteenth-century pedagogues generally forbade the use of unno-
tated arpeggiation. C. A. Ehrenfechter did so in his Technical Study in the Art
of Pianoforte-Playing (London, 1891).171 Referring to the first movement of
Beethoven’s Moonlight Sonata (Fig. 3.122 ), Ehrenfechter discusses the diffi-
culty of playing melody notes and accompaniment in the same hand with stretched
fingers. Should the effect be unsuccessful due to “insufficient singing quality, or
hardness of tone,” he advises—with a hint of sarcasm—that playing the interval
in “the arpeggio manner” might be a solution. But he is swift to warn that the
adoption of arpeggios not indicated by the composer “can hardly be defended.” He
does, however, permit arpeggiation when some intervals (for example the ninth
at c) are too large for pianists who have small hands.172 But in spite of this and
other warnings, many pianists employed unnotated arpeggiations frequently and
to great expressive effect.
The complexity of bringing out the melody notes in a chordal texture is made
clear in an enlightening description in the Magazine of Music (1891). With refer-
ence to Chopin’s Nocturne Op. 48 No. 1, the anonymous author gives an example
(Fig. 3.123 ) and advises that for the legato chords, all the notes must be played
with equal strength (apart from the melody notes, which must be played more

The secret lies in the position of the right hand, and in the worked out
strength of the last two fingers, the 3rd and 4th. The weight of the hand
must lie on the outer position of the hand, the thumb and first fingers
touching the keys more lightly than the finger playing the topmost note,
which finger, must strike the key boldly, and in the centre of the ivory. By
this we get, as Chopin intended, the following melody clear, unclouded,
and singing [Fig. 3.123 ].173

This type of technical advice finds its most detailed manifestation in Technique
and Expression in Pianoforte Playing (London, 1897) by Franklin Taylor (1843–
1919). Taylor was an influential teacher and interpreter, regarded very highly by

Ehrenfechter was the English disciple of the influential pianist and pedagogue Ludwig Deppe
(1828–90). The importance of his technical advice is discussed in Reginald Gerig, Famous Pianists and
Their Technique (Washington, D.C.: Luce, 1974).
C. A. Ehrenfechter, Technical Study in the Art of Pianoforte-Playing (Deppe’s Principles) (c. 1889),
3rd ed. (London: Reeves, 1891), 64–65.
Anon., “Touch—What Rubinstein Says about It,” Magazine of Music (1891): part 2, 172.
Unnotated A r peg g iation 181

pupils and colleagues. His legacy remains in an outstanding series, Progressive

Studies for the Pianoforte. His Technique and Expression in Pianoforte Playing is cur-
rently still in use. Advice such as the following undoubtedly influenced many musi-
cians throughout the twentieth century and into the twenty-first. Taylor describes
the difficulties of balancing melody, bass, and inner accompaniment, particularly
when one or other of the hands has to play two parts. The trick is “to acquire the
power of striking two notes with one hand and at the same moment, the strength
of one of the notes being decidedly in excess.”174 Providing finger exercises
(Figure 3.124 ), Taylor offers highly detailed solutions, censuring any tendency
to make unnotated arpeggiations. Such advice became the focal point of expres-
sion in piano playing, particularly during the second half of the twentieth century:

The strength of the tone depends upon the speed of the blow and the
amount of pressure combined, but if one of the two fingers engaged were
to move decidedly quicker than the other, it would be the first to depress
its key, and the two sounds would not be produced simultaneously, the
weaker sound would follow the other. There is no doubt, however, that
the finger which produces the strong tone does move a little quicker than
the other, but so little that it has just time to give the requisite amount
of pressure at the precise moment at which the weaker finger arrives at
the depth of its key without any pressure whatsoever. Such minute dif-
ferences cannot be calculated; it is a question of sense of touch, which
can be cultivated and developed in this direction to a surprising extent,
given the necessary perseverance and attention.
It may be of benefit to suggest here certain forms of exercise by which
the necessary control over the fingers may be acquired, it being observed
that such exercises may be multiplied and amplified to any extent, and
always repay the labour bestowed on them. In practising them it must be
observed that it is far easier to play two notes of dissimilar strength one
after the other than both, together, and that there is consequently a
temptation to spread the notes, in the manner of arpeggio; this tendency
must, of course, be carefully guarded against. In the examples [Fig. 3.124
], the notes with open heads are to be made strong and the black notes
weak, and the open notes with quaver stems are to be played of the value
of quavers only.175

In an earlier publication, Primer of Pianoforte Playing (1877), Taylor provided

two examples (Fig. 3.125 ) in which melody and accompaniment are played by
the same hand. Example 1 poses no problem, but Example 2 needs assiduous

Franklin Taylor, Technique and Expression in Pianoforte Playing (London: Novello, 1897), 61.
Taylor, Technique and Expression 61–62.
182 off the record

study “to produce two different qualities of sound in the same hand at the same
time.” Taylor is emphatic that arpeggiating the first chord of each group—in other
words, playing the B after the D that accompanies it and so on in Example 2—is
by no means permitted. He concedes, however, that this is a very common habit
and an easy means of effecting “difference in the strength of the two sounds.” Like
others, he warns that arpeggiation of a chord “is very rarely permissible unless it
is indicated by the composer.”176
It is worth noting that pianists who used arpeggiation also valued other means
of emphasizing particular notes in chords. Leschetizky, who, as we have seen,
often used arpeggiation to help bring out the melody note, also taught specific
ways of doing this without its aid. According to Breé, one starts by “making the
finger which bears the theme longer than the others.” This will usually be the high-
est note of the chord played by the fifth finger in the right hand. Following this,
the trick is for this longest finger to press “its key down deepest, obtaining a fuller
tone.” Breé explains further that the wrist should support the finger that plays the
theme by bearing down on it while giving less pressure to the other notes.
The effect will be even more successful by engaging the sustaining pedal with the
chord and at the same time releasing all but the thematic note “instantly after
striking the chord.”177
Although Taylor and others forbade the use of unnotated arpeggiation, this
was by no means a universal attitude. As Gerhard has noted, the German pianist,
pedagogue, and scholar Rudolf Maria Breithaupt (1873–1945) advocates, in Die
natürliche Klaviertechnik (1905), the arpeggiation of chords in both hands “espe-
cially to give a chord a special emphasis or a particularly energetic, sharply
accented character.” This produces a brilliant instrumental effect by means of “a
hint of a spread chord” or one that is spread “almost unnoticeably.” One is
reminded here of Bériot’s advice in the mid-nineteenth century. When done suc-
cessfully, one should apparently “neither hear the arpeggio, nor experience it as a
deliberate effect.”178 Perhaps surprisingly, this advice was retained in a new
edition of Breithaupt’s text in 1927.
In the early years of the twentieth century, opinion about the use of arpeggia-
tion was certainly divided. Gerhard also notes that in 1911, the music journalist

Franklin Taylor, Primer of Pianoforte Playing (1877), ed. George Grove (London: 1884), 13.
Breé, The Leschetizky Method, 56.
Rudolf M. Breithaupt, Die natürliche Klaviertechnik, Band 1 (Handbuch der modernen Methodik
und Spielpraxis für Künstler und Berufsspieler, Lehrer und Lehrerinnen, Konservatorien und Institute,
Seminare und Schulen) (2nd ed., Leipzig, Germany: Kahnt, 1912), 246–47, in Gerhard, “Willkürliches
Arpeggieren,” 129: “Die arpeggienartige Brechung von Akkorden findert häufig im doppelhändigen
Akkordspiel eine besondere Verstärkung oder einen besonders energischen, scharf akzentuierten
Charakter zu geben . . . Man muss dabei achthaben, dass man nur den glänzenden Instrumentaleffekt
hinstellt und das Mittel der Brechung nur andeutet, oder am besten fast ganz unmerklich gestaltet.
Man darf die ‘Arpeggie’ weder heraushören noch ihre Form als besondere Absicht empfinden.”
Unnotated A r peg g iation 183

Karl Grunsky (1871–1943) polemicized about “how the awkward player loves
resorting to arpeggiation out of lack of other ways to achieve expression.” Grunsky
scolds Bülow for his “inclination to the sentimental” in notating many arpeggios
in his piano arrangement of Wagner’s Tristan und Isolde. For Grunsky, this was
proof of “overdone emotionality.”179 Yet, in the same year, the German pianist and
music theorist Ludwig Riemann insisted that the arpeggio is “the finest, for the
piano the very best, indicator of musical expression.”180
Amid such polarization, the general feeling that emerged was that solid chords
sounded stronger than arpeggiated chords. Interestingly, Tovey advised in 1929
that “like harpists, and unlike pianists,” string players “can produce twice as much
tone by spreading their chords instead of cutting them short.”181 From this we can
conclude that Tovey considered arpeggiated chords in piano playing to be weaker
in tone than if played unarpeggiated.
By 1900, some pianists were actively trying to eliminate practices such as
unnotated arpeggiation and dislocation from their own playing. As noted in chap-
ter 2, Busoni showed himself to be one of the leaders of this trend. In 1894, he
advised the player to be especially careful to play all the notes in chords together,
with particular reference to the works of J. S. Bach.182 Further to this, in 1914
Busoni was critical of the “incomprehensible Arpeggiando sign” added in many
editions to the final chord of Invention No. 1 BWV 772 (Fig. 3.126 ). In his
opinion, this is “contrary to the manly style of the piece.” He warns the student
against “such effeminacies in this and in analogous cases.”183
Busoni’s 1925 piano roll of his own transcription of the Chaconne from J. S.
Bach’s Suite for Solo Violin, shows a fairly studious avoidance of arpeggios, though
one or two can be heard to enhance harmonically poignant moments. In his 1922
recording of Bach’s Prelude and Fugue in C Major BWV 846 from Book 1 of The
Well-Tempered Clavier, there is absolutely no arpeggiation. On the other hand, in

Karl Grunsky, Die Technik des Klavierausrugs. Etwickelt am dritten Akt von Wagners Tristan
(Leipzig, Germany: 1911), 85–86; in Gerhard, “Willkürliches Arpeggieren,” 132–33: “wie der unbe-
holfene Spieler aus Mangel anderer Wege zum Zweck des Ausdrucks gern arpeggiert”; “Hang zum
Sentimentalen”; “Belege für Empfindelei.”
Ludwig Riemann, Das Wesen des Klavierklanges und seine Beziehungen zum Anschlag. Eine
akustich-äethetische Untersuchung für Unterricht und Haus dargeboten (Leipzig, Germany: 1911), 89;
in Gerhard, “Willkürliches Arpeggieren,” 133: “der feinste, dem Klavier ureigene Gradmesser der
musikalischen Gefühlswerte.”
Tovey, “Brahms,” 165.
Busoni, “Interpretation,” 87.
Johann S. Bach, “Two-Part Inventions BWV 772–786,” Klavierwerke Busoni-Ausgabe iv
Zweistimmige Inventionen (BWV 772–786), ed. Feruccio Busoni, trans. L. Elson (Leipzig, Germany:
Breitkopf & Härtel, c. 1930–50), 3: “Das ungebriefliche Arpeggiandozeichen, welches man vor diesem
Schlussakkorde vielfach in Ausgaben antrifft, widerspricht durchaus dem männlichen Stile des Stückes
und ist, im Bach’-schen Sinne, als ‘Stillosigkeit’ zu qualifizieren. Vor solchen Verweichlichungen soll
der Schüler an dieser und anderen analogen Stellen besonders gewarnt werden.”
184 off the record

his recording of Chopin’s Nocturne Op. 15 No. 2 made in the same year, he cer-
tainly uses it in the section from bars 17 to 21 in the left hand (Audio Ex. 3.43 ),
but to a much lesser extent than pianists such as Saint-Saëns and Pugno. Although
Busoni may have been trying to rid piano playing of the so-called effeminacies
associated with unnotated arpeggiations, he continued to make them on certain
Despite Busoni’s leanings, at least one musician closely associated with him
advocated unnotated arpeggiation. In 1941, Egon Petri (1881–1962)—Busoni’s
student—edited J. S. Bach’s French Suites arranged by Busoni. He suggests that in
the Sarabande from BWV 814, in which Busoni has augmented the sound and
harmonies with large chords (Fig. 3.127 ), “the chords can also be played with
gentle arpeggio.”184
Although many pianists at the turn of the twentieth century continued to
make considerable use of arpeggiation, some, such as Grieg, were already using it
infrequently. Table 3.5 shows the number of occasions on which Grieg made

Table 3.5 Grieg’s Unnotated Arpeggiations in His 1903 Recordings

Work Notated Arpeggios Unnotated Arpeggios

To Spring Op. 43 Bars 11 and 13—curved line No unnotated arpeggiations

No. 6 arpeggios in the chords in the left

Bars 23 and 27—bass note

acciacatura octave additions in the
left hand

Bars 45 to 68—acciacatura grace

notes indicating the arpeggiation of
the octaves in the right hand

Bar 71—crenellated line arpeggio

indicating separation from lowest
note to highest note

Bar 72—arpeggiated chord with

notes notated separately

Johann S. Bach, French Suites, arr. Feruccio Busoni, ed. Egon Petri (London: British & Continental
Music Agencies, 1941), 37.
Unnotated A r peg g iation 185

Table 3.5 Cont’d

Work Notated Arpeggios Unnotated Arpeggios

“Finale” from Bar 85—crenellated line arpeggio Bar 64 to 66—variation of the
Sonata Op. 7 indicating separation from lowest notated rhythms creating
note to highest note arpeggiations

Bar 75—possible arpeggiation

from lowest to highest note of
second dotted quaver chord
Gangar Op. 54 No notated arpeggiations No unnotated arpeggiations
No. 2
“Alla Menuetto” No notated arpeggiations Bar 17—arpeggiation from
from Sonata the lowest to the highest note
Op. 7 of the chord in the right hand
Wedding Day at No notated arpeggiations in the Bar 19—arpeggiation from
Troldhaugen section that Grieg recorded lowest to highest note of the
Op. 65 No. 6 first chord in the bar.

Bar 56—arpeggiation from

lowest to highest note of the
first chord in the bar
Humoreske No notated arpeggiations Bar 41—possible arpeggiation
Op. 6 No. 2 from lowest to highest note of
the chord
Bridal Procession Bars 68 and 72—grace-note No unnotated arpeggiations
Op. 19 No. 2 arpeggio figures in the left hand
at the beginning of each bar
Remembrances No notated arpeggiations Bar 58—arpeggiation from
Op. 71 No. 7 lowest to highest note of the
chord on the second beat in
the right hand

unnotated arpeggiations in his 1903 recordings of his own works compared with
those arpeggiations notated in the music. Clearly, Grieg made unnotated arpeg-
giations very infrequently in the music he recorded. This is one of the reasons why
his playing sounds more synchronized than many pianists of a similar generation.
Though it is possible that in other repertoire, such as Chopin or Schumann, he
might have arpeggiated more frequently, this does not seem to accord with the
evidence of his piano playing discussed in chapter 2.
186 off the record

During the first half of the twentieth century, pianists such as Hofmann,
Hambourg, and Gieseking railed against the arpeggio manner. Their playing shows
correspondingly less use of unnotated arpeggiation than other pianists. At cer-
tain key moments, however, they could not resist the technique. In his 1912
recording of Chopin’s Nocturne Op. 9 No. 1, Hofmann makes unnotated arpeg-
giations at bar 12 in the right hand and twice in bar 19. And during bars 30 and
31, he arpeggiates the octaves in the right hand. In Schumann’s Warum? Op. 12
No. 3, he makes a noticeable arpeggiation at the beginning of bar 10 and arpeg-
giates the chord in the left hand at bars 21 and 25. In bar 34, he effects a rather
curious arpeggiation by adding a D flat to the texture of the first chord and making
a separation between it and the notated C flat. And similarly to Reinecke, he
arpeggiates the first chord in the right hand at bar 38.
In 1922, Mark Hambourg made a particular a point of the deleterious effect of
unnotated arpeggiation especially in slow movements. Providing the annotations
cited in chapter 2,185 he described the effect as a “drawling” or “stuttering” in
speech, which led to a loss of symmetry.186 Significantly, Hambourg can be seen to
have opposed the style of playing indispensable to his teacher Leschetizky.
In spite of the efforts of Busoni, Hofmann, Gieseking, and Hambourg to eradi-
cate unnotated arpeggiation, some pianists continued to practice it as late as the
1950s. Unnotated arpeggiation can be heard particularly in the recordings of pia-
nists associated with Clara Schumann and Brahms. For example, Eibenschütz
makes a few unnotated arpeggiations in Brahms’s very energetic Ballade Op. 118
No. 3 recorded in 1903. The chords in the right hand on the last beat of bar 10
(Audio Ex. 3.44 ), and the repeat at bar 86 are arpeggiated to mark the begin-
ning of a different mood. And by making a sweeping arpeggiation from the lowest
note in the fourth quaver beat to the highest note of the chord in the middle of
bars 33 and 35, Eibenschütz produces a very effective emphasis on the diminished
seventh harmony (Audio Ex. 3.45 ). She also makes unnotated arpeggiations in
her 1953 recording of Brahms’s Intermezzo in C Op. 119 No. 3 and her 1950
recordings of the second movement from Beethoven’s Sonata Op. 109 and
Schumann’s Romance Op. 28 No. 2. As Crutchfield explains, “Eibenschütz almost
always uses broad arpeggios to fill out rallentando, and often uses quicker, tighter
ones to propel an upbeat back into tempo.”187 In spite of the change in taste,
Eibenschütz clearly retained late-nineteenth-century style practices.
De Lara makes frequent unnotated arpeggiations in her 1951 recording of
Brahms’s Intermezzo Op. 117 No. 1. These include arpeggiations of the chords in
the left hand at bar 3, and very noticeable separations of the octaves in the left
hand at bar 7 (Audio Ex. 3.46 ), as well as the sixths and octaves between bars

See chapter 2, page 95.
Hambourg, How to Become a Pianist, 57.
Crutchfield, “Brahms, by Those Who Knew Him,” 18.
Unnotated A r peg g iation 187

18 and 20. She also arpeggiates the chords in the left and right hands on the
second dotted crotchet beat of bar 12, producing an agogic emphasis. In bar 26,
she expresses the first chord with great delicacy by playing the lowest bass note
first and gently, but swiftly, arpeggiating the remaining notes in the left and right
hands. These unnotated arpeggiations together with frequent dislocations give
the entire performance a feeling of continuous syncopation, much in the style of
Reinecke. Although there are no earlier recordings of de Lara, she, like Eibenschütz,
does not appear to have modified her style significantly.
Early recordings reveal, therefore, that unnotated arpeggiation was, for many
pianists, an indispensable performing practice around the turn of the twentieth
century. Many of the oldest generation of pianists, whose careers reached their
peak during the second half of the nineteenth century, can be heard making fre-
quent expressive arpeggiations that are not marked in the score. Though many
pianists of the era have been accused of overindulging in the practice, this cannot
be verified. In fact, written documentation shows that the tendency to arpeggiate
was endemic throughout the nineteenth century and very probably stemmed
from earlier keyboard practices. Therefore, the frequency of unnotated arpeggia-
tion preserved in many early recordings probably gives, if anything, a glimpse of a
practice that had already passed its zenith. In this light, it is significant that the
oldest recorded pianist, Reinecke, uses it most.
Although early recordings reveal a widespread employment of unnotated
arpeggiation, contemporaneous written texts fail to document clearly its impor-
tance and characteristics. Indeed, the impression of the practice from written
texts alone does not correspond to, and in some cases is completely divergent
with, reality. Many pianists can be seen to have practiced something entirely dif-
ferent from what they advised or marked in the score. Early recordings also show
that some pianists had already adopted a more synchronous style of playing, with
far less use of unnotated arpeggiation. Changing tastes and attitudes must
account for such a change, which, however, did not take a firm grip until the
second half of the twentieth century.
Unnotated arpeggiation provides me with a wonderful means of varying
expression and texture in a wide range of repertoire formats. In my own perfor-
mances, I am using it more and more in appropriate circumstances. A good exam-
ple of this comes from a recent performance of Brahms’s Sonata in G Major Op.
78 for violin and piano. I feel that the solo piano sequence that opens the second
movement cries out for arpeggiation (Audio Ex. 3.47 ).188 Without it, a dimen-
sion of roundness and undulation is missing; Brahms’s particularly heartfelt
music can sound a little straight and bland. Arpeggiation helps bring out the
harmonic hierarchy and allows the music to breathe and to sing.

Live performance with violinist Robin Wilson, recorded November 19, 2010, in Recital Hall
East, Sydney Conservatorium of Music.
This page intentionally left blank

Metrical Rubato and Other Forms

of Rhythmic Alteration
On another occasion in 1899, I had played a Tarantella, one
of Leschetizky’s own compositions, at a party and he complained
bitterly afterwards that it wasn’t in the slightest degree Italian. As
I was only thirteen it was hardly surprising that I had no idea of
what “being Italian” was. He was very conscious of nationality and
used to talk about the slight rhythmical falsification with which a
Polonaise should be played. He said that all the German pianists,
except d’Albert, played their Polonaise accompaniments too
strictly in time, whereas they need a characteristic rubato for
repeated chords (a quaver followed by two semiquavers). Of
course, once you start pulling the rhythm about it’s very difficult
not to go too far and do it bar after bar until it becomes an irritat-
ing mannerism. But Leschetizky was very critical of this also.
—Frank Merrick (1886–1991)1

Metrical rubato is a term I have coined to describe the old bel canto type of
tempo rubato commonly described as the rhythmic alteration of melody notes
while essentially preserving the metrical regularity of the accompaniment. This
expressive device and other forms of rhythmic alteration continued to be used in
piano playing around the turn of the twentieth century. Early recordings reveal
that many pianists, in some cases entirely contrary to twenty-first-century con-
ventions, displace single melody notes or multiple adjacent melody notes within a
bar by lengthening or shortening them. In some cases, larger scale displacement
occurs from one bar to the next. The device can also be heard in the playing of
other instrumentalists and singers. This flexible placement of melodic material
often leads to asynchrony between notes of the melody and accompaniment that
are vertically aligned in the notation. Sometimes, too, there is a subtler bending
of rhythms in a style similar to the so-called Baroque practice of notes inégales,
or inequality. These practices correspond strikingly with a number of written
descriptions and musical illustrations from the second half of the nineteenth
century and earlier, though the manner in which they occur on early recordings
could hardly have been predicted from such evidence alone. On early recordings,

Merrick, “Memories of Leschetizky,” 13.

190 off the record

rhythmic alterations can be heard most frequently in music of a slow or tender

character, but can also be heard in faster music. Although these share with
dislocation and arpeggiation the characteristic of displacement, I believe they
constitute significantly different practices that warrant separate discussion.
By the second half of the twentieth century, metrical rubato was generally con-
sidered old-fashioned. Edward Sackville-West’s 1962 discussion of Rosenthal’s
performance of Chopin’s Piano Concerto No. 1 in E Minor Op. 11 highlights this
fact. Examining two examples taken from the cantabile second subject of the first
movement (Fig. 4.1a , 4.1b , 4.1c , and 4.1d ), he notes a curiosity of
Rosenthal’s rubato—namely, his habit of altering Chopin’s text by “introducing
dotted notes into phrases that were written without them.” Such liberties, he
explains, are regarded in “shockingly bad taste.” He notes that they would sound
like mistakes in the hands of contemporary pianists, being out of sync with “the
neat technique and hard-boiled style” that had by then become customary as
well as the modern attitude of absolute faithfulness to the score. Nevertheless,
Sackville-West acknowledges that these liberties characterized the phrasing choices
of pianists of an earlier generation, introduced naturally as “a kind of decoration.”2
Sackville-West describes one of numerous types of rhythmic alteration that
can be heard on early recordings. The modern attitude he mentions was increas-
ingly adopted during the twentieth century and became the hallmark of
late-twentieth-century style. Rhythmic alteration is now seldom employed in
mainstream classical performance in which synchrony between melody and
accompaniment, among other things, is taken for granted. Any significant devia-
tion from the notation is usually considered to be a mistake, a technical deficiency,
or a sign of poor taste. Yet in other spheres, such as early music, folk, jazz, and
popular music, it is an intrinsic expressive device.
Metrical rubato and other forms of rhythmic alteration can be heard in the
playing of the oldest generation of pianists on record such as Reinecke, Leschetizky,
Saint-Saëns, Grieg, and Brahms, as well as of a younger generation (Table 4.1).
Indeed, there are often similarities between what these pianists do and the advice
of contemporary and earlier sources. But their playing also reveals features too
subtle to be conveyed by written texts and gives us a sense of how much particular
artists employed rhythmic alteration. Early recordings reveal that metrical rubato
was used in Classical and Romantic repertoire in which the character and texture
of the accompaniment is sufficiently different from the melody to allow rhythmic
independence, less so in Baroque and some types of Contemporary repertoire for
which a stricter style seems to have been preferred. It is noticeable that some
pianists used metrical rubato to a lesser extent than others; their playing, which
sounds more synchronized, may represent either a divergent tradition or the first
stage of a move away from a practice that many still considered important.

Sackville-West, “Rosenthal,” 216.
Metrical R ubato and Other Forms of R hy thmic Alteration 191

Table 4.1 Some Early Recordings in Which Metrical Rubato Is Evident

Pianists Recordings

Johannes Brahms Brahms Hungarian Dance No. 1, acoustic recording, 1889

Camille Saint- Beethoven Sonata Op. 31 No. 1, 2nd movement—Adagio grazioso,
Saëns piano roll, 1905; Chopin Nocturne Op. 15 No. 2, piano roll, 1905
Theodor Chopin Nocturne Op. 27 No. 2, piano roll, 1906; Mozart Fantasia K
Leschetizky 475, piano roll, 1906
Carl Reinecke Mozart Piano Concerto K 537, 2nd movement, Larghetto arr.
Reinecke, piano roll, 1905
Edvard Grieg Grieg Sonata Op. 7, 3rd movement—“Alla Menuetto,” acoustic
recording, 1903; Humoresque Op. 6 No 2, acoustic recording, 1903
Raoul Pugno Chopin Sonata No. 2 Op. 35—Marche funèbre, acoustic recording,
1903; Chopin, Nocturne Op. 15 No. 2, acoustic recording, 1903
Landon Ronald Grieg Bridal Procession Op. 19 No. 2, acoustic recording, 1900
Vladimir de Chopin Nocturne Op. 27 No. 2, acoustic recording, 1916
Ignacy Jan Chopin Nocturne Op. 15 No. 1, acoustic recording, 1917
John Powell Chopin Nocturne Op. 27 No. 2, piano roll, 1929
Isidor Philipp Saint-Saëns Sonata No. 1 Op. 32, 2nd movement, electrical
with Paul Bazelaire recording, 1935
Carl Friedberg Schumann Symphonic Etudes, 6th Variation and Finale, 1953;
Beethoven Sonata Op. 14 No. 2, 2nd movement, 1953
Adelina de Lara Brahms Intermezzo Op. 117 No. 1, 1951
Fanny Davies Schumann Piano Concerto Op. 54, 1st movement, 1951
Etelka Freund Brahms Sonata Op. 5, 1st, 2nd, and 4th movements, 1953

Written Texts
Considering its widespread use, particularly by generations of pianists whose
careers flourished in the second half of the nineteenth century, the detailed
discourses by Lussy, Riemann, and others curiously neglect rhythmic alteration.
On the other hand, Taylor espoused in 1897 that it is “the greatest service
to expression when suitably introduced,” cautioning at the same time its
degeneration into affectation. Referring to an excerpt from Chopin’s Ballade
No. 4 Op. 52 (Fig. 4.2 ), he describes a compensating rubato (metrical rubato)
in which one part of a phrase is accelerated while another is proportionally
192 off the record

decelerated. In this, the duration of the phrase should theoretically be the same as
if it were played strictly in time. And for its success, the accompaniment “must
always keep strict time,” which might result in not a single melody note, apart
perhaps from the first in each bar, coinciding with the corresponding note in the
accompaniment. Like many others, Taylor emphasizes that this type of rubato is
“too delicate and subtle to be expressed in notation” and must rely “entirely on
the discretion of the player.”3 Taylor’s advice suggests that metrical rubato was
much valued in piano playing of his era and when artistically handled could lead
to conspicuous asynchrony between melody and accompaniment. His explana-
tion and musical example appear, however, to refer to two separate but related
practices. The first involves rhythmic alterations to melody notes by the per-
former; the second produces a similar effect of rhythmic freedom, but is notated
by the composer.
At the beginning of the twentieth century, metrical rubato was also valued
in violin playing. Referring particularly to music of earlier periods with regular
continuo-type bass lines, Joachim and Moser advise about how best to express
“the living spirit of the work.” This requires a certain freedom and, wherever the
melody demands it, a slackening of the rhythmic structure of the bar. But they
warn of the “offence against all musical feeling” when the base line, in notes of
equal time value, is altered in tempo or rhythm in order to keep in time with the
Joachim’s 1903 recordings show that he used rhythmic alteration in a range of
music, not confined to earlier repertoire. These can frequently be heard on his
recording of his Romance in C, as well as in his recordings of Brahms’s first and
second Hungarian Dance and of two movements from Bach’s Solo Sonatas and
Partitas.5 Of course, for the solo Bach movements, his alterations are understood
when one imagines a bass line or underlying pulse.
Christiani preceded Taylor in describing various types of alteration that come
under the general heading of rubato. In 1885 he advised the following:

1. Any temporary retardation or acceleration is rubato.

2. Any negative grammatical accentuation (for example, syncopation), by which
the time becomes robbed of its regular accents, is a rubato.
3. That capricious and disorderly mode of performance by which some notes are
protracted beyond their proper duration and others are curtailed, without,
however, changing the aggregate duration of each measure, is a rubato.

Taylor, Technique and Expression, 72–73.
Joachim and Moser, Violinschule, vol. 3, 16.
See Figures 4.23, 4.37, 4.38, 4.78, 4.79, 4.80, 4.81, 4.82, and 4.83.
Metrical R ubato and Other Forms of R hy thmic Alteration 193

Attributing the third means to Chopin, Christiani praises its artistic value and
beauty when used with propriety, but considers it “ugly and pernicious when out
of place, or exaggerated.” He advises that these types of rubato may be executed
in two ways: both hands can accelerate and decelerate together in sympathy
(a type of tempo modification discussed in chapter 5), or the accompanying hand
keeps strict time, while the other effects rubato against it (metrical rubato). The
latter is rated as the more beautiful of the two—“the truly artistic rubato.”6
To achieve this, the pulse of the underlying accompaniment must be preserved
so that the character of certain genres—waltzes, and character pieces such as
marches, mazurkas, polonaises, barcaroles, lullabies, serenades, nocturnes,
romances, and songs without words—is not destroyed. Christiani regards this as
the basis of highly expressive performance, and compares it to the practices of
first-rate Italian opera singers who retard and accelerate “at almost every moment”
while the orchestral accompaniment proceeds steadily. It is this “general not-
giving-way of the accompaniment” that the soloist requires, expecting to coincide
perhaps only at tenuto or fermate points: “just so should the pianist keep time,
and yet be free in time.”7
Such a frequency suggests that metrical rubato imparted an improvisatory
character to the performance and was not reserved merely for special or unusual
moments. Christiani considered Thalberg’s piano playing exemplary in this
respect. Thalberg, he explains, would make a melody stand out in “bold dynamic
relief” by accelerating, retarding, or embellishing it while the accompaniment
remained steady. For Christiani, this method is rational, musical, gratifying,
attractive, and possesses peculiar charm. It is “never provoking and exasperating,
as out-of-time playing with both hands.”8
With little doubt, Christiani considered this style a model to be emulated.
Unfortunately, however, his descriptions are too general to discern the true
character of such rhythmic alterations.
The descriptions of Chopin’s playing by those in his close circle are likewise too
general to gauge how his rubato practices actually sounded. For example, Mikuli
attempts to dispel the misconception that Chopin’s playing was “distorted out of
all rhythmic form by an incessant tempo rubato”:

In keeping time Chopin was inflexible, and many will be surprised to

learn that the metronome never left the piano. Even in his oft-decried
tempo rubato one hand—that having the accompaniment—always

Christiani, The Principles of Expression, 299.
Christiani, The Principles of Expression, 300–301.
Christiani, The Principles of Expression, 301.
194 off the record

played on in strict time, while the other, singing the melody, either
hesitated as if undecided, or, with increased animation, anticipating
with a kind of impatient vehemence as if in passionate utterances,
maintained the freedom of musical expression from the fetters of strict

This view is corroborated by Mme Camille Dubois’s explanation reported by

Georges Mathias in 1882. Chopin frequently demanded “that the left hand,
playing the accompaniment, should maintain strict time, while the melodic line
should enjoy freedom of expression with fluctuations of speed. This is quite fea-
sible: you can be early, you can be late, the two hands are not in phase; then you
make a compensation which re-establishes the ensemble.”10 In 1868, Wilhelm von
Lenz put forward a similar explanation. According to Lenz, Chopin considered the
left hand as the conductor, which “must not waver, nor lose ground,” and he told
his students to “do with the right hand what you will and what you can.” Although
this manner would result in deviations to the details of a composition, the overall
length of performance remains unaltered.11 And in 1879, Kleczyński compares
Chopin’s rubato style with that of Paganini, who, when playing with an orchestra,
apparently demanded that the instrumentalists “observe the time, whilst he
himself departed from it.”12
In light of the foregoing, it can hardly be denied that Chopin used metrical
rubato. Like Christiani, however, only the general principle of rhythmic flexibility
of the melody within a metrical framework is preserved, leaving the individual
features of Chopin’s style to the imagination. As we will see later, many writers
found the difficulties of describing or notating clearly the subtleties of such varied
rhythmic nuances insurmountable.
The features of metrical rubato and other forms of rhythmic alteration in
piano playing are rarely discussed in any useful degree in written texts during the
nineteenth century. Fortunately, a more comprehensive understanding may be
gained by referring to the practices in singing. During the nineteenth century,
the application of these to piano playing was considered entirely appropriate.
For example, the precepts of the influential singing teacher Manuel García were

Mikuli, “Introductory Notes,” unpaginated [1].
Isidor Philipp, Exercises quotidiens tirés des oevres de Chopin, with preface by Georges Mathias
(Paris: 1882), 5. See also Eigeldinger, Chopin, 49–50.
Wilhelm von Lenz, “Die grossen Pianoforte-Virtuosen unserer Zeit aus persönlicher
Bekanntschaft: Liszt, Chopin, Tausig, Henselt,” Neue Berliner Musikzeitung XXII/38 (1868): 302;
republished as Die grossen Pianoforte-Virtuosen unserer Zeit aus persönlicher Bekanntschaft: Liszt, Chopin,
Tausig, Henselt (Berlin: 1872); trans. M. Baker as The Great Piano Virtuosos of Our Time: Liszt, Chopin,
Tausig, Henselt (New York: 1899); modern ed., ed. Philip Reder (London: Kahn and Averril 1983),
Kleczyński, How to Play Chopin, 57.
Metrical R ubato and Other Forms of R hy thmic Alteration 195

cited in an anthology of historical keyboard works titled Le Trésor des pianistes

(Paris, 1861–72).13 And titles such as Thalberg’s L’Art du chant appliqué au piano
clearly describe the close relationship between singing and piano playing. It is to
García’s various publications with their illuminating descriptions and annotated
musical examples that we will now turn.
García’s Traité complet de l’art du chant was published in two parts (Paris, c.
1840 or 1841/1847). A version combining both parts appeared in a condensed
English translation as Garcia’s New Treatise on the Art of Singing (London, 1857),
and in French and German two years later. Toward the end of his life, García also
published another treatise called Hints on Singing (London, 1894). Henceforth,
the original French edition will be referred to as the Traité complet, and the English
edition of 1857 as the New Treatise.
In the second part of New Treatise, García provides a general description of
metrical rubato as “the momentary increase of values, which is given to one or
several sounds, to the detriment of the rest, while the total length of the bar
remains unaltered.” This method, which redistributes the note values into pat-
terns of long and short, safeguards against monotony and can be used to give
“greater vehemence to bursts of passion.”14 Such alterations “will stand out in
relief and change the character of certain phrases” but only if the accompaniment
remains strict in time.15 García provided several interesting examples, which
together with those of other writers I have divided into three categories: namely,
small-scale alteration, inequality, and large-scale alteration.

Small-Scale Alteration: Written Texts

Small-scale alteration describes any rhythmic modification made to one or a
few adjacent notes. Sometimes this causes adjacent notes of equal value to
become significantly unequal by creating dotted or triplet figures and so on. At
other times, dotted figures are, so to speak, ironed out into equal-note figures.
Often these modifications help emphasize a particular melody note that requires
heightened expression. Unless otherwise stated, all alterations refer to melody
Before continuing with García, I want to divert for a moment to Heinrich
Germer’s Wie spielt man klavier? Op. 30 (1881). This little-known work provides
a useful point of departure for discussion about small-scale alteration in piano
playing. Noting that Mozart, Beethoven, and especially Chopin (according to their

Aristide Farrenc and Louise Farrenc, eds., Le Trésor des pianistes, 23 vols. (Paris: 1861–72;
reprint, New York: Da Capo Press, 1977).
García, New Treatise, 50.
García, New Treatise, 51.
196 off the record

contemporaries) often made an extreme use of tempo rubato, Germer details

some of the extraordinary practices associated with such a style:

1. Often only one, the melodically most important note in the bar, will be
prolonged. Because the accompanying hand goes on strictly in the meter—it
is described, for example of Mozart as an admired characteristic—one has to
rob the following notes of their notated length as much as was given to the
lengthened note, i.e., they have to be played that much faster. This way of
rubato happens very often in the performance of melodies as also passages;
because of course an accentuated note with simultaneous lengthening will be
presented as the most important [note] because of the resulting overtones,
than when it is only emphasized by the accent.
2. Sometimes smaller groups of notes are changed rhythmically. Three notes that
for example are written as triplet quavers are executed as a quaver and two
semiquavers and vice versa; or quintuplet quavers will be interpreted as
two semiquavers, one quaver and two semiquavers. Above all, the thought of
lengthening the most important melodic notes (maybe also the highest note)
of the figure is the reason for such modification.16

Germer describes the prolongation of a single note in order to give it added

emphasis, and the rhythmic alteration of equal-value notes resulting in the pro-
longation of one or more important notes. As we will see later, his description
corresponds closely with certain practices preserved in early recordings. Several
pianists, however, made alterations that modified the original notation even more
radically. Significantly, Germer notes that the use of small-scale alteration was not
limited to Romantic repertoire but applied more widely, a practice that is also
preserved on early recordings.
García treats the subject of small-scale alteration much more comprehensively
than Germer. In the first part of his Traité complet, García highlights the temps
d’ârret (Fig. 4.3a and Fig. 4.3b)—“a momentary prolongation of the value given to
any note in a sequence formed of equal-value notes.” While lending support to the
voice, this method allows the singer “to render distinctly that which would other-
wise be passed over,” giving heightened effect to the sequence.17 In Figure 4.3a,
crosses indicate the notes to be lengthened. The resulting variants showing the
approximate effect are indicated in Figure 4.3b. Prolongation of a particular note

Heinrich Germer, Wie spielt man klavier? Op. 30, Fünf didaktische Abhandlungen über Tonbildung,
Accentuation, Dynamik, Tempo und Vortrag (Leipzig, Germany: Leede, 1881), 36. Note that the original
German text is provided in Appendix E .
Manuel García, Traité complet de l’art du chant (Paris: 1847), part 1, 49: “Le temps d’arrêt est
une prolongation momentanée de valeur donnée à une note prise au hazard dans un trait composé de
notes d’égale valeur . . . Le temps d’arrêt, en donnant un appui à la voix, lui permet de rendre distinct
ce qui aurait manqué de netteté, et les traits y gagnent beaucoup d’effet.”
Metrical R ubato and Other Forms of R hy thmic Alteration 197

Figure 4.3a The temps d’ârret, annotated by García.

Figure 4.3b The temps d’ârret showing modifications, annotated by García.

requires a compensatory modification of the notes coming before or after it.

Presumably there are cases in which notes both before and after are affected.
Clearly, García considered this the most important and basic rule because in the
second part of the Traité complet he describes it as “the first element of tempo
In the New Treatise, García shows how small-scale alterations help prevent
monotony and enhance passion with annotated examples from Donizetti’s
Anna Bolena and Rossini’s Gazza Ladra (Fig. 4.4).19 Notably, in the first example,
the sequence that precedes the altered section is in E flat major; the rhythmic
alterations therefore help to make the chromatic notes D flat and C flat more
prominent. Without García’s annotations, such features as the triplet rhythms in
the first example, or the equalizing of the back-dotted rhythms in the second
example, might never have been envisaged.
In the New Treatise, García prescribes the occasional use of small-scale
alteration to give variation to repeated passages, for example, where “the second

García, Traité complet, part 2, 24: “Le temps d’arrêt . . . est le premier élèment du tempo
García, New Treatise, 51.
198 off the record

Figure 4.4 Donizetti Anna Bolena and Rossini Gazza Ladra, showing García’s

section of a phrase is composed of the same values as the first.” On rare occasions
it was also to be used in place of dynamic and tempo changes where “the identical
thought is repeated several times in succession” or where “the thought pursues an
ascending or descending progression.”20
In the Traité complet, he provides an annotated example of this from Mozart’s
Le nozze di Figaro, in which the musical material has been previously presented
(Fig. 4.5).21 For the sake of variety, the note E on the syllable “giar” is altered to a
G and lengthened. The descending sequence of notes that follows is naturally
robbed of time. Similar alteration occurs to the note on the syllable “gra” and to
the notes that follow. Here, variation is effected not only by prolongation but
also by embellishment, though García made no mention of the latter.
In the New Treatise, García classes as “a serious error” the use of tempo modifi-
cation (ritardando) rather than tempo rubato (metrical rubato) during the
penultimate bar “in order to give spirit to the final cadences of a piece” because,
“while aiming at spirit and enthusiasm,” the result is that the singer “only becomes
awkward and dull.”22 In the Traité complet, García’s annotated example from
Rossini’s Il barbiere di Siviglia reveals prolongation of the note on a particular
syllable and subsequent diminution of those that follow to create the requisite
energizing effect (Fig. 4.6a with alteration in Fig. 4.6b).23 Hudson has pointed out
that the orchestra in a different register doubles the vocal pitches and rhythms
and that therefore the melody is heard in two different forms simultaneously.24
Again, without García’s annotation, the possibility of this type of doubling would
probably not have been realized.

García, New Treatise, 55.
García, Traité complet, part 2, 24.
García, New Treatise, 51.
García, Traité complet, part 2, 24.
Hudson, Stolen Time, 72.
Metrical R ubato and Other Forms of R hy thmic Alteration 199

Figure 4.5 Mozart Le nozze di Figaro, showing García’s alterations.

Figure 4.6a Rossini Il barbiere di Siviglia.

Figure 4.6b Rossini Il barbiere di Siviglia, showing García’s alterations.

This type of disjunction with a doubling accompaniment had historical

precedents, examples of which can be found in late-eighteenth-century sources.
Richard Maunder has pointed out two, of many, examples in Domenico Corri’s
Select Collection of the Most Admired Songs, Duetts, &c. i–iii (Edinburgh c. 1779).
The first appears in bar 19 of Corri’s arrangement of the popular accompanied
aria “Se placar” from Antonio Sacchini’s Perseo (Fig. 4.7 ),25 and the second
appears in bar 15 in Corri’s arrangement of the accompanied aria “Nel partir”
from J. C. Bach’s La Clemenza di Scipione (Fig. 4.8 ).26 Maunder explains that

Domenico Corri, A Select Collection of the Most Admired Songs, Duetts &c., 3 vols. (Edinburgh:
c. 1779), vol. 1, 49. Reprinted in Richard Maunder, Domenico Corri’s Treatises on Singing, 4 vols, ed.
Richard Maunder (New York: Garland, 1993).
Corri, A Select Collection, vol. 1, 90.
200 off the record

while Corri really only hints at the use of rhythmic alteration in his introduction,
“considerable rhythmic license is allowed not only in recitatives but also in arias,
though in the latter the accompaniment continues in strict time, and measures
are not extended.” Often this results “in the voice and accompaniment failing
to coincide exactly in what is otherwise a unison passage.”27
For García, the lengthening of particular types of notes could invest melodies
with particular color and variety. In the New Treatise, he states that such prolon-
gation is usually made “to appoggiaturas, to notes placed on long syllables, and
those which are naturally salient in the harmony.” He insists that any time
lost must be regained by shortening other notes.28 Although this usefully outlines
specific places that were habitually altered, García implies that there are other
places that might receive similar treatment. In the New Treatise, his example,
taken from Donizetti’s Lucia (Fig. 4.9), uses doubled note stems to indicate the
lengthening of the first in each group of four notes of a coloratura sequence.29
Unusually, for García, the annotation provides no sense of the actual rhythmic
relationship. Four possible interpretations of the foregoing example are given in
Fig. 4.10 . The last shows a variety of alterations with the highest melody note
lengthened to a greater extent than in the other interpretations.
The English pianist Caroline Reinagle (1818–92) discusses a similar metrical
rubato practice in A Few Words on Pianoforte Playing (1855). Giving as an example
bar 74 of the second movement of Beethoven’s Sonata Op. 10 No. 3 (Fig. 4.11),
she explains that “the fourth group of notes, and, to a less degree, the third,
seize, in right of their evidently stronger expression, on a short portion of time
not justly belonging to them, of which the remainder must be robbed.”30 Reinagle
advises that the chromatically altered notes C sharp in the third group, and
G sharp in the fourth group, ought to be lengthened expressively because of their
dissonant effect. An indication of the alteration would have made the intended
effect clearer, though perhaps, like García, Reinagle felt that the notation could
not do justice to the subtleties of the inflection.
García also considered that rhythmic alteration was useful in the preparation
of a trill, “by permitting this preparation to take place on the preceding notes.” In
the annotation to the accompanying example from Rossini’s Il barbiere di Siviglia
(Fig. 4.12), the trill commences earlier than notated, presumably to give enough
time to achieve the requisite rapidity and brilliancy.31 Significantly, Louis Spohr
made reference to this technique in 1833. As he explains, “In order to produce the

Richard Maunder, “Introduction,” Domeninco Corri’s Treatises on Singing, ed. Richard Maunder
(New York: Garland, 1993), vol. 1, x.
García, New Treatise, 51.
García, New Treatise, 51.
Caroline Reinagle, A Few Words on Pianoforte Playing (London: 1855); reprint in The Musical
Times vol. 10 (1862): 242.
García, New Treatise, 51.
Metrical R ubato and Other Forms of R hy thmic Alteration 201

Figure 4.9 Donizetti Lucia, annotated by García.

Figure 4.11 Beethoven Sonata Op. 10 No. 3, second movement, bar 74.

Figure 4.12 Rossini Il barbiere di Siviglia, showing García’s alterations.

shakes full and brilliant, the half of the value of the preceding note has been taken
and added to the shake note.”32 To illustrate this, he presents an annotated exam-
ple of the solo violin part from Rode’s Violin Concerto No. 7 (Fig. 4.13) in which
the lengths of the trills in bars 17, 18, and 19 are effectively doubled compared
with the original notation (Fig. 4.14).
In summary, García and others prescribed small-scale alterations for the

• Variation of passages containing even notes

• Intensification of passionate melodies
• Variation of repeated phrases or identical thoughts including those in
ascending or descending progressions
• Energizing of final cadences

Louis Spohr, Violinschule (Vienna: [1832]); trans. C. Rudolphus as Louis Spohr’s Grand Violin
School (London: Wessel, 1833), 183.
202 off the record

• Emphasis of single notes such as appoggiaturas, notes on long syllables, and

those of harmonic significance
• Variation of phrases
• Preparation and enhancement of trills

In addition, small-scale alterations could help effect the following:

• Prevent monotony
• Emphasize and support notes that would ordinarily be ignored
• Change the character of a melody
• Enhance the color and variety of melodies

Figure 4.13 Rode Violin Concerto No. 7, bars 17 to 19, showing Spohr’s alterations.

Figure 4.14 Rode’s original rhythms against Spohr’s alterations.

Metrical R ubato and Other Forms of R hy thmic Alteration 203

Clearly, however, these are just a few of the types of rhythmic alterations that
were expected and enjoyed by singers and instrumentalists of the time. Other
interesting examples of small-scale alterations are preserved in the Méthode de
chant, composée pour ses classes du Conservatoire (1849) by Laure Cinti-Damoreau
(1801–63). Cinti-Damoreau—considered one of the greatest singers of her era—
shows how to make vocal variations to passages from standard nineteenth-
century operas. Hudson has remarked that “some of the variants simply replace
the composer’s original with a completely new one that fits the same accompani-
ment passage. Others, however, remain close enough to the original melody for
one to recognize the technique of melodic variation.”33 It is in these that metrical
rubato alterations are clearly recognizable. Although Cinti-Damoreau did not
offer verbal explanations, her variations on an excerpt from an aria sung by the
character Isabelle in Meyerbeer’s Robert le diable demonstrate the anticipation and
delay of the original melody notes over a simple accompaniment (Fig. 4.15 ).34
Such alterations seem extreme by modern standards but were obviously consid-
ered quite normal in the mid nineteenth century.
Another annotated example of small-scale alteration in singing is found in
Jean-Baptiste Faure’s (1830–1914) La Voix et le chant: traité pratique (1886).
At the top of the varieties of rhythmic variation, he places the anticipation of
notes, singing them earlier than notated and therefore borrowing from the note
before a “little of its value.” These and other variations give the “rhythm a greater
freedom of movement” and impart to the melody “the stirring character of
improvisation.”35 His accompanying illustration (Fig. 4.16 ) shows alterations,
made in bar 3 of an excerpt from Donizetti’s Lucia di Lammermoor, by the Italian
baritone Giorgio Ronconi (1810–90).36 In the given example, Ronconi anticipates
the orchestra (which plays the rhythms of the original text) on the syllable “ta.”
Small-scale alteration to material in the accompaniment was also prescribed in
order to characterize certain musical genres, particularly—but not exclusively—
dance movements. Marie Prentner, one of Leschetizky’s teaching assistants, hinted
at this. For the “various dance rhythms . . . the waltz, polonaise, mazurka, gavotte,
menuet,” and so on, she advised that the pianist should have in depth knowledge
of their peculiarities and retain “the characteristics of race and the period of

Hudson, Stolen Time, 81.
Laure Cinti-Damoreau, Méthode de chant, composée pour ses classes du conservatoire (Paris:
1849), 97.
Jean-Baptiste Faure, La Voix et le chant: traité pratique (Paris: 1886), 182: “Parmi les variétés de
rhythm, il faut placer en première ligne les anticipations. C’est le procédé qui consiste à emprunter à un
temps un peu de sa valeur, pour la reporter sur le temps qui suit. Ce que les Italiens appellent: le tempo
rubato . . . Employées avec discernement, les anticipations laissent au rythme une plus grande liberté
d’allure et communiquent au chant, tout en lui conservant le sentiment de la mesure, le caractère
entraînant de l’improvisation.”
Faure, La Voix et le chant, 183.
204 off the record

time to which they belong.”37 One is reminded here of the “slight rhythmical
falsification” that Leschetizky promoted, as reported by Merrick at the beginning
of this chapter.38 In this regard, Breé clearly explained Leschetizky’s views:

Rhythm does not depend on a strict observance of the measure, but per-
mits on the contrary, of freer disposal over the beats, but only between
the boundaries of the bars. Thus individual beats may be abbreviated
to the profit of others, or lengthened at their expense, but not whole
measures in proportion to other measures.

Giving an example (Fig. 4.17 ), she notes that at the appearance of the
asterisk, “the quarter-note is prolonged a little at the expense of the following
eighth-note.” For waltz rhythms, she advises the opposite effect—an abbreviation
of the first beat “by accenting the bass tone in the accompaniment and rapidly
carrying it over to the second beat,” with the third beat played “somewhat
more lightly, staccato, and in exact time.” Although this gives the accompaniment
swing, Breé warns against exaggeration. For mazurkas, she explains that the
accent (with a presumably slight prolongation) sometimes falls on the first,
second, or third beats (Figs. 4.18 , 4.19 , 4.20 ). And for polonaises, “the
bass tone must be accented and then followed by a minute retardation” with
the loss of time made up in the next two semiquavers. The second and third beats
fall in time (Fig. 4.21 ).39

Small-Scale Alterations: Audible Evidence

Early recordings reveal that techniques similar to those discussed by García,
Germer, and others are exemplified in the playing of several generations of
pianists. In these recordings, a range of melodic material, from single notes to
more extended sequences of notes, is rhythmically modified, creating in some
cases conspicuous asynchrony between melody and accompaniment. In other
cases, a flexibility of rhythm and expression not inherent in the original notation
can also be heard.

Modification of Notes of Equal Value and Unequal Value

Despite the poor sound quality of Brahms’s 1889 wax cylinder recording of his
Hungarian Dance No. 1, one can hear that he significantly altered the rhythmic

Marie Prentner, Die Leschetizky Methode (1903); trans. M. de Kendler and A. Maddock as The
Leschetizky Method, ed. M. Greenwalt (London: Curwen, 1903), 73.
Merrick, “Memories of Leschetizky,” 13.
Breé, The Leschetizky Method, 70–71.
Metrical R ubato and Other Forms of R hy thmic Alteration 205

values in certain passages (Audio Ex. 4.1 ). As a result of their expert analysis,
Berger and Nichols conclude that between bars 25 and 36 containing the pattern
of a dotted crotchet followed by a quaver, “Brahms alters this group considerably
in his performance.” The subversion creates two crotchet beats instead and is
“consistent with the composer’s predilection for metric ambiguity.” Overall, the
dotted crotchet is given its full value only once, in measure 38.40 The aural
effect preserved in the recording is stark because Brahms evens out the strongly
characteristic Hungarian dance dotted rhythm. The most obvious moment
when this occurs is during bars 39 and 40 (Fig. 4.22 ).41
Joachim, too, subverted the dotted rhythms in his 1903 recording of the
same dance. This can be heard particularly in bar 21 (Audio Ex. 4.2 ). And
a similar effect is heard in his 1903 recording of his own Romance in C, where
he changes the dotted rhythm in bar 154 to equal-value notes (Fig. 4.23 ;
Audio Ex. 4.3 ).42
Numerous types of small-scale alteration are preserved in Saint-Saëns’s 1905
piano roll of the second movement of Beethoven’s Sonata Op. 31 No. 1. In bar 7,
the lengthening of the second dotted semiquaver C causes the displacement of
the following E and G (Fig. 4.24). In bar 71, containing similar material, this dis-
placement starts earlier with the lengthening of the second dotted semiquaver C
(Fig. 4.25). In bar 70, Saint-Saëns lengthens the first tied A. This results in
the delay and hemiola-like syncopated placement of the repeated A, giving it a

Figure 4.24 Beethoven Sonata Op. 31 No. 1, second movement, bar 7, Saint-Saëns,
1905, piano roll (Audio Ex. 4.4 ).

Berger and Nichols, “Brahms at the Piano,” 27. The Brahms cylinder together with denoised ver-
sions can be heard at
Johannes Brahms, “Ungarische Tänze No. 1,” Klavier-Werke (Leipzig, Germany: Breitkopf &
Härtel, c. 1890), 2.
In this and all figures that follow, the top line represents an approximation to the perceived
rhythmic alterations.
206 off the record

Figure 4.25 Beethoven Sonata Op. 31 No. 1, second movement, bar 71, Saint-Saëns,
1905, piano roll (Audio Ex. 4.5 ).

peculiarly expressive emphasis (Fig. 4.26). Saint-Saëns’s alterations in bars 53 and

54—containing identical material—cause an unusual effect that is disturbing by
modern standards (Fig. 4.27). Here, the C in the inner voice and its accompani-
ment note A flat in the second half of each bar are played approximately one semi-
quaver beat early. The resulting syncopation accentuates the poignant shift in
tonality from G to A flat. Although both melody and accompaniment are altered,
the metrical rubato effect is retained because of the preservation of the underly-
ing pulse. In a similar way, Saint-Saëns’s placement of the F sharp in bar 92,
approximately one semiquaver beat early, creates a syncopation that emphasizes
the commencement of the rising chromatic scale (Fig. 4.28). In bars 27, 65, and
91, Saint-Saëns lengthens trills to the extent that their terminations (nachschlag)
and the notes that follow are significantly displaced (Figs. 4.29 and 4.30 ;

Figure 4.26 Beethoven Sonata Op. 31 No. 1, second movement, bar 70, Saint-Saëns,
1905, piano roll (Audio Ex. 4.6 ).
Metrical R ubato and Other Forms of R hy thmic Alteration 207

Figure 4.27 Beethoven Sonata Op. 31 No. 1, second movement, bar 53, Saint-Saëns,
1905, piano roll (Audio Ex. 4.7 ).

Figure 4.28 Beethoven Sonata Op. 31 No. 1, second movement, bars 91 and 92,
Saint-Saëns, 1905, piano roll (Audio Ex. 4.8 ).

Audio Ex. 4.10 ). Although there appears to be no historical written evidence

supporting this practice, its frequent use here effects small-scale alteration that
cannot simply be a technical aberration.
Saint-Saëns’s piano roll of Chopin’s Nocturne Op. 15 No. 2 also preserves
examples of small-scale alteration. In bar 2, the F sharp is doubled in length
and the following two notes, A and D, are halved to compensate (Fig. 4.31).
Figure 4.29 Beethoven Sonata Op. 31 No. 1, second movement, bar 27, Saint-Saëns,
1905, piano roll (Audio Ex. 4.9 ).

Figure 4.31 Chopin Nocturne Op. 15 No. 2, bars 1 to 5, Saint-Saëns, 1905, piano roll
(Audio Ex. 4.11 ).

Metrical R ubato and Other Forms of R hy thmic Alteration 209

The A sharp in bar 5 is lengthened, creating a syncopation that is compensated

for by shortening the following two melody notes, F double sharp and D sharp
(Fig. 4.31). Both of these examples correspond closely with García’s temps d’ârret
cited earlier. In bar 3, the quintuplet is played in such a way as to sound like two
triplets. Within the second triplet, there is a lengthening of the D natural and a
shortening of the following C sharp (Fig. 4.31). This example corresponds closely
with Germer’s description of such practices, though Saint-Saëns’s playing shows a
more complex combination of alterations. Saint-Saëns alters the sextuplet figure
at the end of bar 12 by shortening the preceding C sharp and commencing the G
double sharp earlier. In addition, the E natural is made more expressive by length-
ening it and shortening the following D sharp (Fig. 4.32 ; Audio Ex. 4.12 ).
Similarly, the septuplet in bar 10 is modified to form a sextuplet by shortening the
preceding C sharp and commencing the following A sharp earlier (Fig. 4.32 ).
Again, the similarities with techniques mentioned by García and Germer are obvi-
ous; however Saint-Saëns’s treatment is more varied and subtle than is suggested
by their texts.
Leschetizky’s 1906 piano roll of Chopin’s Nocturne Op. 27 No. 2 reveals that
he, too, frequently made expressive small-scale alterations. In bar 31 the tied E,
embellished with a trill, is played a quaver beat early and extended beyond its
notated length. In compensation, the rising melodic figure in the second half of
the bar is played in a hurried manner resembling a quintuplet of demisemiquavers
(Fig. 4.33). Leschetizky’s own edition of the work shows that he changed Chopin’s
original notation to match his modifications (Fig. 4.34), providing incontrovert-
ible proof that this effect was fully intended and not the product of a technical
aberration.43 In bar 36, Leschetizky plays the second chord earlier than notated
(Fig. 4.35). This creates a hemiola-type rhythm that gives the chord emphasis and
is similar to the effect heard in Saint-Saëns’s playing.
In bar 57, Leschetizky gives the syncopated chord (marked con forza by Chopin)
on the third quaver beat a poignant emphasis by playing it earlier than notated
(Fig. 4.36). The chords of the following descending sequence are shortened, caus-
ing the end of the bar to sound suddenly and dramatically accelerated.
A similar effect is heard in the anticipations made by Joachim in his Romance in
C, particularly in bars 40 and 41 and between bars 116 and 120 (Figs. 4.37 ;
Audio Ex. 4.16 and 4.38 ; Audio Ex. 4.17 ). Here, the effect is of a passion-
ate and fiery snap. In her 1906 recording of Bellini’s Casta Diva, Patti makes simi-
lar anticipations creating angular rhythms in bar 45 and in the embellished
melody line in bar 46 (Fig. 4.39 ; Audio Ex. 4.18 ).
In Chopin’s Nocturne Op. 27 No. 2, Leschetizky shortens the chords of
the triplet and quintuplet figures in the second half of bar 39, creating passages
of demisemiquavers that produce the effect of a sudden accelerando (Fig. 4.40).

Chopin, “Nocturne Op. 27 No. 2,” ed. Leschetizky, 20.
Figure 4.33 Chopin Nocturne Op. 27 No. 2, bars 31 and 32, Leschetizky, piano roll,
1906 (Audio Ex. 4.13 ).

Figure 4.34 Chopin Nocturne Op. 27 No. 2, bar 31, ed. Leschetizky.

Figure 4.35 Chopin Nocturne Op. 27 No. 2, bar 36, Leschetizky, piano roll, 1906
(Audio Ex. 4.14 ).

Metrical R ubato and Other Forms of R hy thmic Alteration 211

Figure 4.36 Chopin Nocturne Op. 27 No. 2, bar 57, Leschetizky, piano roll, 1906
(Audio Ex. 4.15 ).

Figure 4.40 Chopin Nocturne Op. 27 No. 2, bar 39, Leschetizky, piano roll, 1906
(Audio Ex. 4.19 ).

In a similar way, Leschetizky creates an accelerando effect in the second half of

bars 42 and 43 (Fig. 4.41 ; Audio Ex. 4.20 ). In both bars, the highest note, G,
is extended beyond its notated length and the following triplet semiquavers are
modified into demisemiquavers. Noticeably, the three-against-two rhythm in
Chopin’s notation is radically altered in Leschetizky’s performance. The effect is
212 off the record

Figure 4.42 Chopin Nocturne Op. 27 No. 2, bar 50, Leschetizky, piano roll, 1906
(Audio Ex. 4.21 ).

further exaggerated by the slight truncation of the last beat in each bar. In the
second half of bar 50, Leschetizky shortens the values of the first two triplet
figures (Fig. 4.42). The B flat, which is in effect the resolution of the long appog-
giatura C flat, is played earlier, increasing its length and thereby giving it an
expressive quality that would have been absent had the sequence had been played
as notated. A feature common to the three previous examples is that one sequence
of equal-value notes is transformed into a sequence of equal-value notes differing
from the original.
Leschetizky’s student Powell makes abundant use of small-scale alteration
for expressive effect in his 1929 piano roll of the same work. In bar 14, the
second chord is lengthened, emphasizing its dissonant quality (Fig. 4.43 ;
Audio Ex. 4.22 ). Powell very probably adopted this from Leschetizky’s edition
(Fig. 4.44 ).44
Similarly, Powell enhances the effect of the dissonant chord on the penulti-
mate quaver beat of bar 40 by prolonging it (Fig. 4.45 ; Audio Ex. 4.23 ). In
this case, however, the alteration is not marked in Leschetizky’s edition of the
work. In bar 58, it is the consonant effect of the penultimate chord that is height-
ened by prolongation (Fig. 4.46 ; Audio Ex. 4.24 ). Additionally, in bars 58
and 59, Powell also modifies the accompaniment, creating an intermittent
dotted effect (Fig. 4.46). Although this might sound extreme or erratic to us
now, rhythmic alteration of bass rhythms was advocated by Leschetizky and his
circle as cited earlier. In bar 53, Powell rhythmically inflects the dissonant C flat,
giving it increased poignancy. And in bar 54, the second chord, also dissonant, is
significantly lengthened (Fig. 4.47 ; Audio Ex. 4.25 ). In bars 11 and 14,

Chopin, “Nocturne Op. 27 No. 2,” ed. Leschetizky, 19.
Metrical R ubato and Other Forms of R hy thmic Alteration 213

Powell lengthens the first chord in the triplet, creating as a result more noticeably
angular rhythms (Fig. 4.48 ; Audio Ex. 4.26 and Fig 4.43 ). A rather
extraordinary example of Powell’s small-scale alterations occurs in bar 36. Here,
he plays the double-note chord D–F sharp in the first half of the bar slightly
early and shortens the triplets in the second half of the bar, producing again
very angular rhythms (Fig. 4.49). Notably, these alterations are not marked in
Leschetizky’s edition of the work.

Dotting and Tripletizing

Saint-Saëns created dotted effects in his 1905 piano roll recording of the second
movement of Beethoven’s Sonata Op. 31 No. 1. In bar 15, he alters the triplet by
prolonging the D and shortening the E (Fig. 4.50). This dotted rhythm seems to
give emphasis to the poignant 7th at the point of arrival on the second dotted
crotchet beat of the bar. Saint-Saëns also alters the semiquaver octaves at the end
of bar 37, creating a dotted (long/short) figure that has the effect of giving empha-
sis to the downbeat of bar 38 (Fig. 4.51). He uses this technique again at the end
of bar 78. Here, the transformation of the equal-value melody notes F sharp and
G into a dotted figure emphasizes the angularity of the following leap of a seventh
down to A (Fig. 4.52 ; Audio Ex. 4.30 ). And in bars 12, 23, and 24 of Chopin’s
Nocturne Op. 15 No. 2, Saint-Saëns lengthens the penultimate notes, creating
dotted figures (Figs. 4.53 and 4.54).
In Pugno’s 1903 recording of the same work, he alters the quintuplet in bar 3,
creating two semiquavers and a triplet within which further alteration creates a
dotted figure (Fig. 4.55). And in bar 13, he alters the quintuplet to form a triplet
followed by two semiquavers. He also tripletizes the dotted figure on the third
quaver beat of the bar (Fig. 4.56).

Figure 4.49 Chopin Nocturne Op. 27 No. 2, bar 36, Powell, piano roll, 1929 (Audio
Ex. 4.27 ).
Figure 4.50 Beethoven Sonata Op. 31 No. 1 second movement, bar 15, Saint-Saëns,
piano roll, 1905 (Audio Ex. 4.28 ).

Figure 4.51 Beethoven Sonata Op. 31 No. 1 second movement, bar 37, Saint-Saëns,
piano roll, 1905 (Audio Ex. 4.29 ).

Figure 4.53 Chopin Nocturne Op. 15 No. 2, bar 12, Saint-Saëns, piano roll, 1905
(Audio Ex. 4.31 ).

Figure 4.54 Chopin Nocturne Op. 15 No. 2, bars 23 and 24, Saint-Saëns, piano roll
recording, 1905 (Audio Ex. 4.32 ).

Figure 4.55 Chopin Nocturne Op. 15 No. 2, bar 3, Pugno, recorded in 1903 (Audio
Ex. 4.33 ).

Figure 4.56 Chopin Nocturne Op. 15 No. 2, bar 13, Pugno, recorded in 1903 (Audio
Ex. 4.34 ).

216 off the record

Overdotting is also a characteristic of some early piano recordings. In his 1912

recording of Chopin’s Nocturne Op. 27 No. 2, La Forge overdots the rhetorical
figure in bar 20, giving it more dramatic effect (Fig. 4.57). Rosenthal does a similar
thing in his 1936 recording of the same work. But he also overdots in more gentle
moments, such as bars 10 and 26 (Figs. 4.58 and 4.59). Interestingly, the conduc-
tor of the Meiningen Court Orchestra Fritz Steinbach (1855–1916)—who worked
closely with Brahms—advised that from bars 11 to 18 of the third movement of
Brahms’s Symphony No. 1 in C Minor Op. 68 (Fig. 4.60 ), “it would be better for
the woodwinds to treat their sixteenth-notes as thirty-second-notes, so as not to
sound like sloppy triplets.”45 Clearly, this instruction put into action produces the
effect of overdotting. Brahms’s friend and biographer Max Kalbeck attests
Brahms’s absolute approval of Steinbach’s interpretations.
In his 1905 piano roll of Chopin’s Nocturne Op. 15 No. 2, Saint-Saëns modifies
dotted figures into triplets, particularly noticeable during the first half of bars 19
and 21 (Fig. 4.61). This appears similar to Brahms’s practice of altering dotted
figures into equal-value notes, mentioned earlier. In fact, tripletizing is the
most common feature of Saint-Saëns’s performance of this work. This is clearly
audible in bars 1, 3, 4, 13, 19, and 21 and similar places, in which equal-value
notes—quintuplet figures and dotted figures—are altered to form triplets
(Fig. 4.62).
Grieg modifies equal-value notes into triplet figures in his 1903 recording of
the third movement of his Sonata Op. 7. Between bars 17 and 27, the quavers
of the tenor voice are tripletized (Fig. 4.63 ; Audio Ex. 4.39 ). In bars 18, 20,

Figure 4.57 Chopin Nocturne Op. 27 No. 2, bar 20, La Forge, recorded in 1912 (Audio
Ex. 4.35 ).

Jonathan R. Pasternack, “Brahms in the Meiningen Tradition—His Symphonies and Haydn
Variations According to the Markings of Fritz Steinbach, Edited by Walter Blume: A Complete
Translation with Background and Commentary” (DMA diss., University of Washington, 2004), 34.
Figure 4.58 Chopin Nocturne Op. 27 No. 2, bar 10, Rosenthal, electrical recording,
1936 (Audio Ex. 4.36 ).

Figure 4.59 Chopin Nocturne Op. 27 No. 2, bar 26, Rosenthal, electrical recording,
1936 (Audio Ex. 4.37 ).

Figure 4.61 Chopin Nocturne Op. 15 No. 2, bars 19 to 21, Saint-Saëns, piano roll
recording, 1905 (Audio Ex. 4.38 ).

218 off the record

Figure 4.62 Chopin Nocturne Op. 15 No. 2, bars 1 to 5, Saint-Saëns, piano roll
recording, 1905 (Audio Ex. 4.11 ).

and 22, Grieg modifies the dotted figures in the right hand into triplets so that
they align with the tripletized tenor voice. In a similar way, Grieg tripletizes the
dotted figures in bars 21, 22, 25, and 26 in his 1903 recording of his Humoresque
Op. 6 No. 2 (Fig. 4.64 ; Audio Ex. 4.40 ). In Chopin’s Nocturne Op. 15 No. 1
recorded in 1917, Paderewski tripletizes dotted figures in bars 3 and 5. And he
also creates dotted figures within triplets in bars 5, 6, and 7 (Fig. 4.65).
Such assimilation between dotted and triplet figures was practiced in earlier
eras. As Brown has pointed out, in 1765, Löhlein advised that “when triplets and
dotted figures occurred together, the latter should be played with a triplet rhythm.”
He provides parallel examples from the works of Haydn, Beethoven, and Schubert
in which there is little doubt that assimilation was necessary.46 For example,
in “Erstarrung” from Schubert’s Winterreise, the notation implies assimilation
(Fig. 4.66 ).

Brown, Classical and Romantic Performing Practice, 614–21.
Metrical R ubato and Other Forms of R hy thmic Alteration 219

Figure 4.65 Chopin Nocturne Op. 15 No. 1, bars 1 to 8, Paderewski, recorded in 1911
(Audio Ex. 4.41 ).

Commencement of Trills
As noted earlier, García and Spohr described the technique of commencing trills
slightly earlier than their notated position in order, it seems, to make them more
brilliant. This can clearly be heard in bars 3, 27, 29, 65, 67, 91, 93, 99, and 104 of
Saint-Saëns’s 1905 piano roll of the second movement of Beethoven’s Sonata Op.
31 No. 1. In bars 27 (Audio Ex. 4.9 ), 29, 91 (Audio Ex. 4.8 ), 99, and 104, the
aural effect is that the trill continues on from the preceding figure. This slightly
220 off the record

early commencement seems to serve as a technical means of energizing the trill

before the entrance of the corresponding accompaniment. Although by today’s
standards the effect in Saint-Saëns’s playing sounds sloppy, the written historical
evidence indicates that this was probably wholly intentional. On the other hand,
without his recording, this effect (presumably one of many created by the early
commencement of trills) could scarcely have been envisaged.
The early commencement of trills is also heard in bar 32 of Bellini’s Casta Diva
(Fig. 4.67 ; Audio Ex. 4.42 ) recorded by Patti in 1906. Here, instead of
Bellini’s extended turn, she sings a rapid trill, starting significantly earlier and
on a higher note than the one notated. This allows her to show off her technical
agility and at the same time to effect a metrical rubato.
In summary, the foregoing examples reveal a range of small-scale alterations,
the features of which would not have been discernible from written texts alone.
Typically, these alterations include the following:

• Modification of equal-value notes to different notes of equal value

• Modification of unequal-value notes to equal-value notes
• Tripletizing of figures that were originally equal-value or dotted
• Creation of dotted figures from equal-value notes
• Overdotting of certain notes
• Commencement of trills before their notated position

Inequality: Written Texts

Another type of small-scale alteration—sometimes referred to as inequality, or
notes inégales—in which equal-value notes are played slightly unequally, must
also be considered here. Today, such a style is generally associated with music
predating the Classical period or in folk, jazz, or popular music. But such altera-
tions were also considered important during the nineteenth century. As Brown
points out, Bériot troubled to describe “the subtle flexibility that a musician of the
mid-nineteenth century, whether singer or instrumentalist, might have been
inclined to introduce into passages of equal-length notes.”47 Under the title of
“Syllabation,” and with reference to two examples (Fig. 4.68 ), Bériot advises
in 1858 the following:

In very soft music, composers do not always mark the long and short
notes, for fear that the song could take too rhythmical a form. In such
cases they leave to the singer the care of marking the syllables with that
infinite delicacy that lends so great a charm. So, for instance, if we sang
with absolute equality the two quavers that begin each bar of the following

Brown, Classical and Romantic Performing Practice, 162–63.
Metrical R ubato and Other Forms of R hy thmic Alteration 221

Romance [Hérold’s Pré Aux Clers], our diction would be flat and cold.
But if the composer had written those notes as dotted notes this sweet
song would be too jerky in effect and would agree only a little with the
sentiment of its poem. It is here that a medium form is required, which
the feelings alone can understand, and which no sign can express. It is
sufficient for the first quaver to be a little longer than the second and that
the small interval which separates them should be almost insensible.48

This type of rhythmic inflection of adjacent notes had strong historical precedents.
As early as 1550, Loys Bourgeois (c. 1510–60) describes how to make inequality
in singing sequences of semiminims (crotchets). One should “sing them two by
two, dwelling some little bit of time longer on the first, than on the second—as
though the first had a dot and the second were a fusa [quaver].” This manner cre-
ates a musically effective hierarchy by lengthening the first note, which is a conso-
nance and shortening the second note, which is most often a dissonance. Bourgeois
also recommends this manner of execution because the notes sound more grace-
ful than if sung equally.49 As noted by Hudson, in 1565 Thomás de Santa María
(c. 1510–70) gives examples in which either the first or the second note of a pair
is lengthened. In 1602, Giulio Caccini (1551–1618) states that passages per-
formed in this way are more graceful. Frescobaldi recommends in 1616 that in
passages in which one hand plays quavers against semiquavers in the other, the
second of each pair of semiquavers should be “somewhat dotted.” And, Couperin
advises in 1713 that the second of the two slurred quavers in a coulé should be
Indeed, in 1717, Couperin noted deficiencies in the French way of notating
music that correspond to those in writing the French language. “We write differ-
ently to the way we play,” he opines, and therefore foreigners “play our music less
well than we play theirs.” He feels that the French are enslaved by a custom in
which sequences of crotchets in stepwise motion are generally swung or played

Bériot, Méthode, 211: “Dans la musique très douce, les auteurs ne marquent pas toujours les
longues at les brèves de peur que le chant ne prenne une expression trop rhythmée. Ils laissent dans ce
cas aux chanteurs le soin de la syllaber avec cette délicatesse infinie qui doit en faire le charme. Ainsi
par exemple si l’on disait avec une égalité absolue les deux croches qui commencent chaque mesure
de la romance suivante, cette diction serait plate et décolorée. Cependant si l’auteur avait écrit ces
deux notes en notes pointées, ce chant suave aurait une couleur trop saccadée peu en rapport avec le
sentiment qu’il exprime. C’est ici qu’il faut un terme moyen que le sentiment seul peut comprendre;
mais qu’aucun signe ne peut exprimer. Il suffit que la première croche soit un peu plus longue que la
deuxième et que le petit intervalle qui les sépare soit presqu’insensible.”
Loys Bourgeois, Le droict [sic] chemin de musique (1550); trans. in Stephen E. Hefling in Rhythmic
Alteration in Seventeenth- and Eighteenth-Century Music (New York: Schirmer, 1993), 3.
Tomás de Santa María, Arte de tañer fantasia (1565); Giulio Caccini, Le nuove musiche (1602);
Girolamo Frescobaldi, Toccate e partite d’intavolatura (1616); François Couperin, Pièces de Clavecin,
Premier livre (1713); in Hudson, Stolen Time, 26.
222 off the record

unequally although they are notated as equal-value notes. The Italians, by

contrast, notate their music in the rhythms that they truly play.51 Whether the
latter was really the case, however, is open to debate. In this respect, Hudson
explains that during this period, all countries developed individual concepts of
accentuating “good” notes in a series of even notes by slightly lengthening them
and de-emphasizing the “bad” notes by shortening them. He notes, too, that some
theoretical sources give an approximation to this style by notating a dotted note.
But in reality, the lengths “might actually be more or less than a dotted note, and
in any event would vary.”52 This concept of good and bad notes was clearly described
in the eighteenth century by Quantz, who advised that the player

must know how to make a distinction between the principal notes,

ordinarily called accented or in the Italian manner, good notes, and those
that pass, which some foreigners call bad notes. Where it is possible, the
principal notes always must be emphasized more than the passing. In
consequence of this rule, the quickest notes in every piece of moderate
tempo, or even in the Adagio, though they seem to have the same value,
must be played a little unequally, so that the stressed notes of each figure,
namely the first, third, fifth, and seventh, are held slightly longer than
the passing, namely the second, fourth, sixth, and eighth, although this
lengthening must not be as much as if the notes were dotted.53

The similarity between this and Bériot’s description above is obvious, although
Bériot does not make a rule of it.
The use of inequality in the performance of jazz has been well explained by
Mick Hammer, who asserts that “no one plays music exactly as it is written, just
as no two people would read a passage from a book the same way.” In his opinion,
“real musicians shorten one note, lengthen another, delay a third and accent
notes,” all of which adds to the individuality of the artist. Jazz musicians take
such things to extremes:

Faced with a row of eighth notes on a sheet of music a straight musician

plays a series of more or less equal notes. A jazz musician plays the eighth

Couperin, L’Art de toucher le clavecin, 39–40: “Il y a selon moy dans notre facon décrire la musique,
des deffauts qui se raportent à la manière d’écrire notre langue. C’est que nous écrivons diffèrement
de ce que nous éxècutons: ce qui fait que les étrangers joüent notre musique moins bien que nous ne
fesons la leur. Au contraire les Italiens écrivent leur musique dans les vrayes valeurs qu’ils l’ont pensée.
Par exemple, nous pointons plusieurs croches de suites par degrés-conjoints; Et cependant nous les
marquons égales; notre usage nous a asservis; Et nous continüons.”
Hudson, Stolen Time, 26–27.
Johann J. Quantz, Versuch einer Anweisung die Flöte traversiere zu spielen (Berlin: 1752); trans.
Edward R. Reilly as On Playing the Flute (1966), 2nd ed. (London: Faber, 1985), 123.
Metrical R ubato and Other Forms of R hy thmic Alteration 223

notes alternately long and short. The long note coincides with the basic
beat, the note clipped short is off the beat. There is a similar but less
tendency to play notes long and short in folk and baroque music as well
as in popular music.54

Inequality: Audible Examples

Some early recordings preserve the use of inequality in the manner described by
Bériot and other writers cited earlier. In the passage between bars 15 and 17 in his
1905 piano roll recording of his arrangement of Mozart’s Larghetto, Reinecke
lengthens and shortens adjacent notes in various combinations (Fig. 4.69).
The rhythmic relationship is too subtle to notate with any accuracy but sounds
approximately in the proportion 3:2. In the following example, the letters L and S
represent lengthened and shortened notes. Notably, in bars 15 and 17, the pat-
tern of short followed by long notes gives the effect of a Lombardic rhythm, or
“scotch snap.” Very similar inequality is heard when the material is repeated
between bars 78 and 80, showing that this was not an aberration of some
sort, but an intended rhythmic inflection. Reinecke’s use of inequality is again
evident in the descending sequence of quavers in bars 31 and 33 (Fig. 4.70 ;
Audio Ex. 4.44 ) and in the descending semiquaver sequences at the end of bars
46 and 48 (Fig. 4.71 ; Audio Ex. 4.45 ).
Planté is another pianist who uses inequality. This is particularly noticeable in
his interpretation of the adjacent semiquaver melody notes in the “Etwas lang-
samer” section (bars 33 to 62) of Schumann’s Romanze Op. 32 No. 3 recorded in
1928. Here, the falling melodic sequence is played with more or less the following
rhythmic nuance on each appearance (Fig. 4.72 ; Audio Ex. 4.46 ).
At the appearance of the portato sign at bars 2, 4, 28, and 30 in Beethoven’s
Sonata Op. 31 No. 1, Saint-Saëns makes subtle inflections by lengthening and
shortening notes in the sequence of chromatically rising semiquavers (Fig. 4.73).
Saint-Saëns’s interpretation of the portato may have some link with the instruc-
tions of Adam and Pollini mentioned in chapter 2, in which they advise that
melody notes ought to be somewhat displaced. Although their examples show
melody notes delayed after the corresponding note of the accompaniment, the
underlying principle of inflection is clear. But Saint-Saëns playing certainly cor-
responds very closely with Bériot’s description of inequality, and the proportions
he uses are almost exactly those recommended for eighteenth-century inequality
as described, for example, by Quantz.

Mark Hamer, “It Don’t Mean a Thing if It Ain’t Got That Swing. But What Is Swing?” New Scientist
(December 2000): 48.
Figure 4.69 Mozart Larghetto arr. by Reinecke, bars 15 to 17, Reinecke, piano roll,
1905 (Audio Ex. 4.43 ).

Figure 4.73 Beethoven Sonata Op. 31 No. 1, second movement, bar 2, Saint-Saëns,
piano roll, 1905 (Audio Ex. 4.47 ).

Metrical R ubato and Other Forms of R hy thmic Alteration 225

Large-Scale Alteration: Written Texts

I use the term large-scale alteration to describe the displacement of melodic
material over extended sequences within a bar, and from one bar into the next. In
his New Treatise, García describes this style promoted by his father Manuel del
Pópulo Vicente Rodriguez García (1775–1832) and the violinist Niccolò Paganini
(1782–1840), both of whom apparently excelled at it. Although the orchestral
accompaniment remained steadily in time, “they would abandon themselves to
their inspiration” until a change of harmony or to the end of the phrase. This style
was really successful only in places where the harmony did not change, or at most
was only slightly varied. And it required of the player “an excellent perception of
rhythm, and great self-possession.”55 To illustrate his point, García provides an
annotated example from Rossini’s Il barbiere di Siviglia (Fig. 4.74).56 García makes
it clear that the introduction of such extended rhythmic alterations was largely
dependent on the underlying rate of change of the harmony. His annotated exam-
ple reveals the lengthening of the highest notes toward the phrase climax and the
creation of poignant suspensions. Continuous rhythmic alteration results in the
extraordinary displacement of melodic material from the first bar into the second
bar and further displacement into the third bar. This is the only extensive example
I have found in nineteenth-century texts illustrating the practice. In García’s
example, the addition of notes at the end of bar 2 and throughout bar 3, causing
more extensive alteration to the original melody, is noticeable. And in bar 4, the
trill is clearly notated commencing much earlier than Rossini notated.
García’s annotation of large-scale alteration reveals practices that would have
been very difficult to describe. It is tantalizing that he provided only one example;
the plethora of other effects that may have been produced can only be imagined.
Acknowledging the impossibility of notating the subtleties of such alterations,
García admits that his annotations are only an approximation of the tempo rubato
his father employed.57

Figure 4.74 Rossini Il Barbiere di Siviglia, showing García’s alterations.

García, New Treatise, 51.
García, New Treatise, 51.
García, New Treatise, 51.
226 off the record

To conclude, García warns in the New Treatise that when used inappropri-
ately—“affectedly, or without discretion”—large-scale alteration (tempo rubato)
destroys the balance in the music and distorts the melody unacceptably.58 But
how are we to evaluate this comment from today’s standpoint? In the large-scale
displacements shown in his example, the original melody is distorted in ways
that would now undoubtedly be considered extreme or even grotesque. Imagine
how much more alien we would find the practices that García found affected or

Large-Scale Alteration: Audible Examples

Large-scale alterations can be heard in Patti’s 1906 recording of Bellini’s Casta
Diva. In bar 20 (Fig. 4.75 ; Audio Ex. 4.48 ), bar 41 (Fig. 4.76 ; Audio Ex.
4.49 ), and bar 44 (Fig. 4.77 ; Audio Ex. 4.50 ), she alters noticeably the
position of notes in descending sequences. Each version is slightly different,
giving variety to the repeated material. In the case of bar 41 (Fig. 4.76), Patti also
employs melodic variation. And in a similar way to García’s example, Patti dis-
places the last note of bar 44 into bar 45 (Fig. 4.77). The subtle inflections of
Patti’s alterations may demonstrate the style that García expected, but that could
be notated only in an approximate fashion. At the same time, García’s annotation
provides an historical context for Patti’s alterations.
Joachim also made large-scale alterations that displaced melodic material from
one bar into the next in his 1903 recording of his Romance in C. This is particularly
clear between bars 67 and 68 (Fig. 4.78 ; Audio Ex. 4.51 ), and bars 133 and
134 (Fig. 4.79 ; Audio Ex. 4.52 ). Elsewhere, he used anticipation rather than
delay to displace material from one bar to another. Shortening of the note values
(effecting an accelerando) in the first bar causes notes at the beginning of the
second bar to arrive early. This can be heard between bars 19 and 20 (Fig. 4.80 ;
Audio Ex. 4.53 ), 23 and 24 (Fig. 4.81 ; Audio Ex. 4.54 ), 41 and 42
(Fig. 4.82 ; Audio Ex. 4.55 ), 52 and 53 (Fig. 4.83 ; Audio Ex. 4.56 ), and
in bar 66 (Fig. 4.78 ).
Similar large-scale alterations are preserved on early piano recordings. For
example, Saint-Saëns’s 1905 piano roll of the second movement of Beethoven’s
Sonata Op. 31 No. 1 shows that he substantially alters the position of notes in the
descending figure of the final dotted crotchet beat in bar 13. Here, the lengthen-
ing of the B causes ongoing displacement (Fig. 4.84). The identical material in bar
77 both the B and the following A are lengthened, whereas the F sharp is short-
ened and the E is displaced into bar 78 (Fig. 4.85).

García, New Treatise, 50–51.
Metrical R ubato and Other Forms of R hy thmic Alteration 227

Figure 4.84 Beethoven Sonata Op. 31 No. 1 second movement, bar 13, Saint-Saëns,
piano roll, 1905 (Audio Ex. 4.57 ).

Figure 4.85 Beethoven Sonata Op. 31 No. 1 second movement, bars 77 and 78,
Saint-Saëns, piano roll, 1905 (Audio Ex. 4.58 ).

In chapter 2, I discussed Leschetizky’s approach to playing the sequential

melodic material in the second half of bar 6 of Chopin’s Nocturne Op. 27 No. 2,
preserved in his 1906 piano roll of the work. There I proposed that he was dislo-
cating many of the notes. But another plausible explanation for this particular
example is that he made metrical rubato alterations. Indeed, he may have used
both techniques in tandem. Whatever the explanation, the C in the second half of
bar 6 sounds somewhat lengthened, causing continuous displacement into bar 7
(Fig. 4.86). A similar technique is used again in the second half of bar 8, causing
displacement into bar 9.
228 off the record

Figure 4.86 Chopin Nocturne Op. 27 No. 2, bars 6 to 9, Leschetizky, piano roll, 1906
(Audio Ex. 4.59 ).

Significantly, in Leschetizky’s edition of the work, he notated the rising scale in

the second half of bar 8 to be played cantando. For many composers, these types
of expressions provided a coded message to make, among other things, expressive
alterations. Apparently, this was the practice of the composer/pianist Jan Dussek
(1760–1812). According to an anonymous contemporary review in Le Pianiste
(March 1834),

Dussek very much liked the Rubato, although never wrote the word in
his music; Dussek tried to make it visible by means of [notating] synco-
pation; but, when one faithfully executed these syncopations, one was
far from rendering his suave and delectable manner. He renounced this
method, and contented himself with writing the expression espressivo.59

Anon., Le Pianiste vol. 1, no. 5 (Mar. 1834): 78: “Dussek, qui aimant beaucoup le Rubato, quoiqu’il
n’ait jamais écrit ce mot dans sa musque; Dussek avait essayé de le rendre visible au moyen des
Metrical R ubato and Other Forms of R hy thmic Alteration 229

Clearly, in this case, a strict adherence to the letter of Dussek’s scores (in other
words, his notated syncopations) would produce a result rather different
from what he intended. Of particular interest here is that expressions such as
espressivo, and presumably others, were used in Dussek’s era to indicate practices
that are no longer in general use. There is an obvious parallel between this and
Corri’s advice cited in chapter 3 that certain musical terms indicated the use of
frequent arpeggiation.
Like Leschetizky and Saint-Saëns, Pachmann made alterations effecting larger
scale displacement. This can be heard clearly in bar 28 (Fig. 4.87) and bar 30
(Fig. 4.88) of Chopin’s Nocturne Op. 27 No. 2 recorded in 1916.
Saint-Saëns’s delay of the termination of trills effecting large-scale alteration is
particularly evident in his 1905 piano roll of the second movement of Beethoven’s
Sonata Op. 31 No. 1. In bar 65, the final C is clearly displaced into bar 66. Similar
displacement occurs between bars 67 and 68 (Fig. 4.89) and again between bars
91 and 92. In bar 71, Saint-Saëns lengthens the first dotted semiquaver C, displac-
ing the following four notes. Further to this, the penultimate note F is also length-
ened, causing the displacement of the final note B into bar 72 (Fig. 4.90). And in
bars 93 and 94, Saint-Saëns makes an extraordinary large-scale alteration. Here,
the G in bar 93 is lengthened to the extent that part of the following descending

Figure 4.87 Chopin Nocturne Op. 27 No. 2, bars 28 and 29, Pachmann, recorded in
1916 (Audio Ex. 4.60 ).

syncopes; mais, lorsqu’on exécute fidèlement ces syncopes, on était bien loin de rendre sa manière
suave et délicieuse. Il y renonça lui-même, et se contenta d’écrire espressivo.”
Figure 4.88 Chopin Nocturne Op. 27 No. 2, bars 30 and 31, Pachmann, recorded in
1916 (Audio Ex. 4.61 ).

Figure 4.89 Beethoven Sonata Op. 31 No. 1 second movement, bars 65 to 68,
Saint-Saëns, piano roll, 1905 (Audio Ex. 4.62 ).

Figure 4.90 Beethoven Sonata Op. 31 No. 1 second movement, bars 71 and 72,
Saint-Saëns, piano roll, 1905 (Audio Ex. 4.63 ).

Figure 4.91 Beethoven Sonata Op. 31 No. 1, second movement, bars 93 to 94,
Saint-Saëns, piano roll, 1905 (Audio Ex. 4.64 ).

232 off the record

tirade is significantly displaced into bar 94 (Fig. 4.91). There is an obvious

correspondence between this and García’s annotation of large-scale alteration
cited earlier.
Similarly, Saint-Saëns’s 1905 piano roll of Chopin’s Nocturne Op. 15 No. 2
demonstrates his use of large-scale alteration. In bars 18 and 20, he broadens the
descending tirade grace notes, substantially displacing the final A. He also alters
the dotted figure in the second half of bar 21, displacing the final note into bar 22
(Fig. 4.92). The aural effect is somewhat like a hemiola across the bar line. And in
the second half of bar 23, Saint-Saëns broadens the end of the quintuplet, displac-
ing the final note into bar 24 (Fig. 4.93).
In Leschetizky’s 1906 piano roll of Chopin’s Nocturne Op. 27 No. 2, he broad-
ens the notes toward the end of the fioritura in bar 52. This results in the displace-
ment of the final C into bar 53, creating an expressive appoggiatura (Fig. 4.94).
In his 1929 recording of the same work, Powell lengthens the penultimate
note of bar 21, causing the final note to be displaced into bar 22 (Fig. 4.95 ;

Figure 4.92 Chopin Nocturne Op. 15 No. 2, bars 18 to 22, Saint-Saëns, piano roll,
1905 (Audio Ex. 4.65 ).
Metrical R ubato and Other Forms of R hy thmic Alteration 233

Figure 4.93 Chopin Nocturne Op. 15 No. 2, bars 23 and 24, Saint-Saëns, piano roll,
1905 (Audio Ex. 4.66 ).

Figure 4.94 Chopin Nocturne Op. 27 No. 2, bars 52 and 53, Leschetizky, piano roll,
1906 (Audio Ex. 4.67 ).

Audio Ex. 4.68 ). And at bars 28 and 30, Powell lengthens the penultimate notes
of the compound embellishments derived from Leschetizky, causing the displace-
ment of the final notes into the following bar (Fig. 4.96 ; Audio Ex. 4.69 ).

The Problems Inherent in Descriptive Language

and Musical Notation
The foregoing examples reveal features of practices that could hardly have
been surmised from written texts alone. In the case of piano playing, metrical
rubato received a cursory treatment in contemporary written documentation.
234 off the record

The complex and varied small- and large-scale alterations that important
late-nineteenth- and early-twentieth-century pianists made for expressive pur-
poses are simply not codified, though the similarities with references to vocal
practice are clear. The difficulty of conveying the features of such alterations with
musical notation and/or descriptive terminology was noted on several occasions
during the nineteenth century. Writing in the Harmonicon in 1823, an anonymous
author mentioned the many “delicate shades in music” that cannot be described
in writing and must “be learnt and felt by the genius and practise [sic] of a
performer.” Among these are “the tempo rubato, or occasional retardation of
the time for the purpose of enforcing the expression.” Such shades are only effec-
tive in “the hand of a master.”60 At the turn of the twentieth century Edwin H.
Lemare (1865–1934) reiterated a similar sentiment in “The Art of Organ Playing”
(c. 1900). For him, that the art of rubato “is so subtle and almost mystic that it
is very difficult, and well-nigh impossible in writing, to give much help to the
Dussek, we have seen, abandoned his attempts to notate metrical rubato
because he felt that its numerous inherent subtleties could not be sufficiently
conveyed by notation. In 1834, Pierre Baillot (1771–1842) noted that few com-
posers had indicated it in their scores. He warned, too, that it could be notated
only to a certain degree and that “like all impassioned accents, it will lose a lot of
its effect if it is executed cold bloodedly.”62 And we have seen that García, whose
examples of alterations are extremely informative, makes it clear that these are
approximations to actual practice.
Other writers felt that earwitness experience was essential for the apprecia-
tion of the subtleties of metrical rubato. In the Introduction to Le Trésor des
pianistes (1861), Aristide Farrenc (1794–1865) finds García’s musical examples to
be insufficient because of the plethora of “combinations and nuances of value
which cannot be notated.” To appreciate these, one has to hear “a great virtuoso.”63
Concerning the latter point, Théophile Lemaitre offered a similar opinion in his
translation of Tosi’s Opinioni (1874). He describes the great effect of Italian sing-
ers who make tasteful and artistically sophisticated delays or lose “the precision

Anon., “On Musical Colouring,” The Harmonicon vol. 1, no. 11 (1823): 162.
Edwin H. Lemare, “The Art of Organ Playing,” The Musical Educator vol. 4 (c. 1900): xiii–xiv.
Pierre Baillot, L’Art du violon: nouvelle méthode (Paris: 1834); facs. reprint, Méthodes et Traités:
Violon Vol. 3 (Courlay, France: Fuzeau, 2001), 136: “On peut noter, jusqu’à un certain point cet artifice,
mais, comme tous les accens passionnés, il perdra beaucoup de son effet à être execute de sang froid.”
See also Robin Stowell in Violin Technique and Performance in the Late Eighteenth and Early Nineteenth
Centuries (Cambridge: Cambridge University Press, 1985), 274.
Farrenc, Le Trésor des pianistes, vol. 1, 4: “Quant aux examples qui accompagnent ses precepts,
je dirai qu’ils sont insuffisants, car il y a cet artifice de l’exécution des combinations et des nuances de
valeur qui ne peuvent [pas] se noter; il n’y a que l’audition d’un grand virtuose qui puisse en donner
Metrical R ubato and Other Forms of R hy thmic Alteration 235

of the time at will,” while the orchestra remains undisturbed. But he admits that
it is impossible to give a notated example of this: “it is necessary to observe it in
Whereas Dussek eventually adopted the musical term espressivo to denote met-
rical rubato, Chopin felt that musical terminology fell short of capturing its
essence. In Frédéric Chopin (1852), Liszt used highly poetic analogies to describe
Chopin’s idiosyncratic manner of expressing the melody “which gave so individual
an impress to his virtuosity.” Liszt notes that Chopin, in his early career, indicated
this by writing the term tempo rubato. But he eventually relinquished any verbal
indication after realizing that the terminology “taught nothing to him who knew,
said nothing to him who did not know, understand, and feel.” In Liszt’s opinion,
Chopin became persuaded “that if one understood it, it was impossible not to
divine this rule of irregularity.”65 Presumably, understanding of it came through
firsthand experience of Chopin’s playing or a highly developed artistic sense based
on knowledge of accepted practice.
Although Chopin renounced the use of terminology, it is clear that he tried to
notate metrical rubato alterations in his compositions. This met with severe oppo-
sition that highlighted the weakness inherent in the notation of such alterations.
The anonymous reviewer in Le Pianiste (March 1834) felt that Chopin’s inspired
works had been ruined by

a manner of affectation to write the music almost as it should be played

(we say almost, for completely is impossible)—to write this swaying, lan-
guid, groping style, this style which no known arrangement of note
values can well express; the Rubato—the terror of young women, the
bogeyman of beginners.66

Other texts warn, too, that musical notation is imperfect in preserving the subtle
inflections of metrical rubato. In La voix et le chant (1886), Faure describes the

Théophile Lemaitre, L’Art du chant traduit de l’Italien (Paris: 1884); trans. Reginald Gatty as
“Tempo Rubato,” The Musical Times vol. 53, no. 829 (Mar. 1912): 161.
Franz Liszt, Frédéric Chopin (1852); 2nd ed. (1879); trans. Frederick Niecks as Chopin as a Man
and Musician 2 vols. (London: Novello, 1888), vol. 2, 101: “He [Chopin] always made the melody undu-
late like a skiff borne on the bosom of a powerful wave; or he made it move vaguely like an aerial appa-
rition suddenly sprung up in this tangible and palpable world . . . tempo rubato: stolen, broken time—a
measure at once supple, abrupt, and languid, vacillating like the flame under the breath which agitates
it, like the corn in a field swayed by the soft pressure of a warm air, like the top of trees beat hither and
thither by a keen breeze.”
Anon., Le Pianiste vol. 1, no. 5 (1834); trans. Hudson, Stolen Time, 190: “une sorte d’affectation
à écrire la musique Presque comme il faut l’exécuter,—(nous disons Presque, car tout-à-fait est
impossible.)—à écrire ce genre balance, languissant, tâtonné, ce genre qu’aucun arrangement de
valeurs connues ne peut bien exprimer; le Rubato enfin, ce Rubato l’effroi des jeunes filles, le Croque-
Mitaine des mazettes!”
236 off the record

practice of anticipation, which when skillfully employed gives the rhythm “a greater
freedom of movement” and the melody “an arousing improvisatory character”
while maintaining a sense of pulse. “One could arguably write the anticipations
as one does syncopations, with which they offer some analogy,” he explains, “but
this would be giving it the form rather than the spirit.”67 This observation is
especially significant when one considers the numerous compositions in which
syncopations appear to have been used to represent a certain rhythmic freedom
in the melody line. Whether in such cases the composer expected the melody to be
played exactly as written, or simply intended the notation to convey its approxi-
mate position, remains conjecture. Certainly, the current penchant for strict
adherence to musical texts results in a literal interpretation that does not always
sound free.
The foregoing references show that by their nature, written texts could not
meaningfully preserve important features of metrical rubato. This may account
for the conspicuous lack of detail and information contained in them. There is
no doubt, however, that metrical rubato was a primary method of enhancing

Historical Precedents
The practices of metrical rubato had historical precedents dating from as early
as the fourteenth century. The Robertsbridge Fragment (c. 1320) preserves
“instrumental intabulations of vocal music” in which a more elaborate line is cre-
ated by the rhythmic alteration of the original notation. Later sources such as
Sylvestro di Ganassi’s (1492–mid-sixteenth century) Opera intitulata Fontegara
(Venice, 1535) and a version of the madrigal Anchor che col partire (Bovicelli, 1594)
by Cipriano de Rore (1515/1516–65) show alterations made through embellish-
ment. And during the seventeenth century, various sources show the displacement
of melodic material through anticipation, delay, and embellishment.68
By the eighteenth century, the principle of metrical rubato as rhythmic free-
dom in the melody against a strictly rhythmic accompaniment was certainly
established. It was generally prescribed for sophisticated performance, and always
to be guided by good taste and experience. In 1723, the castrato Pier Francesco
Tosi (1653–1732) noted in his Opinioni de’cantori antichi e moderni that it was

Faure, La Voix et le chant, 182: “À la rigueur, on pourrait écrire les anticipations comme on le fait
pour les syncopes, avec lesquelles elles offrent quelque analogie; mais ce serait en donner la lettre et
non l’esprit. Employées avec discernement, les anticipations laissent au rhythme une plus grande lib-
erté d’allure et communiquent au chant tout en lui conservant le sentiment de la mesure, le caractère
entraînant de l’improvisation.”
Hudson, Stolen Time, 13–14.
Metrical R ubato and Other Forms of R hy thmic Alteration 237

imperative for singers to know how “to steal the time” and to make restitution
ingeniously in music of a pathetic character.69 John Ernst Galliard amplified this
idea in his 1742 translation of Tosi’s text. The upper part “retards or anticipates in
a singular manner, for the sake of expression but after that returns to its exact-
ness, to be guided by the bass,” a practice most appropriate for vocalists or solo
instrumentalists in music of the pathetic or tender style.70 While confirming the
practice, the general nature of Tosi’s and Galliard’s words leaves the intended aural
effect to the imagination of the reader.
Importantly, Galliard provided realizations of a fascinating vocal ornament
referred to by Tosi as the drag or strascino (Fig. 4.97 ). According to Tosi, “there
is no invention superior or execution more apt to touch the heart.” To effect this
ornament, “a singer begins with a high note, dragging it gently down to a low one,
with the forte and piano, almost gradually with inequality of motion (that is to
say, stopping a little more on some notes in the middle, than those that begin or
end the strascino or dragg [sic]).”71 Galliard’s example demonstrates how a singer
of the time might embellish the melody with rhythmically inflected descending
note patterns. It reveals aspects of the improvisatory and irregular nature of the
strascino that could not necessarily have been deduced from Tosi’s explanation
alone. Nevertheless, the aural effect of the strascino, its subtle and varying
qualities, and the frequency of its use, remains obscure.
Other eighteenth-century texts document the importance of metrical rubato,
without making its features clear. In 1756 Leopold Mozart (1719–87) strongly
encouraged the accompanist not “to be beguiled by the postponing or anticipating
of the notes” that the soloist shapes “so adroitly and touchingly.” The accompani-
ment must not follow the soloist, otherwise the effect “would be demolished.”72
This style is “more easily shown than described.”73 In other words, it was best
transmitted aurally, a fact that helps to explain the lack of detail in written texts.
Later in the century, Türk made it clear that many types of rhythmic modifica-
tions were not only appropriate but also vital for a musically sophisticated per-
formance. Giving examples (Fig. 4.98), he explained that (a) metrical rubato
“is understood as a kind of shortening or lengthening of notes, or displacement
(dislocation) of these.” A note may be shortened causing the anticipation of others
as in (b), or it may be lengthened causing others to be delayed as in (c). Türk’s

Pier F. Tosi, Opinioni de’cantori antichi e moderni (Bologna, Italy: 1723); trans. John E. Galliard as
Observations on the Florid Song (London: 1742); modern ed. ed. Michael Pilkington (London: Stainer
& Bell, 1987), 70.
Tosi, Opinioni, 70–71.
Tosi, Opinioni, 84–85.
Leopold Mozart, Versuch einer gründlichen Violinschule (Augsburg, Germany: 1756); trans. Editha
Knocker as A Treatise on the Fundamental Principles of Violin Playing, 2nd ed. (Oxford: Oxford University
Press, 1951; reprint, 1988), 224.
Mozart, Treatise, 244.
238 off the record

Figure 4.98 Türk, metrical rubato alterations.

Figure 4.99 Türk, metrical rubato alterations.

examples (e) and (f) show how notes in a melody (d) can be anticipated and delayed
by embellishment (Fig. 4.99).74 In all of these, the tempo and rhythm of the bass
are to remain unaltered.
Indeed, Türk was critical of those musicians for whom it had become fashion-
able to disregard the underlying beat while extemporizing ornaments (presum-
ably extended ornamental figures). He remarks that the notes in such figures
may be “played a little too soon or a little too late for the sake of the affect” but
that even for the most extensive of these the pulse “must be maintained in the
strictest manner.”75 There is definite correlation between this and C. P. E. Bach’s
explanation of tempo rubato. He explains it as “the presence of more or fewer
notes than are contained in the normal division of the bar.” This compositional
distortion might affect a whole bar or part thereof, or even several bars in which
equal-valued notes must be given “exactly the same duration.” A desirable effect
is created when “one hand seems to play against the bar and the other strictly
with it” so that rarely do all parts coincide.76 We have seen that Taylor, at the
end of the nineteenth century, also referenced this style in a musical example.

Türk, School of Clavier Playing, 363–64.
Türk, School of Clavier Playing, 313.
Bach, Essay, 161.
Metrical R ubato and Other Forms of R hy thmic Alteration 239

While documenting the use of metrical rubato, the foregoing references could
not show the subtleties of rhythmic inflection that must have graced an artistic
The same is true of the oft-quoted reference to W. A. Mozart’s rubato style. He
explained to his father in 1777, “Everyone is amazed that I can always keep strict
time. What these people cannot grasp is that in tempo rubato in an Adagio, the left
hand should go on playing in strict time. With them the left hand always follows
suit.”77 Mozart appears to be criticizing keyboard players who could not properly
achieve independence of the hands or who, in any case, had become accustomed
to tempo modification made in both hands simultaneously. One can only imagine
the flexibility and beauty of Mozart’s right-hand alterations.
Apart from those already discussed, several other nineteenth-century refer-
ences document the ongoing tradition of metrical rubato. Adam states in 1804
that successful expression requires a slowing down or hastening (anticipation or
delay) of certain melody notes, a practice to be applied only at some places “where
the expression of a sad melody or the passion of an agitated melody requires” it.78
As noted in chapter 2, Adam, and later Pollini and Lichtenthal, provided verbal
and pictorial evidence that the portato sign signified not only a type of articula-
tion, but also a continuous displacement between the melody and accompani-
ment. But it is improbable that their notation captured the subtlety of the
intended effect.
In his The Singer’s Preceptor (1810), Domenico Corri described metrical rubato
(tempo rubato) in much the same way as the aforementioned eighteenth-century
sources: as the rhythmic alteration of notes while preserving the regularity of
time and the laws of harmony. Interestingly, he advised its use “in any species
of music where there is a leading or predominant melody.” As usual, the artful
rendering of such effects was left to the “the skill and prudence” of the player.79
Rhythmic alterations to melody notes in metrical rubato must sometimes
have been rather extreme. In 1837, Herz noted that where the character of the

Wolfgang A. Mozart, “Mozart to his Father (Oct. 23, 1777),” The Letters of Mozart and His Family,
ed. Emily Anderson 3 vols. (London: Macmillan, 1938), vol. 2, 497. Mozart’s words are “dass ich immer
accurat im tact bleybe. über das verwundern sie sich alle. Das tempo rubato in einem Adagio, dass die
lincke hand nichts darum weiss, können sie gar nicht begreifen. Bey ihnen giebt die lincke hand nach”;
in Mozart Briefe und Aufzeichnungen: Gesamtausgabe, ed. Ulrich Konrad, 8 vols. (Kassel: Bärenreiter,
2005), vol. 2, 83.
Adam, Méthode, 160: “Sans doute l’expression exige qu’on ralentisse ou qu’on presse certaines
notes de chant, mais ces retards ne doivent pas être continuels pendant tout un morceau, mais seule-
ment dans quelques endroits où l’expression d’un chant langoureux ou la passion d’un chant agité
exigent un retard ou un mouvement plus animé. Dans ce cas c’est le chant qu’il faut altérer, et la basse
doit marquer strictement la mesure.”
Domenico Corri, The Singer’s Preceptor, or Corri’s Treatise on Vocal Music (London, 1810), vol. 1, 6.
Reprinted in Richard Maunder, Domenico Corri’s Treatises on Singing, ed. Richard Maunder (New York:
Garland, 1993), vol. 3.
240 off the record

composition requires a “different rhythmic effect in each hand. . . . [T]he right

hand seems to lose its way in fantastic variations.” This requires both indepen-
dence of the hands and “a different soul in each of them.” According to Herz,
Dussek’s playing best exemplified this style in that he “produced a hazy and mel-
ancholy tint on certain sequences by letting the right hand sing in a vague and
nonchalant manner, whereas the left executed the arpeggiated chords rigorously
in time.” He laments the recent demise of this practice.80 As noted earlier, Dussek
abandoned attempts to notate such effects in his music and used the term espres-
sivo instead. In this light, the implications for the value of other musical illustra-
tions, for example by Galliard and Türk noted earlier, are serious.
In 1834, Baillot described the effect of seemingly extreme alterations, which
he notated as a type of syncopation expressing the trouble and agitation implied
by the expressions tempo rubato or disturbato, or temps dérobé or troublé. According
to him, these produce a great effect but will become fatiguing and insupportable
if used too often. When carried away by the expression, the performer loses
all sense of pulse and is “delivered by this means from the trouble that besets
him.” This carries with it the usual caution to remain aware of and to respect the
harmony and pulse.81 To demonstrate his meaning, Baillot made annotations to
an extract from the first movement of Giovanni Battista Viotti’s (1755–1824)
Violin Concerto No. 19 (Fig. 4.100a with alterations shown in Fig. 4.100b).82 By
modern standards, some of the alterations are fairly extreme and would no doubt
have sounded even more flexible than they appear on the page. One is reminded
of the variations in Cinti-Damoreaux’s examples (Fig. 4.15 ).
Other references highlight the liberties taken by some musicians. In 1836,
Potter described the soloist’s “peculiar division of the bar” as being of great beauty
in cantabile or slow passages. But he was critical of the extreme licenses in time
taken by some instrumentalists and singers, to which the orchestra “is too often

Herz, Méthode complète, 20: “quelquefois . . . le double caractère de l’accompagnement et de la
mélodie exige de chaque main un effet rhythmique différent. Ainsi, tandis que la droite semble s’égarer
en de folles variations, la gauche, appuyant à contre temps sur les basses, la suit à pas pesants et
pas notes syncopées. Ce cas, comme tous ceux où l’expression est complexe, exige non seulement des
mains parfaitement indépendantes l’une de l’autre, mais, si je puis le dire, une âme différent dans
chacune d’elles. C’est ainsi que Dussek répandait une teinte vaporeuse et mélancolique sur certaines
périodes en laissant chanter la main droite d’une manière vague et nonchalante, tandis que la gauche
exécutait des batteries rigoureusement en mesure. J’ignore pourquoi cette manière de phraser, tant
prônée naguère, est tombée maintenant dans l’oubli.”
Baillot, L’Art du violon, 136–67: “Il tend à exprimer le trouble et l’agitation et peu de compositeur
l’ont noté ou indiquér; le caractère du passage suffit dire, en faire usage que malgré lui, lorsqu’entrainé
par l’expression, elle l’oblige à perdre, en apparence, toute mesure et à se délivrer ainsi du trouble qui
l’obsède. Nous disons qu’il ne perd la mesure qu’en apparence, c’est-à-dire, qu’il doit conserver une
sorte d’aplomb qui le retienne dans les limites de l’harmonie du passage et qui le fasse rentrer à propos
dans la mesure exacte des temps.”
Baillot, L’Art du violon, 136–37.
Metrical R ubato and Other Forms of R hy thmic Alteration 241

Figure 4.100a Temps dérobé in Viotti Concerto No. 19, annotated by Baillot.

Figure 4.100b Temps dérobé in Viotti Concerto No. 19, showing alterations by Baillot.
Source: Extract of the facsimile edition Méthodes et Traités, violon, Baillot, France
1800–1860 (Réf. 5816), Anne Fuzeau Production—

obliged to submit.” Such licenses—presumably alterations made without regard

for the harmony, rhythm, or tempo—make it impossible to accompany some singers
and reduce the expression to “a caricature of the intention of the author.”83
Other soloists were more judicious and effected more tasteful alterations. In
1911, the conductor Bernhard Scholz (1835–1916) recounted his pleasure in
accompanying the baritone Julius Stockhausen (1826–1906) in 1859, while
directing the orchestra or playing the piano. At first Scholz tried to follow
Stockhausen’s every minute inflection. Eventually Stockhausen requested that
he “remain peacefully and strictly in time.” Stockhausen introduced “small
deviations here and there, for which he would later compensate.” This was an
ear-opening experience for Scholz. For the first time, the true character of the
tempo rubato—“freedom of phrasing on a steady rhythmic foundation”—became
completely clear to him.84 The fact that Scholz had to be instructed to play and
conduct in time suggests the variety of practices that existed side by side. Clearly,
musicians such as Stockhausen continued to use metrical rubato while others had
abandoned it.
The texts presented throughout this chapter document the survival of metrical
rubato and other forms of rhythmic alteration for several centuries until the turn

Cipriani H. Potter, “Companion to the Orchestra; or Hints on Instrumentation,” The Musical
World vol. 4, no. 41 (1886): 4.
Bernhard Scholz, Verklungene Weisen (1911); trans. Hudson, Stolen Time, 86.
242 off the record

of the twentieth century. It was certainly an intrinsic element in keyboard playing

and music making in general. However, few texts preserve the subtleties that
must have been the hallmark of trained artists. Their descriptive language and
brevity impede a fuller appreciation of its aural impression, the frequency of its
use or the range of situations in which it was considered appropriate.

Hidden Meanings
Central to an appreciation of metrical rubato are the possible hidden meanings in
its descriptive terminology. Written texts frequently state that in metrical rubato,
the accompaniment must remain “strictly” or “exactly” in time. In this respect,
several matters need to be considered. Did such terms imply absolute strictness?
Or did they simply imply that the pulse ought always to be recognizable despite
the vacillation of the melody? The latter would permit a degree of flexibility in the
placement of notes of the accompaniment, while maintaining a perception of
unvarying pulse. Another matter for consideration is the extent to which metro-
nomical strictness was considered truly artistic or indeed possible. In this regard,
Brown has observed that “a degree of deviation from absolutely mechanical adher-
ence to a constant beat is inevitable in a musically effective performance of any
reasonably extended piece, even if the performer’s primary intention is to adhere
strictly to the initial tempo.”85
The purpose of metrical rubato was to create an expressive harmonic or rhyth-
mic tension by playing a melody note sooner or later than the corresponding note
in the accompaniment. This can still be achieved when both parts fluctuate but do
not coincide. The commonly prescribed “strictly in time” accompaniment may,
therefore, have been simply a convenient and concise way of describing and ensur-
ing a displacement between the parts, but could also engender a certain degree of
During the first half of the nineteenth century, tempo modification appears to
have been increasingly employed as a standard expressive device. Czerny comments
that strict time keeping has been “almost entirely forgotten” and that arbitrary
changes of tempo are “now often employed even to caricature.”86 He also explains
that in almost every line of music there are notes or passages “where a small and
often almost imperceptible relaxation or acceleration” is needed to enhance the
expression and interest.87
Hummel, too, favored some flexibility in tempo. He explains that, in using the
metronome, many still mistakenly believe that “they are bound to follow its equal

Brown, Performing Practice, 375.
Czerny, Pianoforte School, vol. 3, 29.
Czerny, Pianoforte School, vol. 3, 31–32.
Metrical R ubato and Other Forms of R hy thmic Alteration 243

and undeviating motion throughout the whole piece, without allowing themselves
any latitude in the performance for the display of taste or feeling.”88
Being metronomically “in time” and musically “in time” certainly appear
to represent two quite different modes of performance. Toward the end of the
nineteenth century, Grimaldi encouraged “playing in time” as “most essential and
important.” For her, playing with taste and expression did not require too much
liberty with regard to rhythm and tempo. In the most extreme tempo rubato, “the
underlying general drawing must be always observed and felt.” But she warns the
student against going to the other extreme, of counting and playing the accompa-
niment rigidly, adding that “this is a good practice to make an accompanist, but
never an artist.”89
Within a framework of relative strictness then, a certain degree of tempo mod-
ification was considered desirable. Terminology such as strict and exact needs to
be interpreted in this context. The underlying tempo in metrical rubato may often,
therefore, have been more flexible than is implied by a face-value interpretation of
its descriptive terminology. Certain early-twentieth-century musicians who con-
cluded that metrical rubato was a purely theoretical concept did not take such
factors into account. For example, in 1928 John McEwen published results based
on evidence preserved on Duo-Art piano rolls. By measuring the distance between
perforations (and therefore note lengths and positions), he sought to discover,
among other things, whether when “playing an independent accompaniment to a
rubato melody or phrase,” the artists kept strict time. McEwen found that pia-
nists such as Pachmann, Busoni, and Teresa Carreño (1853–1917) did not do so;
their accompaniments wavered in tempo. Based on this, he concluded that theo-
ries of metrical rubato with strict time in the accompaniment were, as Philip
describes it, “inventions of theoreticians, rather than reflections of actual
practice.”90 One point can be seen to mar McEwen’s conclusions. He interpreted
the term strict far too literally. When one listens to the roll performances, the ear
recognizes elements of metrical rubato despite the fluctuations of tempo.
In 1937, Leroy Ninde Vernon made an empirical study based on piano rolls.
Focusing, like McEwen, on a limited number of rolls, he showed that when “a
clearly defined and continuous melody has an accompaniment of chords con-
trasted in rhythm and rather separate from the melody, the two are seldom played
together.”91 However, he concluded that because the accompaniments were not
steady in time, there were few if any examples of the Chopin style of rubato.
Again, the literal interpretation of descriptive terminology, without consideration
of historical context, seems to have clouded the issue.

Hummel, Art of Playing the Piano Forte, vol. 3, 65.
Grimaldi, The Art of Piano Playing and Teaching, 22–23.
Philip, Early Recordings, 46.
Hudson, Stolen Time, 333.
244 off the record

Historical texts can be easily misinterpreted when subjected to the boundaries

of our differing modern taste. In general, we have now become accustomed to a
style of performance that is very faithful to the score. Absolute precision, syn-
chrony of parts, and very subtle fluctuations of tempo are reigning characteris-
tics. Currently, the word strict means “very exact” or “literal.” However, for
nineteenth- and early-twentieth-century musicians, it apparently had a wider
meaning. In the context of tempo and rhythm, it incorporated a certain flexibility
that was still perceived as being strict. Certainly, in the foregoing recorded exam-
ples, metrical rubato alterations are clearly evident even when the accompani-
ment is not metronomically in time.
A good example of the hidden meaning in written texts is seen in the comparison
between Saint-Saëns’s description of the true tempo rubato (see chapter 2) and his
own playing. This style was properly and artistically effected when “the accompani-
ment remains undisturbed while the melody floats capriciously, rushes or retards,
sooner or later to find again the support of the accompaniment.” Such a manner is
hard to achieve and requires “a complete independence of the two hands.”92
The strong impression here is that in employing metrical rubato, Saint-Saëns’s
left hand would always have been exactly in time. His piano rolls show, however,
that in repertoire by Chopin and Beethoven, he wavers between a strict and a
more flexible tempo in the accompaniment. In spite of this, a sense of pulse is
always evident and the asynchrony caused by alterations to melodic material is
always clear. The fact that Saint-Saëns’s flexibility is not implied in his verbal
description shows how misleading the latter is for appreciating the true features
of his metrical rubato. Saint-Saëns very probably considered his playing to be
strict, but this was within a wider boundary than is currently acceptable.
Another example of the misleading impression given by written texts is found
in Eduard Hanslick’s 1879 review of Patti. According to him, Patti was always
rhythmically strict with regard to the overall length of a bar. However, the rhythms
within each bar were treated “with individual freedom—nothing is dragged, noth-
ing is rushed, and yet everything is animated right down to the softest vibrations
of tone.”93 In her 1906 recording of Bellini’s Casta Diva, Patti certainly treats the
rhythm within certain bars with individual freedom. However, by modern stan-
dards, her tempo, in this and other works, is not always strict. Often, she stretches
and contracts the tempo within a bar or phrase; she lengthens particular trilled
notes at final cadences in order it seems to show off her agility; and she makes
noticeable ritardandi in the final bars of songs and arias. None of these practices
render the composition unrecognizable, but she certainly does not sing strictly in
tempo. Here, as in abundant cases already mentioned in chapters 2 and 3, the
written description gives a different impression to the audible evidence.

Saint-Saëns, “Quelques mots,” 386–87.
Eduard Hanslick, “Adelina Patti (1879),” Music Criticisms 1846–99, trans. and ed. Henry
Pleasants, rev. ed. (London: Penguin Books, 1963) 179.
Metrical R ubato and Other Forms of R hy thmic Alteration 245

Changing Tastes
Metrical rubato continued to be a matter for discussion in the first half of the
twentieth century. In Some Reflections on Piano Playing (Paris, c. 1900), the pianist
Isidor Philipp (1863–1958) encourages the use of rubato for “expressive piano
playing.” According to him, it is often misunderstood and “does not mean playing
out of time.” He advises that “any ritenuto must be compensated by a correspond-
ing accelerando and also the opposite, the bass keeping exactly the time.”94
In an example from the second movement of Saint-Saëns’s Sonata No. 1
Op. 32 recorded in 1936, Philipp’s metrical rubato—extraordinary by modern
standards—is clearly audible. While his cellist Paul Bazelaire plays the walking
bass line between bars 6 and 10 exactly in time, Philipp makes displacements by
lengthening and shortening particular chords in the melody line (Fig. 4.101). It is
obvious that for him, ritenuto and accelerando signified lengthening and shorten-
ing or anticipation and delay. Without the audible evidence, this type of alteration
within the context of chamber music might never have been appreciated.
Other early-twentieth-century writers prescribed a very limited use of metri-
cal rubato. In 1909, Hofmann implied that it was more theoretical than practical
in nature and found that the usual explanation, in which the melody moves with
complete freedom while the accompanying hand keeps strictly in time, to be mis-
leading. This, he argues, is merely an assertion or an allegation and can be applied
only to “a very few isolated cases.” The pianist must decide whether alterations should
be extended “over both hands or only over one.” In the end, he sees little opportunity
for independence of the hands and still less advantage in its employment.95
Although Hofmann did not value the technique very highly, he was not entirely
opposed to it. Such explanations, however, must have accelerated the demise of a
practice that had long been in existence. Hofmann’s recordings preserve little use
of metrical rubato. His playing is markedly more synchronized than pianists noted
earlier, and his inflection of rhythms, while evident, is much more subtle and
bound to the notation. Listen for example to his 1912 recording of Chopin’s
Nocturne Op. 9 No. 2 (Audio Ex. 4.71 ).
An appraisal of Hofmann’s playing by the critic Harold Schonberg highlights
the difference between his style and that of earlier generations:

As a representative of the nineteenth-century school of piano playing,

Hofmann was well aware of the romantic tradition. He himself was
a bridge pianist, one who modified the romantic approach to the new
philosophies of the twentieth century. His rhythms were straightforward
whereas the rhythms of the Liszt and Leschetizky pupils tended to

Isidor Philipp, Some Reflections on Piano Playing (Paris: c. 1900), 11.
Hofmann, Piano Questions Answered, 100–102.
246 off the record

Figure 4.101 Saint-Saëns Sonata No. 1 Op. 32, second movement, bars 7 to 11,
Philipp and Bazelaire, electrical recording, 1935 (Audio Ex. 4.70 ).

be capricious. He played the notes as written, whereas the Lisztianers

and Leschetizkianers took a remarkably free view toward the printed
note. Indeed, Hofmann in early years was accused, often, of being a “cold”
pianist, just as Toscanini at the same time was being accused of being a
“cold” conductor. Of course neither was cold. Both merely discarded some
of the excess romanticism then in vogue.96

The recommendation for a literal interpretation of musical notation during the

early twentieth century is no clearer than in Grainger’s advice given in 1920 about
performing Grieg’s Norwegian Bridal Procession. On several occasions, he warns
that the dotted rhythms must be played exactly as written. Referring to bar 5, he
advises the player to

Harold C. Schonberg, [Sleeve Notes] The Complete Josef Hofmann: Volume One The Chopin
Concertos, Vai Audio/International Piano Archive 1002 (1992): unpaginated [2].
Metrical R ubato and Other Forms of R hy thmic Alteration 247

be scrupulous to preserve the exact rhythmic relationship between the

dotted sixteenth-notes and the thirty-second-notes. Too often this pas-
sage [Fig. 4.102 ] is played with the sound of triplets. This tendency
can be corrected by practicing the passage as indicated [Fig. 4.103 ].
Count four to every eighth-note, and be sure that the thirty-second-
notes are not played before “four” is counted. Inexperienced musicians
are apt to cut the duration of the dotted notes too short in cases such as
these. This error can also be corrected by practicing the passage with a
metronome ticking four times in each measure, and playing the thirty-
second notes like very quick grace notes [Fig. 4.104 ].97

Grainger also recommended this fairly rigid approach for other passages:

What was remarked regarding the rhythm of measure 5 applies with par-
ticular force to the continuous figure of dotted sixteenth-notes and
thirty-second-notes that are found throughout the following measures:
25–59, 62–72, 74–76, 78, 80–89. Take care not to let this degenerate
into the triplet rhythm [Fig. 4.105 ].

To safeguard against this tendency he advises the player to “think of each (not
to the preceding) dotted sixteenth-note, and practice the passage [in the same way
as for Figure 4.103 and Figure 4.104].”98 Further to this, for bars 49, 50, and 51,
Grainger states that the player must “play the thirty-second-notes in the left hand
well after the third note of the right hand triplet. In particular avoid the slovenly
performance of measure 51 [Fig. 4.106 ].”99 Grainger’s 1925 recording of the
work shows that he did observe his own advice, though in order to achieve rhyth-
mic incisiveness, he played it at a suitably moderate tempo of approximately
crotchet = 69MM. Grieg’s 1903 recording of the work shows that he was not so
strict and that, in any case, his faster tempo of approximately crotchet = 88MM,
often precluded the production of the sharp rhythms in his notation (Audio Ex.
4.72 ). In any case, as we have seen, Grieg did not always play the rhythms as
notated in his own music. Dotted figures, for example, are sometimes played in a
lilting, almost tripletized fashion.
Gieseking, too, demanded a much more literal interpretation of the musical
text. Discussing natural interpretation, he explains that pianists often believe
they “must alter the musical notation of a composition, especially as regards
rhythm.” This, he espouses, is very often done unconsciously, either through

Percy Grainger, “Grieg’s ‘Norwegian Bridal Procession’—A Master Lesson by Percy Grainger,”
The Etude vol. 38, no. 11 (1920): 42.
Grainger, “Grieg’s ‘Norwegian Bridal Procession,’” 742.
Grainger, “Grieg’s ‘Norwegian Bridal Procession,’” 742.
248 off the record

incorrect reading or superficiality in the playing. Indeed, some pianists think it

more expressive or interesting to play, for example, “a succession of sixteenth
notes unevenly and strongly rubato, although the composer has written them all
of equal value.”100
In criticizing the uneven rendition of passages of equal-value notes, Gieseking
may have been referring to the survival of a technique remarkably similar to the
practice of inequality discussed earlier, of which in his youth he must have
had direct experience. Elsewhere, he registered a strong dislike for the fact that
“triplets are rarely played correctly,” explaining that

they very often create a false impression when heard by trained ears. A
rhythmically rendered triplet is a thing unknown to many musicians. In
opposition to the intentions of the composer, the triplet is very often not
played precisely on the beat. Moreover, it is generally taken too fast and
finished too soon. In order, therefore, somewhat to balance the rhythm,
the player generally lingers a while before striking the note following the
triplet. I therefore go minutely into this matter and insist upon absolute
equality in the execution of the three notes. It is only by strictly following
this rule that such phrases can be rendered with the right effect.
Otherwise, they will always appear uneven and jerky. This may seem
pedantic to many, but it is a perfectly natural thing to a musical ear. The
correct rendering of triplets is a greater help to technique than is gener-
ally supposed.101

Like that of Hofmann and Grainger, Gieseking’s playing exhibits a stricter adher-
ence to the musical notation than that of pianists of an earlier generation. An
example highlighting the difference between his style and that of Freund, who
continued to make conspicuous rhythmic alterations as late as the 1950s, is seen
in the comparison of their recordings of Brahms’s Intermezzo Op. 117 No. 2. In
her 1953 recording of the work, Freund plays the adjacent semiquavers in the alto
and tenor parts between bars 27 and 29 unequally (Audio Ex. 4.73 ). In his circa
1939 recording of it, on the other hand, Gieseking plays the sequence literally as
notated (Audio Ex. 4.74 ).
Like Gieseking, Maurice Cauchie was highly critical, in 1929, of those instru-
mentalists and singers who employ rhythmic alteration and therefore substitute
“a work of their own composition for the one they imagine themselves to
be performing.” He suggests that in so doing, they believe they are being

Gieseking and Leimer, The Shortest Way to Pianistic Perfection, 43.
Gieseking and Leimer, The Shortest Way to Pianistic Perfection, 35.
Metrical R ubato and Other Forms of R hy thmic Alteration 249

more expressive. He warns that “the greatest care must be taken that the various
values (crotchets, quavers, &c.) last exactly the times that are intended.”102
As late as 1958, Leschetizky’s student Merrick was scathing of a “so-called def-
inition of rubato” in which the left hand is in time and the right hand free. He
notes, giving examples similar to that of Taylor, in which the number of notes in
the melody exceed the normal division of the bar, that this is “a phenomenon
which is often called for by the notation itself.” But he likens this style to that of
“an insensitive accompanist who cannot keep together with the soloist.”103 If
Merrick had at one time emulated aspects of Leschetizky’s style of metrical rubato,
he appears to have adopted (at least in theory) a more modern approach in which
any modification of rhythm or tempo occurred in both hands simultaneously.
In spite of this and other warnings, rhythmic alteration was still employed by
certain pianists at least until the mid-twentieth century. Freund is an excellent
case in point. In her 1953 recording of Brahms’s Sonata Op. 5, she makes altera-
tions to great effect, particularly in the first section of the second movement.
Here, the semiquavers in the left hand are played unequally, almost as triplets
(Audio Ex. 4.75 ). From bar 12, the repeated semiquavers marked with portato
articulation are played in a variety of dotted rhythms (Audio Ex. 4.76 ). And in
the section commencing from bar 68, each pair of semiquavers is subverted to
form a tripletized quaver/semiquaver (Audio Ex. 4.77 ). In the first movement
of the same Sonata, she plays the more expressive passages, for example bars 27
to 38, rather unequally (Audio Ex. 4.78 ). And in Brahms’s Intermezzo Op. 116
No. 2 recorded in 1953, Freund assimilates the pairs of quavers in the right hand
to match the left-hand triplets (Audio Ex. 4.79 ). Another fine example, and
there are many more, comes from De Lara’s 1951 recording of Brahms’s Intermezzo
Op. 117 No. 1. She tripletizes pairs of semiquavers forming the upbeat to most
bars (Audio Ex. 4.80 ).
Early recordings reveal, therefore, that around the turn of the twentieth cen-
tury, metrical rubato and other forms of rhythmic alteration remained indispens-
able expressive devices in piano playing. Many pianists—particularly but not
exclusively the oldest generation—frequently made rhythmic alterations of vary-
ing complexity for expressive purposes in a range of music. Yet few late-nine-
teenth-century written texts mention such practices. And those that do are
generally lacking in detail. In the case of metrical rubato, the texts rarely describe
more than its underlying principle. Thus the recordings reveal many features that
would have been impossible to deduce from written texts alone.
The close correspondence between the alterations preserved in early piano
recordings and those detailed by García and others provides strong evidence that
such alterations were, contrary to the views put forward in recent scholarship, not

Maurice Cauchie, “Respect for Rhythm,” The Musical Times vol. 70 (1929), 891.
Merrick, Practising the Piano (London: Rockliff, 1958), 73–74.
250 off the record

simply remnants of an earlier style. Clearly, there is a historical basis for what can
be heard in the recordings. In addition, certain similarities with earlier practices
suggest that many aspects of metrical rubato in late-nineteenth-century piano
playing had been in existence for several centuries. The recordings show clearly
that although some pianists employed metrical rubato more sparingly, Brahms,
Saint-Saëns, Reinecke, Leschetizky, Grieg, and other important pianists up until
the 1950s made noticeable use of it.
I am much bolder now than I used to be about altering the rhythm of notes in
a musical phrase. The more comfortable I get with it, the more I value it as means
of giving emphasis to particular notes and of varying the expression. I use rhyth-
mic alteration in a wide range of repertoire when the music will allow it or the
sentiment requires it. In a recent recording of Brahms’s Sonata for cello and piano
Op. 38, my cellist colleague and I were committed to using rhythmic alteration
and other devices (sometimes subtly, sometimes more blatantly) to bring out the
character and structure of the music.104 The exposition of the first movement can
be heard in Audio Example 4.81 . Having experimented with this, it becomes
almost inconceivable to play this music in the straightjacketed manner nowadays
frequently heard. Such a way sounds to me emotionally restricted: devoid of the
living, breathing expression that one can so easily imagine Brahms having
When time travel becomes possible, we will at last be able to experience first-
hand the rubato style of Mozart, Chopin, and other revered musicians. Until then,
an educated intuition will have to guide us. I feel certain though that few compos-
ers before the early twentieth century would have expected the notes in their
scores to be played exactly as notated.

Live unedited recording with cellist Daniel Yeadon, recorded November 17, 2010, in Verbrugghen
Hall, Sydney Conservatorium of Music.

Tempo Modification
[In today’s music making there is] a preference for strict tempo
and rhythm. . . . [A]ny tendencies of themes, or even single-bar
motives to speed up or slow down, become rhythmically two
dimensional. This kind of music making is boring, even if it is daz-
zling and virtuosic, because it is completely devoid of tension. . . .
Thus a microscopically refined feeling for tempo modification
must not be forgotten. . . . Tempo modifications protect the
rhythm from becoming too motoric and the melody from becom-
ing too lethargic. Expressed positively: Tempo modifications are
essential in giving rhythm and melody musical life.
—Walter Blume1

The extent to which, and the situations in which, tempo modification was
employed is yet another factor that distinguishes the style of piano playing around
the turn of the twentieth century from the present style. Nowadays, tempo
modification—also referred to as tempo rubato, or rubato—is used in fairly subtle
ways to enhance the ebb and flow of a musical phrase. And this generally takes
place whether indicated by the composer or not. Yet few specific rules govern its
application: musicians tend to develop an idiosyncratic manner or emulate the
practices of others. During the late twentieth century, rubato was commonly
defined as “some distortion of the strict mathematical tempo applied to one or
more notes, or entire phrases, without restoration; and also to time added as
pauses or breaks in the continuity of the tempo, to mark the separation of phrases
more conspicuously than merely by a silence of articulation within the tempo.”2
Such principles still hold true now, but the words used to describe them do not
convey the boundaries within which such distortions of the tempo are considered
appropriate or tasteful.
Throughout the ages, some degree of tempo modification must always have
graced musically sophisticated performances, albeit subject to the dictates

Pasternack, “Brahms in the Meiningen Tradition,” 3–4.
Robert Donington, “Rubato,” The New Grove Dictionary of Music and Musicians, ed. Stanley Sadie
(London: Macmillan, 1980), vol. 16, 292. More recently, Hudson has described tempo modification
as “a later type” of rubato involving “rhythmic flexibility of the entire musical substance,” in Richard
Hudson, “Rubato,” The New Grove Dictionary of Music and Musicians, 2nd ed., ed. Stanley Sadie (London:
Macmillan, 2001), vol. 21, 832.

252 off the record

of fashion. As a window into the past, piano recordings from the turn of the
twentieth century often reveal a style of tempo modification that is radically
different to the present. For the uninitiated, these recordings give the general
impression of exaggerated temporal waywardness. Yet many important pianists
of the era evidently considered such a style to be highly expressive. In contrast,
most pianists today adhere more faithfully to the dictates of notation: any
modification tends to stay within close proximity of the prevailing tempo.
Some appreciation, at least, of the extent to which the style of tempo modifica-
tion has changed over the past hundred years may be gained by examining a cross-
section of piano recordings of the same work. Table 5.1 presents certain
calculations3 based on the tempo modifications made by various pianists between
bars 1 and 9 of Chopin’s Nocturne Op. 27 No. 2 (Fig. 5.1 ).4 The elapsed time
between the downbeats of successive bars has been calculated using the music-
editing program Adobe Soundbooth CS4.5 Where smaller sections such as half
bars are involved, elapsed time has been provided accordingly. Because the effect
of increase or decrease (acceleration/deceleration) of tempo is perceived in rela-
tion to what has preceded, it is useful to know the rate of change of the length of
a bar, from one bar to the next, or one half bar to the next half bar, expressed as a
percentage of the first. This method provides a quantitative impression of the
degree to which individual pianists change the tempo during an accelerando or a
ritardando. The higher the value, the more radical is the perceived change. All
values have been rounded to one decimal place.
In Table 5.1, the upper numeral represents the times lapse in seconds for each
bar. The numerals marked in bold indicate the amount of increase (positive value)
or decrease (negative value) in length of the bar in relation to the one that
precedes. We can see that, in the majority of cases, no two bars are the same
length. More significantly, the variation in some bar lengths is quite a bit larger
for pianists such as Diémer, Leschetizky, La Forge, Pachmann, and Rosenthal,
and, to a lesser extent, Godowski and Powell than Livia Rév (b. 1916), Daniel
Barenboim (b. 1942), and Katherine Stott (b. 1958). Solomon Cutner (1902–88)6
plays consistently more in time, whereas Adam Harasiewicz (b. 1932) and particu-
larly Alex Weissenberg (b. 1929) show some localized large variation. Where
tempo has been modified, the earlier pianists generally make larger and therefore
more noticeable changes from bar to bar, whereas pianists more recently

Note that Diémer’s recording starts in bar 2. Pachmann’s 1915 recording commences at bar 26 of
the work and cannot therefore be included in this table, but his 1925 recording is included. Note that
permission was not granted for the use of Weissenberg’s (1969) recording on the companion website.
See discography for details.
Frédéric Chopin, “Nocturne Op. 27 No. 2,” Pianoforte-Werke, ed. Carl Mikuli (Leipzig, Germany:
Kistner, 1879), vol. 2, 30.
The downbeat is taken as the first bass note in each bar.
Solomon Cutner was known professionally as Solomon.
Te mpo Modif i cati on 253

Table 5.1 Chopin Nocturne Op. 27 No. 2, Bars 1 to 9, Duration in Seconds

Bar 1 Bar 2 Bar 3 Bar 4 Bar 5 Bar 6 Bar 7 Bar 8 Bar 9

Diémer c. 1903–4 3.7 3.6 3.8 3.2 3.5 3.0 4.2 3.4
(Audio Ex. 5.1 ) ‒0.1 ‒0.2 ‒0.6 0.3 ‒0.5 1.2 ‒0.8
Leschetizky 1906 5.8 5.3 4.7 5.5 5.1 5.2 4.5 6.4 5.6
(Audio Ex. 5.2 ) ‒0.5 ‒0.6 0.8 ‒0.4 0.1 ‒0.7 1.9 ‒0.8
La Forge 1912 5.3 4.6 4.1 4.3 4.4 4.1 3.7 6.0 5.3
(Audio Ex. 5.3 ) ‒0.7 ‒0.5 0.2 0.1 ‒0.3 ‒0.4 2.3 ‒0.7
Pachmann 1925 3.4 3.3 2.9 3.6 3.0 3.2 2.7 3.9 3.5
(Audio Ex. 5.4 ) ‒0.1 ‒0.4 0.7 ‒0.6 0.2 ‒0.5 1.2 ‒0.4
Godowski 1928 4.6 4.0 3.6 4.0 4.0 4.7 4.3 4.2 4.8
(Audio Ex. 5.5 ) ‒0.6 ‒0.4 0.4 0 0.7 ‒0.4 ‒0.1 0.6
Powell 1929 5.3 4.5 4.2 4.2 4.9 4.4 4.2 4.9 5.2
(Audio Ex. 5.6 ) ‒0.8 ‒0.3 0 0.7 ‒0.5 ‒0.2 0.7 0.3
Rosenthal 1936 4.0 4.1 3.4 3.7 2.9 3.6 2.8 4.6 4.3
(Audio Ex. 5.7 ) 0.1 ‒0.7 0.3 ‒0.8 0.5 0.2. 0.8. ‒0.3
Solomon 1942 5.4 4.8 4.4 4.5 4.7 4.6 4.4 4.6 4.8
(Audio Ex. 5.8 ) ‒0.6 ‒0.4 0.1 0.2 ‒0.1 ‒0.2 0.2 0.2
Harasiewicz 1963 4.3 3.7 3.3 3.7 3.7 3.7 3.6 4.0 5.0
(Audio Ex. 5.9 ) ‒0.6 ‒0.4 0.4 0 0 ‒0.1 0.4 1
Weissenberg 1969 5.9 5.7 4.9 5.9 5.2 5.2 5.6 6.5 6.7
‒0.2 ‒0.8 1 ‒0.7 0 0.4 0.9 0.2
Barenboim 1982 4.2 4.4 4.3 4.6 4.1 4.5 4.9 5.1 5.7
(Audio Ex. 5.10 ) 0.2 ‒0.1 0.3 ‒0.5 0.4 0.4 0.2 0.6
Rév 1988 5.0 4.5 4.0 4.2 4.2 4.3 4.2 4.9 4.9
(Audio Ex. 5.11 ) ‒0.5 ‒0.5 0.2 0 0.1 ‒0.1 0.7 0
Stott 1992 5.1 4.8 4.6 4.8 4.9 5.1 5.0 4.9 5.4
(Audio Ex. 5.12 ) ‒0.3 ‒0.2 0.2 0.1 0.2 ‒0.1 ‒0.1 0.5

make consistently far less variation. It seems that by the mid-twentieth century,
tempo modification was kept to a minimum, as represented in the figures for
Solomon and Harasiewicz. This accords with the general move toward a stricter
style that is described in written texts, discussed later. Weissenberg’s figures may
represent a remnant of earlier practices or perhaps a move toward a slightly more
flexible style. In this respect, the figures for Rév, Barenboim, and Stott show
more variation than those of Solomon and Harasiewicz. These conclusions do not
take into account the differences in overall tempo from one pianist to another.
254 off the record

Further to this, closer examination of the tempo modifications in bar 8 of the

Nocturne reveals even more interesting trends. Here, the majority of pianists
examined broaden the tempo in one of two ways in order, it seems, to heighten
the expression. Either they broaden the first few beats, including the high G
and the descending ornamental arpeggio that follows, or they broaden the
right-hand ascending scale in the second half of the bar that culminates in the
poignant G natural at the beginning of bar 9. Table 5.2 lists the increase in
the length of bar 8 expressed as a percentage of bar 7. The percentage change
figures shown have been calculated by dividing the increase or decrease in bar
lengths by the first bar length and multiplying by 100. All percentage change
figures are rounded to the nearest whole number. From this it is clear that
Diémer, Leschetizky, La Forge, Pachmann, and Rosenthal slowed down to a much
greater extent than Harasiewicz, Weissenberg, and Rév, and significantly more
than Barenboim and Stott.
To gain a clearer impression of the percentage changes that are typical of
individual pianists, it is useful to examine a cross-section of excerpts from the
Nocturne. As no two pianists necessarily make tempo modifications in the same
place, it is not the intention here to provide comparative figures. The percentage
changes for the most noticeable tempo modifications of each pianist are listed
in Tables 5.3 to 5.15.

Table 5.2 Chopin Nocturne Op. 27 No. 2, Bars 7 and 8, Percentage Change
Bar 7 Bar 8 Percentage Change

Diémer 3.0 4.2 40

Leschetizky 4.5 6.4 42
La Forge 3.7 6 62
Pachmann 2.7 3.9 44
Godowski 4.3 4.2 2
Powell 4.2 4.9 17
Rosenthal 2.8 4.6 64
Solomon 4.4 4.6 5
Harasiewicz 3.6 4.0 11
Weissenberg 5.6 6.5 16
Barenboim 4.9 5.1 4
Rév 4.2 4.9 17
Stott 5.0 4.9 2
Te mpo Modif i cati on 255

Table 5.3 Diémer, Percentage Change Figures

Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2

Bar 6 3.6
Bar 7 3.0 17
Bar 8 3.9
Bar 9 3.4 13
Bar 24 3.2
Bar 25 4.4 38
Bar 25 3.7
Bar 26 3.2 14
Bar 39 3.3
Bar 40 2.4 27
Bar 40 (first half) 1.0
Bar 40 (second half) 1.4 40
Bar 42 (first half) 1.4
Bar 42 (second half) 1.1 21
Bar 43 (first half) 1.3
Bar 43 (second half) 1.0 23
Bar 45 (first half) 1.4
Bar 45 (second half) 3.5 150
Bar 57 2.9
Bar 58 2.1 28
Bar 60 3.5
Bar 61 5.6 60

The figures given in these tables provide an overview of the extent to which the
pianists examined modify tempo in one particular work. This evidence reveals
that earlier pianists, such as Diémer, Leschetizky, La Forge, Rosenthal, and Powell,
employ fairly frequent tempo changes, sometimes extending a particular bar up
to twice as long as the preceding bar. The figures for Pachmann, too, accord with
this trend. In general, this is more radical than for later pianists such as
Weissenberg, Harasiewicz, Rév, Barenboim, and Stott, whose tempo modifica-
tions at their extreme cause a particular bar to be around one and a half times the
length of the preceding bar. Although there may be some modern pianists now
who make more radical modifications, it appears from this study that in recent
times, tempo has been varied within narrower boundaries than it was during the
first half of the twentieth century.
Table 5.4 Leschetizky, Percentage Change Figures
Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2
Bar 7 4.5
Bar 8 6.4 42
Bar 12 4.6
Bar 13 6.4 39
Bar 24 3.2
Bar 25 6.3 97
Bar 30 (first half) 2.4
Bar 30 (second half) 3.1 29
Bar 44 3.6
Bar 45 5.8 61
Bar 51 4.1
Bar 52 5.7 39
Bar 60 4.7
Bar 61 6.1 30

Table 5.5 La Forge, Percentage Change Figures

Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2
Bar 7 3.7
Bar 8 6 62
Bar 12 3.5
Bar 13 4.7 34
Bar 24 3.3
Bar 25 5.5 67
Bar 36 (first half) 1.7
Bar 36 (second half) 2.3 35
Bar 44 2.5
Bar 45 5.0 100
Bar 60 3.4
Bar 61 5.0 47
Bar 62 5.0
Bar 63 3.3 34

Table 5.6 Powell, Percentage Change Figures
Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2
Bar 12 4.0
Bar 13 5.9 48
Bar 18 3.7
Bar 19 4.7 27
Bar 24 5.6
Bar 25 7.3 30
Bar 44 2.3
Bar 45 4.7 104

Table 5.7 Rosenthal, Percentage Change Figures

Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2
Bar 2 4.1
Bar 3 3.4 17
Bar 4 (first half) 2.3
Bar 4 (second half) 1.4 39
Bar 4 3.7
Bar 5 2.9 22
Bar 6 (first half) 1.5
Bar 6 (second half) 2.1 40
Bar 24 3.7
Bar 25 5.0 35
Bar 43 2.9
Bar 44 3.1 7
Bar 45 7.9 155
Bar 60 3.3
Bar 61 4.8 45

Table 5.8 Pachmann 1915 Recording, Percentage Change Figures
Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2

Bar 31 4.2
Bar 32 5.5 31
Bar 29 (first half) 2.4
Bar 29 (second half) 1 58
Bar 34 (first half) 2.3
Bar 34 (second half) 1.8 22
Bar 35 (first half) 1.8
Bar 35 (second half) 2.0 11
Bar 28 4.0
Bar 29 3.4 15
Bar 44 4.5
Bar 45 7.1 58

Table 5.9 Pachmann 1925 Recording, Percentage Change Figures

Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2

Bar 7 2.7
Bar 8 4.1 52
Bar 19 3.0
Bar 20 3.9 30
Bar 32 5.0
Bar 33 4.0 20
Bar 44 2.9
Bar 45 5.0 72
Bar 56 3.7
Bar 57 4.7 27
Bar 59 3.0
Bar 60 5.0 67
Bar 64 3.1
Bar 65 3.7 19

Table 5.10 Solomon, Percentage Change Figures
Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2

Bar 9 4.8
Bar 10 3.8 21
Bar 16 4.0
Bar 17 3.8 5
Bar 18 3.2 16
Bar 19 4.2 31
Bar 20 4.1 2
Bar 23 3.6
Bar 24 4.1 14
Bar 25 5.4 32
Bar 27 4.8
Bar 28 5.4 13
Bar 33 4.9
Bar 34 3.7 25
Bar 36 4.1
Bar 37 3.4 17
Bar 38 3.1 9
Bar 39 4.0 29
Bar 41 2.9
Bar 42 2.4 17
Bar 44 2.41
Bar 45 3.59 49
Bar 71 (first half) 2.47
Bar 71 (second half) 2.94 19
Bar 72 (first half) 3.53 20

Table 5.11 Harasiewicz, Percentage Change Figures

Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2

Bar 12 3.6
Bar 13 4.4 22
Bar 24 3.6
Bar 25 4.3 19
Bar 44 3.0
Bar 45 4.5 50
Bar 57 3.8
Bar 58 2.9 24

Table 5.12 Weissenberg, Percentage Change Figures
Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2

Bar 12 4.7
Bar 13 5.2 11
Bar 18 3.9
Bar 19 6.0 54
Bar 24 5.5
Bar 25 7.3 33
Bar 44 3.0
Bar 45 5.2 73

Table 5.13 Barenboim, Percentage Change Figures

Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2

Bar 7 4.9
Bar 8 5.1 4
Bar 15 4.5
Bar 16 6.0 33
Bar 24 4.1
Bar 25 5.3 29
Bar 29 4.4
Bar 30 5.4 23
Bar 39 4.4
Bar 40 3.8 14
Bar 44 3.3
Bar 45 5.2 58
Bar 60 5.0
Bar 61 6.8 36
Bar 69 5.4
Bar 70 6.2 15

Te mpo Modif i cati on 261

Table 5.14 Rév, Percentage Change Figures

Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2

Bar 24 4.3
Bar 25 5.3 23
Bar 29 3.9
Bar 30 4.0 3
Bar 30 (first half) 2.0
Bar 30 (second half) 2.0 0
Bar 44 3.6
Bar 45 5.1 42
Bar 67 4.7
Bar 68 5.0 6

Table 5.15 Stott, Percentage Change Figures

Chopin Nocturne Seconds Percentage Change
Op. 27 No. 2

Bar 24 5.0
Bar 25 6.3 26
Bar 42 (first half) 2.2
Bar 42 (second half) 1.7 23
Bar 44 3.4
Bar 45 5.2 53
Bar 60 4.7
Bar 61 5.3 13

But even the foregoing calculations do not convey certain features of tempo
modification that contribute to the improvisatory and rhetorical sound of some
early piano recordings. Such features include the sudden short accelerations that
create the effect of erratic forward surges such as can be heard between bars 1 and
9 of Diémer’s circa 1903–4 recording of Chopin’s Nocturne Op. 27 No. 2 (Audio
Ex. 5.1 ), or between bars 9 and 12 of the second movement of Beethoven’s
Sonata Op. 31 No. 1 on Saint-Saëns’s 1905 piano roll (Audio Ex. 5.13 ). These
features also include frequent agogic lengthenings7 that create erratic rhythmic

The accentuation of single notes made by variation of duration as opposed to variation of dynamic.
262 off the record

effects exemplified in bars 16 to 26 of the same performance (Audio Ex. 5.14 ).

In the sequence from bars 16 to 24 of Chopin’s Nocturne Op. 15 No. 2, Saint-Saëns’s
1905 piano roll reveals frequent accelerandos and ritardandos that do not sound
proportioned by today’s standards. During these bars, he also anticipates the
entry of a new phrase in a manner that sounds abrupt (Audio Ex. 5.15 ). And in
the Doppio movimento section of the work from bars 25 to 48, he makes a very
noticeable and erratic-sounding accelerando not indicated by Chopin. In addition,
the rate at which he slows down during the molto rallentando at bar 47 is more
extreme than might have been expected (Audio Ex. 5.16 ). In his 1916 record-
ing of Chopin’s Nocturne Op. 27 No. 2, Pachmann makes similar forward surges
in the left hand in bar 29 (Audio Ex. 5.17 ). Rosenthal produces a similar effect
in the left hand at bar 5 of his 1936 recording of the work (Audio Ex. 5.18 ). And
between bars 10 and 14 (Audio Ex. 5.19 ) and in bar 38 (Audio Ex. 5.20 ), he
accelerates particular right-hand figures, projecting them in an impatient manner.
He also rushes certain notes in the left hand. La Forge’s 1912 recording of the work
reveals abrupt surges in bar 11 and the second half of bar 15 (Audio Ex. 5.21 ).
And Powell makes them in bar 20 (Audio Ex. 5.22 ), bar 32 (Audio Ex. 5.23 ),
and bar 57 (Audio Ex. 5.24 ) on his 1929 piano roll. Such seemingly erratic,
whimsical, and exaggerated modifications are seldom heard in piano playing nowa-
days. Philip has presented further interesting examples of tempo modification in a
range of recordings from solo piano to orchestral that reveal a number of trends:

The most obvious is that a greater range of tempo within movements was
generally used in the 1920s and 1930s than in modern performances.
But the trend over the last 60 years has not been simply a narrowing of
the accepted tempo range. In pre-war performances, slowing down at
points of low tension and speeding up at points of high tension were
both used frequently, and with emphasis. Modern performers still some-
times slow down at lyrical passages, particularly in works of the Romantic
period, but accelerations at energetic passages are generally very
restrained. The degree of acceleration heard in many pre-war recordings
would be considered uncontrolled in modern performance. . . . Over the
succeeding decades there has been a gradual change in attitude to tempo,
and to flexibility of tempo, and this has been part of a more general
change in the rhetoric of musical rhythm. . . . [M]odern taste insists on
careful control, particularly of acceleration. This goes with a requirement
that every detail should be considered and clearly placed. By comparison,
early-twentieth-century performance was more volatile. Theoretical
flexibility was applied not just to overall tempo, but also to the shaping
of phrases and the relationship between individual notes.8

Philip, Early Recordings, 35–36.
Te mpo Modif i cati on 263

Although Philip does not comment about it, a greater range of tempo within
movements was evidently employed earlier than the 1920s. Certainly the earliest
piano recordings provide irrefutable evidence that a style of tempo modification—
one that is no longer considered tasteful—was intrinsic to piano playing in the
second half of the nineteenth century. As with dislocation, unnotated arpeggia-
tion, and rhythmic alteration, the comparison between early piano recordings and
contemporaneous verbal advice reveals several anomalies regarding the application
of tempo modification. Some of these are highlighted in the section that follows.

Early Recordings and Written Texts

The practices of Brahms provide a suitable point of departure, as written refer-
ences to his style of tempo modification may be compared with his 1889 cylinder
recording of his Hungarian Dance No. 1. Fanny Davies, who heard Brahms on many
occasions, recounted that his “manner of interpretation was free, very elastic
and expansive.” And yet it always retained its balance: “one felt the fundamental
rhythms underlying the surface rhythms.” Davies notes that Brahms used tempo
modification to emphasize phrasing particularly in lyrical passages. Here, “a strictly
metronomic Brahms is as unthinkable as a fussy or hurried Brahms in passages
which must be presented with adamantine rhythm.”9
Davies also described Brahms’s practice of lengthening individual notes as well
as making larger-scale modifications, sacrificing an unvarying tempo to create a
beautiful effect:

The sign “< >,” as used by Brahms, often occurs when he wishes to express
great sincerity and warmth, applied not only to tone but to rhythm also. He
would linger not on one note alone, but on a whole idea, as if unable to tear
himself away from its beauty. He would prefer to lengthen a bar or a phrase
rather than spoil it by making up the time into a metronomic bar.10

Further to this, Davies explains that during the third movement of his Piano
Trio Op. 101, “Brahms would lengthen infinitesimally a whole bar, or even a
whole phrase, rather than spoil its quietude by making it up into a strictly metro-
nomic bar.” Interestingly, she describes this apparently infinitesimal lengthening
as “expansive elasticity,” something distinct from “a real rubato” (presumably
metrical rubato). Either practice might be applied according to the musical idea.

Fanny Davies, “Some Personal Recollections of Brahms as Pianist and Interpreter,” Cobbett’s
Cyclopedic Survey of Chamber Music, compiled and ed. Walter W. Cobbett, with supplementary material
ed. Colin Mason, 2nd ed. (London: Oxford University Press 1963), vol. 1, 182.
Davies, “Some Personal Recollections,” 182.
264 off the record

For Davies, such elasticity was a chief characteristic of Brahms’s manner of

Davies’s detailed description of metronome speeds in the fourth movement
of the same work (summarized in Table 5.16), that she verified with Joachim, is
particularly valuable and shows the extent to which Brahms modified tempo:

The last movement about dotted crotchet = 120; at the much discussed
meno allegro about dotted crotchet = 88. Then the tempo broadened
gradually, until at the violin solo with semiquaver accompaniment it had
become about dotted crotchet = 72. The violin solo I marked “very much
brought out,” the cello the same—a real solo. Then came a very fine shad-
ing to pp, a “taking off,” but hardly to be called a ritardando. Tempo 1,
then, of course (dotted crotchet = 120). The song in C major and the first
four bars of the poco stringendo started at about crotchet = 76—at first
quietly, then going on in musical phrases and becoming rather “wild,” as
marked in my copy—through 100 and 108 to 120, as in the beginning.

Davies concludes that although much more could have been detailed, what she
described was “thoroughly typical of Brahms’s conception and performance of his
works.”12 Her figures show a wide variation of tempo that is not evident from
Brahms’s indications alone. For instance, the term poco stringendo results in a
variation from 76MM to 120MM. By today’s standards, this is fairly extreme.

Table 5.16 Brahms Piano Trio Op.101, Fourth Movement, Tempo

Annotations by Davies
Bar 1 Dotted crotchet = 120
Bar 49 (meno allegro) Dotted crotchet = 88
Bar 49–66 Broadening gradually
Bar 46 (violin solo with semiquaver Dotted crotchet = 72
Bars 82–83 Slight ritardando
Bar 84 (Tempo 1) Dotted crotchet = 120
Bar 190 (the song in C Major) + first 4 Starting at crotchet = 76, then becoming rather
bars of poco stringendo bars 213–16 wild through crotchet = 100, 108, and 120

Davies, “Some Personal Recollections,” 184.
Davies, “Some Personal Recollections,” 184.
Te mpo Modif i cati on 265

Florence May gave a similar impression of Brahms’s style of tempo modi-

fication. She recollected that his way of interpreting the music of J. S. Bach
“was always unconventional and quite unfettered by traditional theory.” He did
not share the popular view that Bach’s music required “a simply flowing style.”
In the performance of Bach’s keyboard suites, Brahms favored “variety of tone
and touch, as well as a certain elasticity of tempo.”13
Further corroboration of Brahms’s preference for fluidity of tempo is found in
his response to the English baritone George Henschel (1850–1934), who inquired
whether the metronome markings in the Requiem Op. 45 were to be strictly fol-
lowed. Brahms responded that elasticity of tempo was “not a new invention” and
that the term con discrezione “should be added to this as to many other things.”14
Tempo flexibility appears to have been an indispensable aspect of “Brahmsian”
style. Yet the boundaries within which this flexibility took place remain relatively
unclear. Apart from Davies’s metronome indications in one example, the written
references do not convey how much or how little modification of tempo is appro-
priate. This is also the case with another important source closely connected with
Brahms himself. The textual annotations of the conductor of the Meiningen
Orchestra, Fritz Steinbach, provide invaluable information about the places in
his orchestral works where Brahms apparently welcomed tempo modification.
Steinbach’s pupil, Walter Blume, published a description of these annotations.
For example, during the fourth movement of the Symphony No. 3 Op. 90,
Steinbach instructed the orchestra to plays as follows:

At H the strings should play near the bridge until the f entry on m. 141.
The tempo must above all remain unchanged from m. 141 and on.
Although a ff is reached before K, all instruments should make another
crescendo toward the climax one measure before K. . . . The tempo should
relax at the beginning of the triplet episode in m. 252. Each of the six-
teenth-note figures in the strings at O should be played with a slight
tenuto on the first sixteenth-note.15

Such descriptions abound in the Meiningen document; however, they do not tell
us the extent of the variation in tempo caused by agogic lengthenings and/or
larger-scale modifications. Other references purporting to preserve Brahmsian
practices suffer from a similar lack of detail. Reminiscing about his performance
with Joachim and Rüdel in Berlin in 1902, Tovey recounts that for the first move-
ment of Brahms’s Sonata Op. 108, he learned from Joachim “that at the first forte
Brahms made a decided animato which he might as well have marked in the score;

May, The Life of Johannes Brahms, 16.
Pascall, “Playing Brahms,” 16.
Pasternack, “Brahms in the Meiningen Tradition,” 92.
266 off the record

this, of course, implies that the tempo of the outset must be broad, though, of
course, flowing.”16 And for the second movement, Scherzo, from Brahms’s Trio
for Piano, Violin, and Horn Op. 40, Tovey claims to have discovered from Joachim
that the long-standing custom of playing bars 120 to 152—“the quiet B major
passage where the violin and horn pull the theme out by holding every third note
for an extra bar while the pianoforte interpolates pianissimo arpeggios”—in a
slower tempo, was a mistake. Importantly, however, he notes that such modifica-
tions of tempo became allied to Brahms’s later style of composition in which
the action became more rapid and the texture more concentrated.17 Although this
and other written evidence strongly support the use of tempo modification at
appropriate moments in Brahms’s music, they give us only a vague impression of
the features of its employment.
A remnant (at least) of the long-standing tradition mentioned earlier by Tovey
is preserved in a 1933 recording of the Scherzo from the Op. 40 Trio by Adolphe
Busch, Aubrey Brain, and Rudolf Serkin. In this, the exposition commences at
approximately dotted minim = 112MM (the second subject from bar 49 is very
slightly slower). There is a ritardando from bar 106 in the B major passage, to a
new tempo at bar 109 of approximately 104MM. A further ritardando during the
solo piano passage between bars 113 and 120 reduces the tempo to 100MM for
the passage in question (Fig. 5.2 ; Audio Ex. 5.25 ). From bar 199, a decided
animato eventually reestablishes the original tempo of 112MM. Similar modifica-
tions are made during the repeat of the Scherzo. Many now would consider this
degree of tempo change overdone and in dubious taste.
Brahms’s use of tempo modification as an expressive device in the extract from
his Hungarian Dance No. 1 has been discussed by Crutchfield in “Brahms, by
Those Who Knew Him” (1986), and by Jonathan Berger and Charles Nichols in
“Brahms at the Piano: An Analysis of Data from the Brahms Cylinder” (1994). The
scientific analysis by Berger and Nichols shows that in order to delineate the
structure of the composition, Brahms deliberately broadened the tempo during
certain bars. By graphing the time lapse between the successive first beats for
most bars between bars 13 and 71 (Fig. 5.3 ),18 they explain that it

shows much longer durations for measures 30, 56, 64 and 68, and much
shorter durations for measures 31 and 55. The elongation of measure 30
occurs at the end of a six-bar phrase. The previous measure, 29, is also
lengthened, suggesting a ritardando at the end of a phrase. Measures 56,
64 and 68 are all at the end of four-bar phrases, and are also probably due

Donald F. Tovey, “Brahms’s Chamber Music,” Essays and Lectures on Music, ed. Hubert J. Foss
(London: Oxford University Press, 1949), 264.
Tovey, “Brahms’s Chamber Music,” 249.
Berger and Nichols, “Brahms at the Piano,” 29.
Te mpo Modif i cati on 267

Figure 5.4 Brahms Hungarian Dance No. 1, bars 29 and 30.

Figure 5.5 Brahms Hungarian Dance No. 1, bars 66 to 68.

to rubato or ritardando. The shorter durations, measures 31 and 55,

immediately follow or precede a lengthened measure, suggesting a musi-
cal compensation for time gained or lost.19

The graph in Figure 5.3 clearly shows the elongation of bars such as 30, 56, and 68
that are structurally important. These either form the end of a phrase or contain
the characteristically emphatic Hungarian dance rhythm (Figs. 5.4 and 5.5).20

Berger and Nichols, “Brahms at the Piano,” 29.
Johannes Brahms, Hungarian Dance No. 1, Hungarian Dances, Book 1 (New York: Schirmer,
1893), 3–4.
268 off the record

Berger and Nichols also evaluate Brahms’s tempo modifications by measuring

and plotting the time lapse between the second beats of various bars (Fig. 5.6 ).21
They conclude the following:

Two outstanding second-beat IOIs (inter-onset intervals), in measures

29 and 71, occur during arpeggiation in the penultimate measure of the
phrase. The other longer second beat IOI, in measure 60, is at the end of
a 12-bar phrase. Both cases are easily interpreted as musically motivated

The Berger and Nichols graphs clearly show that some bars are much longer than
others. They also portray a continual variation of tempo.
In his less scientific but no less valuable analysis, Crutchfield describes the
extent of tempo variation during particular sections of Brahms’s performance.
According to him, Brahms “starts off at a tempo of about half-note = 83, but soon
settles to a basic pulse of approximately 78.” Particularly interesting is Brahms’s
way of playing the syncopations and sixteenth-note runs in the B section that are
so typical of nineteenth-century “Gypsy” style (Fig. 5.7 ).23 Syncopated chords
are played very emphatically, or rinforzando, with agogic accent (lengthening) and
strong accentuation. On the fourth appearance of the syncopated figure, Brahms
also gives special emphasis to the first beat. Runs are played at a notably increased
tempo—typically in metronome speeds in the high 80s—creating “a dashing
effect.” For Crutchfield, the best moment is the final cadence, “which is tossed off
with a fiery snap, faster yet than the tempo of the runs.”24 Though Crutchfield
doesn’t mention it, I detect a noticeable increase of tempo in bars 69 and 70,
followed by a broadening from bar 71 to the end.
Brahms used tempo modification to enhance the effect and character of his
Hungarian Dance No. 1. But the boundaries within which he makes tempo varia-
tions, and the frequency with which they occur, are wider than would normally be
acceptable today. His practice contradicts the modern concept of an appropriate
style for his music. Clearly, however, Brahms intended the lengthening of single
notes as well as the broadening of particular bars to increase poignancy and to
delineate various phrase shapes and structures, while the speeding up of certain
phrases certainly adds to their excitement. In conjunction with written texts,
the Brahms cylinder confirms that he did practice what he preached, but in a style
that could not be fully encapsulated by the written word. What can safely

Berger and Nichols, “Brahms at the Piano,” 29.
Berger and Nichols, “Brahms at the Piano,” 29–30.
Brahms, Hungarian Dance No. 1, Hungarian Dances, 3.
Crutchfield, “Brahms, by Those Who Knew Him,” 14.
Te mpo Modif i cati on 269

be extrapolated from this evidence and applied to a genre such as a sonata is

food for further thought and experimentation.
The playing of Reinecke, who was nine years older than Brahms, exhibits sig-
nificant use of tempo modification. In his 1905 piano rolls of Schumann’s Warum?
Op. 12 No. 3 and his own arrangement of the Larghetto from Mozart’s Piano
Concerto K 537, frequent tempo changes enhance the expression of phrases and
help delineate their boundaries. Metronome markings and comments for both
works are provided in Tables 5.17 and 5.18. The most significant aspect of his
performance of Schumann’s Warum? is the wide variation of tempo from 54MM
to 80MM, within a short single-movement work. The metronome readings for
Reinecke’s performance of Mozart’s Larghetto show frequent variation within
wide limits ranging from crotchet = 52MM to 96MM. On first listening, some
of his modifications have the aural effect of lurching and erratic surging. But
clearly, these were intended to give particular character to individual phrases and
phrase parts.
Comparison between Reinecke’s style of tempo modification and his verbal
advice brings to the fore several inconsistencies. In his discussion of Beethoven’s
piano sonatas, he frequently comments about the use of tempo modification.
Regarding the first movement—in variation form—from Beethoven’s Sonata
Op. 26, for example, Reinecke is enthusiastic about slight modifications of tempo
here and there, including the “not quite immediate succession of the fourth to the
fifth variation.” But he warns especially “against ‘perceptible’ changes of tempo
and ‘perceptible’ pauses.”25

Table 5.17 Schumann Warum? Op. 12 No. 3, Sections A and B without

Repeat, Reinecke, Piano Roll Recording, 1905 (Audio Ex. 5.26 )
Bars 1–4 60MM
Bars 4–8 Ritardando to between 54 and 56MM for climax
Bars 8–9 Accelerando to 69MM
Bars 9–10 Somewhat broadened giving poignancy to the falling melodic figures
Bar 12 Ritardando as marked by Schumann
Bars 12–17 60MM
Bar 17 66MM produces effect of reanimation
Bars 17–26 Accelerando to 80MM significantly increasing the momentum
Bars 19 and 23 Agogic lengthenings emphasizing the rising bass figures
Bars 23–42 Ritardando to 66MM and eventually 60MM

Reinecke, The Beethoven Piano Sonatas, 47.
270 off the record

Table 5.18 Mozart Larghetto Arr. by Reinecke, Bars 1 to 60, Reinecke,

Piano Roll Recording, 1905 (Audio Ex. 5.27 )
Bars 1–8 Approximately 66MM with slight broadening at end of each
four-bar phrase—bars 4 and 8, respectively
Bar 9 marked animato Sudden increase to approx. 72MM
Bar 14 Accelerando to approx. 96MM for rising semiquaver
passages, tempo then settles to approx. 84MM
Bar 16 Accelerando during rising figure and expressive broadening
for falling figure that follows
Bar 17 Accelerando during embellished rising figure
Bar 19 marked Tempo reduced to 66MM
un poco slentando
Bar 32 Tempo reduced to 60MM to enhance the effect of the
interrupted cadence
Bar 36 marked as Starts at approx. 60MM with slight accelerando for the rising
con espressione arpeggio at the end of bar 38 and again at the end
of bar 43
Bar 44 Accelerando enhancing the rich harmonies of the climax
Bar 46 72MM
Bar 48 66MM with ritardando to end of bar 49
Bar 50 60MM for the ethereal section
Bar 52 Broadening to between 52 and 54MM for poignant material
Bar 53 Sudden increase of tempo to 72MM followed by increase to
80MM for restless triplet section
Bar 60 Return to 72MM followed by sudden accelerando through
the ascending scale leading to the cadenza at bar 62
Recapitulation Similar tempo modifications

Later, Reinecke remarks that strictly metronomic performance for the entirety
of a sonata movement “is as inconceivable as unlovely.” He notes the great dif-
ference between the imperceptible changes of tempo—both speeding up and
slowing down—that all sensitive artists make appropriately and as a matter of
course, and the “obtrusive changes” that might lead to a “feeling of jerkiness or
forcing.”26 The strong impression here is that any tempo change should be very

Reinecke, The Beethoven Piano Sonatas, 74.
Te mpo Modif i cati on 271

Reinecke explains his avoidance of indications such as ritardando or stringendo:

in his view, these “always lead to exaggeration.” Sensitive performers, he acknowl-
edges, “will introduce those small modifications which might be desirable.” But
he advises less talented or inexperienced players to avoid such changes in case
they sound overdone.27 For Reinecke, subtle tempo changes were occasionally
appropriate when the music specifically warranted it. Referring to a passage in
the second movement of Beethoven’s Sonata Op. 81a Les Adieux—Andante
Espressivo—he singles out bars 17 and 33 (Fig. 5.8 ).28 These contain a prolif-
eration of notes in the second half of each bar for which, in his opinion, “a certain
ritardando” cannot be avoided if the requisite feeling is to be expressed. Reinecke
believed that Beethoven did not give an indication of this, because such a direc-
tion “would drive the majority of players to exaggeration.”29 Reinecke also
prescribed subtle tempo changes to delineate one musical idea from another. For
example, he advises that bar 4 of the first movement of Beethoven’s Sonata
Op. 101 (Fig. 5.9 )30 be played with a “diminuendo and an imperceptible slacken-
ing of the pace” so that the following bar satisfactorily stands out from it.31
Reinecke also recommended tempo modification to help sustain musical inter-
est. As an example, he cites bars 40 to 58 in the development section of the first
movement from Beethoven’s Sonata Op. 110. Here, the two bars taken from the
principal subject are repeated nine times in succession. To avoid monotony,
Reinecke advises “a discreet accelerating of the tempo during the first 14 bars,
while an equally discreet ritardando has then, with the entry of the principal
Subject, to lead again into the original tempo.”32
Reinecke was scathing of exaggerated tempo fluctuation in classical sympho-
nies because he felt these caused the work to become unrecognizably distorted.
He berated conductors for readily taking such licenses, as well as critics who had
“become indifferent to such inartistic runnings after effect” or who would “shrink
from censuring them.”33 It is likely that Reinecke was referring to the type of
interpretative conducting promoted by Wagner and emulated by disciples such
as Bülow.
Taken together, Reinecke’s written references leave no doubt that he consid-
ered tempo modification as an indispensable performing practice. However, a
face-value interpretation of his verbal advice gives the impression that he pre-
ferred performances that did not stray too far from the initial tempo and that any
changes were to be unobtrusive. But Reinecke’s piano rolls preserve frequent and,

Reinecke, The Beethoven Piano Sonatas, 65.
Beethoven, “Sonata Op. 81,” ed. Hallé, 422.
Reinecke, The Beethoven Piano Sonatas, 92.
Beethoven, “Sonata Op. 101,” ed. Hallé, 447.
Reinecke, The Beethoven Piano Sonatas, 96.
Reinecke, The Beethoven Piano Sonatas, 130.
Reinecke, The Beethoven Piano Sonatas, 67.
272 off the record

by today’s standards, very perceptible modifications of tempo that do not

appear to accord with his written advice. For us, his modifications often sound
exaggerated, abrupt, and even jerky. Clearly, his descriptions do not convey the
features and frequency of such tempo nuances. Therefore, a performance style
based on his written advice alone would lead to a result quite at variance with
his own style.
Leschetizky, we have already seen, considered tempo modification an indis-
pensable device. His 1906 piano roll of Mozart’s Fantasia K 475 shows prolific use
of tempo modification, which certainly enhances the improvisatory character of
the work (Table 5.19). While every subtle nuance cannot be described, the
table gives some indication of the most significant modifications for particular
sections. Although percentage change figures for some of Leschetizky’s tempo
modifications in Chopin’s Nocturne Op. 27 No. 2 were given earlier, in Table 5.4,
the following analysis provides further clarity by referring to metronome mark-
ings (Table 5.20). Leschetizky’s use of tempo modification enhances the expres-
sion of individual phrases and sections and helps distinguish their boundaries.
While broadening is a frequent occurrence, accelerations are less frequent. When
accelerations do take place, however, the intention is clearly expressive.
Comparing these modifications with Leschetizky’s own edition of Chopin’s
Nocturne Op. 27 No. 2 reveals a few interesting points. For example, his broaden-
ing during bar 8 corresponds with his notated cantando (Fig. 5.10).34 Evidently,
such terminology implied the use of tempo modification and other expressive
devices during Leschetizky’s era, though it would not now generally be interpreted
in this way. Indeed, the a tempo indication in bar 9 implies that the tempo ought
to have changed. Notable, too, is the fact that Leschetizky’s tempo modifications
between bars 22 and 25 do not completely accord with his notation. In bar 22 he
notates a poco accelerando that he does not make, though the following markings
of calando, poco a poco rall, and molto riten are steadfastly observed. At other places,
such as between bars 37 and 45, where a very noticeable increase in tempo occurs,
Leschetizky gives no indication other than poco a poco crescendo (Fig. 5.11 ).35
But, as mentioned by many writers during the nineteenth century, such dynamic
indications implied a corresponding tempo modification.
In 1957, Merrick noted that tempo modification is nowadays “more severely
frowned on than in my childhood,” underlining a dramatic change in attitude
toward its use during the first half of the twentieth century. Furthermore he
explained that

Leschetizky was sometimes at pains to advocate subtle vacillations, per-

haps in a graded series, that enabled one to achieve desired changes

Chopin, “Nocturne Op. 27 No. 2,” ed. Leschetizky, 18.
Chopin, “Nocturne Op. 27 No. 2,” ed. Leschetizky, 21.
Te mpo Modif i cati on 273

unperceived. I can for once masquerade as a moderate man if I submit

that about the period of 1900 there were too many tempo changes and in
the 1950s there are perhaps too few. Certainly if there is anyone who
advocates an inflexible metronomic constancy of pace regardless of other
considerations, he may be regarded as an extremist.36

Table 5.19 Mozart Fantasia K 475, Bars 1 to 124, Leschetizky, Piano Roll
Recording, 1906 (Audio Ex. 5.28 )
Bars 1–5 Approximately quaver = 60–63MM
Bars 6–18 Sudden acceleration to quaver = 80MM where there is more
activity in the bass
Bar 19 Accelerando to quaver = 100–104MM for the repeated dramatic
bass figurations
Bar 22 Sudden broadening to quaver = 72MM for the new idea
Bar 24 Further broadening to quaver = 56MM for the notated calando
Bar 25 Slightly faster quaver = 69MM
Bars 26–30 Average tempo of about quaver =
including repeat 76–80MM; in general phrase endings are broadened
Bars 30–36 Slower tempo quaver = 66–69MM, with laboring of the figures in
including repeat bars 30 and 31; note that in bar 36 the final poignant figure is
played exaggeratedly slower at about quaver = 46MM
Bars 36–55 In this Allegro section the tempo picks up through bar 36 to about
crotchet = 152M; there are agogic lengthenings, particularly in the
rests at bars 44, 53, and 54
Bars 56–72 The tempo is immediately slower for this lyrical section at about
crotchet =120–26MM; at bar 62, the tempo recedes for the
presentation of the theme in the minor to crotchet = 112MM; after
this, there is a very noticeable accelerando through bars 64 to 68,
reestablishing the tempo of about crotchet = 152MM
Bars 86–101 The prevailing tempo is about crotchet = 52 but within this various
figures are played noticeably slower, such as the falling figures in
bars 88 and 91, and particularly the poignant melodic sequence
in bar 99
Bars 102–24 The tempo crotchet = 52MM is reasserted, but, here again, various
poignant melodic figures such as at bars 107 and 116 are
broadened very noticeably

Merrick, Practising the Piano, 74–75.
274 off the record

Table 5.20 Chopin Nocturne Op. 27 No. 2, Bars 1 to 45, Leschetizky, Piano
Roll Recording, 1906 (Audio Ex. 5.29 )
Bars 1–8 Approximately 72MM; Leschetizky makes agogic lengthenings
of the first bass note and sometimes also the bass note on the
second half of the bar
Second half of Broadening to expressive note on the downbeat of bars 9,
bars 8, 11, and 13 12, and 14
Bars 22–25 There is an exaggerated broadening reducing the tempo from
about quaver = 72MM to 63MM and finally to 48MM for the
rising semiquaver melody at the end of bar 25
Bar 26 Downbeat is approximately doubled in length
Bars 37–45 The tempo increases from about 72MM in bar 38 to 96MM in
bar 42; this matches and greatly enhances the momentum
already built into the music
Bar 45 Exaggerated broadening in bar 45, drawing out and enhancing
the expressive chromaticism

Figure 5.10 Chopin Nocturne Op. 27 No. 2, bars 8 to 9, ed. Leschetizky.

This glacial shift is borne out in the time lapse readings presented in Table 5.1,
particularly those of Solomon Cutner, who made markedly less tempo variation
than pianists before him.
Indeed, written texts by Leschetizky’s teaching assistants discuss the types of
tempo modification that were considered artistic. For example, Prentner pro-
motes rhythmic stability as the chief requisite for artistic piano playing and is
critical of “the constantly increasing haste, and the senseless scrambling” of some
players. To counteract this, she advises “a decided ‘holding back’ at the last part
of a bar, in order to prevent precipitation in reaching the first beat of the next.”37

Prentner, The Leschetizky Method, 73.
Te mpo Modif i cati on 275

In her opinion, this particularly amateurish lack of sense of rhythm was charac-
teristic of many players who indulged in the “conventional hurrying in forte and
slackening in the piano passages.” To give “breadth and swing” to the interpreta-
tion calls for the exact opposite: in exceptional circumstances “an accelerando
accompanies a ff, and a ritenuto a pp.” Later, Prentner confirms that an accele-
rando or ritenuto might occur “in one or several bars,” but that the tempo must
afterward be restored so that any change is inconspicuous.38 Bülow, too, was in
favor of broadening the tempo at powerful moments. In an annotation to
Beethoven’s Sonata Op. 57 Appassionata, he comments that “in general, the quick-
ening of the movement at an increase of power, is likely to paralyze rather than
to promote the energy of the expression.”39
Prentner’s seemingly unconventional marrying of forte with broadened or held
tempo, and piano with increased tempo seems to have had historical precedents.
For example, August L. Crelle states in 1823 that although “an exact and strictly
measured tempo is an essential aspect of music,” certain changes are appropriate.
Among these, all strengthened notes should not hurry and “as a rule, the begin-
ning of a musical unit commences powerfully and importantly, the middle carries
on in a measured and regular manner and the end increases in speed and decreases
in power.”40
Brée’s advice about tempo modification accords with and helps clarify some of
Prentner’s views. According to Brée, Leschetizky approved of “continual changes
in the tempo” based on contrasts in the movement. No composition, she asserts,
is played from start to finish in the same tempo, but any changes are to be incon-
spicuous. Giving the example from Paderewski’s Légende (Fig. 5.12), she advises
the following:

The changes in tempo must be so delicately graded that the hearer notices
neither their beginning nor their end; otherwise the performance would
sound “choppy.” Thus, in a ritardando, calculate the gradual diminution of
speed exactly, so that the end may not drag; and conversely in an accele-
rando, that one may not get going altogether too fast. In a ritenuto, more-
over, many play the final tone a trifle faster, which abbreviates the
ritenuto and gives the hearer a feeling of disappointment. Where an
a tempo follows, it should quite often not be taken too literally at the very
outset, but the former tempo should be led up to gradually;—beginning
the reprise of the theme like an improvisation, for instance. Thus, in the
course of one or two measures, one would regain the original tempo.

Prentner, The Leschetizky Method, 73.
Beethoven, “Sonata Op. 57,” Sonatas and Other Works, ed. Bülow, part 3, vol. 4, 62.
August L. Crelle, Einiges über musikalischen Ausdruck und Vortrag (Berlin: 1823), 61; trans in
Brown, Classical and Romantic Performing Practice, 386.
276 off the record

Figure 5.12 Paderewski Légende, annotated by Brée.

She goes on to advise, however, that “where the character of the composition
requires it, begin the a tempo immediately at the original pace.”41 Her example is
taken from a Prelude by Schütt (Fig. 5.13).
Although providing useful guidance about the use of tempo modification,
Brée’s and Prentner’s descriptions do not clearly indicate the limits within which
it was considered appropriate and in good taste. And their advice certainly does
not convey the flavor of Leschetizky’s practices as evidenced on his piano rolls.
To our ears, his tempo modifications are not always inconspicuous. In many
instances, the rate of ritardando, accelerando, or the effect on bar duration caused
by lengthening individual notes does not seem subtle or unnoticeable. Though,
for many musicians of his era, they may have sounded so.
The playing of Grieg—another of the oldest generation to record—provides a
further interesting case study. Written accounts of Grieg’s playing give the gen-
eral impression that he was a refined musician and an elegant, tender, and neat
player who did not indulge in the type of exaggerated expression that other
virtuosi apparently readily admitted.42 In 1888, the Times (London) noted that

Figure 5.13 Schütt Prelude, annotated by Brée.

Brée, The Groundwork, 69.
Henry T. Finck, Grieg and His Music (New York: Lane, 1910), 110.
Te mpo Modif i cati on 277

Grieg exhibits “a composer’s touch on the piano,” which for his own works “gives
them a peculiar charm of their own.”43 In the same year, the Musical Times (London)
mentioned that in Grieg’s playing of a particular solo, “Nothing could be more
neat, clear, and intelligent.”44 And commenting, in 1897, upon Grieg’s perfor-
mance in New York of some of his own Lyrische Stückchen Vol. 3 Op. 43, the Musical
Courier extols his playing for “the utmost delicacy and a rare sympathy of touch of
softer, finer quality,” as well as “the remarkably strong manner in which he brought
out all that was ‘characteristic’ in each section.”45
Although the foregoing written references give a very favorable impression
of Grieg as a composer/pianist, they do not convey the aural effect of his
musicianship. We are no closer to knowing what distinguished his peculiar
charm; how neat, delicate, and elegant his performances sounded by modern
standards; or in what way he strongly brought out the characteristics of individual
Grieg’s 1903 recordings show that his playing does, to some extent, reflect the
preceding descriptions. As we have seen, he used dislocation and arpeggiation far
less frequently than many other pianists of his generation. In this sense, his play-
ing sounds neat and synchronized. But in some of the works he recorded, Grieg
employs tempo modification more frequently than he notated and in a manner
that is strikingly free. A good example of this can be heard at the beginning of the
“Alla Menuetto” from his Piano Sonata Op. 7, where the tempo dramatically accel-
erates between bars 1 and 8 (Fig. 5.14 ; Audio Ex. 5.30 ). Bar 1 commences at
approximately crotchet = 76MM. By bar 3, the tempo has increased to about
84MM, bar 5 to 92MM, and by bar 8 to about 112MM. Grieg slightly prolongs the
last beat of bar 8 and proceeds with the repeat of the theme at approximately
96MM. This very dramatic and unexpected increase of tempo matches and
enhances the buildup of tension accompanied by the crescendo from piano to for-
tissimo. It would, however, never have been expected or extrapolated from his
musical text alone, especially from the title “Alla Menuetto,” which now suggests
a graceful and measured dance. That Grieg particularly intended this effect is con-
firmed by the fact that where the passage is repeated at bar 72, he makes a very
similar accelerando.
In To Spring Op. 43 No. 6, Grieg’s designated “Allegro appassionata” marking
is achieved by fairly sudden and exaggerated tempo alteration. The first
section commences at and maintains an average tempo of approximately dotted
minim = 112MM. Between bars 11 and 13, there are two noticeable ritardando/
accelerando patterns enhancing the poignant arpeggiated chords in the bass
(Fig. 5.15 ; Audio Ex. 5.31 ). At bar 15, the indicated “rit. molt” is achieved

Finck, Grieg and His Music, 100.
Finck, Grieg and His Music, 100–101.
Finck, Grieg and His Music, 109–10.
278 off the record

less by a real tempo modification than by accentuation of the repeated right-hand

chords. Between bars 16 and 18, there is what can only be described as a frantic
accelerando, at which point, curiously, the notation indicates a tempo (Fig. 5.15).
Here, the tempo increases to about 126MM, giving the feeling of anxious restless-
ness. But perhaps the most significant variation of tempo is heard in the section
between bars 23 and 44 (Fig. 5.16 ; Audio Ex. 5.32 ). The music comes almost
to a halt at the end of bar 22, after which there is an accelerando to approximately
108MM in the middle of bar 25, followed by a slight ritardando. The following
four-bar phrase accelerates in a similar manner to approximately 120MM. By bar
33, the tempo has increased to 132MM. The rate of accelerando and the variation
from approximately 108MM to 132MM would not, according to modern stan-
dards, have been deduced from Grieg’s indication “stretto poco a poco” in bar 27.
Finally, in the section between bars 37 and 44, Grieg elongates very exaggeratedly
each bar by extending the notated crotchet rests to dotted minim rests. The result-
ing agitated effect could not have been appreciated without Grieg’s recording,
nor extrapolated from his notation.
In the lyrical “Tempo di Valse” movement Remembrances Op. 71 No. 7, Grieg
often modifies the internal waltz rhythm so that, instead of three equal crotchet
beats, there is a slightly longer first beat, a slightly shortened second beat and a
regular third beat (Audio Ex. 5.33 ). This is opposite to the Viennese tradition
reflected in Leschetizky’s practice (elucidated by Brée) involving an abbreviated
first beat and a lengthened second beat.46 Perhaps more significant, however, are
the myriad tempo changes that he seemingly employs to delineate individual
phrases. During the first two phrases, with their characteristic syncopated tune,
the tempo is slightly increased toward the minim at the height of each phrase
and is reduced toward the end. During the second half of the third phrase, con-
taining identical material to the first half, the tempo increases very significantly,
resulting in a sense of rushing. As if to compensate, the tempo is subsequently
broadened noticeably, and suddenly the stipulated “poco rit” comes into action in
the penultimate bar of the section (bar 23). The effect is an emphasis of the dis-
sonant harmony preceding the close of the section. This process occurs continu-
ally throughout the rest of the movement. Notably, each new dynamic marking is
subtly varied in tempo so that, for example, pianissimo tends to be very slightly
slower, while crescendo tends to push forward. However, the “cres. molto” at bar
45 is accompanied by a prominent broadening of the tempo. In general, the tempo
modifications used here are subtler than in other works, perhaps because of the
simplicity and waltz-like character of the movement.
In the abridged version of the “Finale” to his Piano Sonata Op. 7, tempo
modification can particularly be heard in the final thundering section of the
movement from bars 313 to 334 (Fig. 5.17 ; Audio Ex. 5.34 ). Here, for the

See chapter 4, page 204.
Te mpo Modif i cati on 279

fff sempre grandioso, the prevailing tempo is broadened from approximately

dotted crotchet = 116MM to 96MM and even more toward the final “Presto.” The
only indication of a tempo change notated by Grieg is the “ritard.” at bar 333,
which hardly conveys such an extreme tempo modification. At the “Presto”
(bar 334), Grieg sets off at a tempo of approximately dotted crotchet = 120MM.
In other works, such as Gangar Op. 54 No. 2, Wedding Day at Troldhaugen Op.
65 No. 6, and Bridal Procession Op. 19 No. 2, Grieg uses less tempo modification,
adhering more closely to his notation. He may perhaps have regarded these works
as requiring less variation, because of their simpler character.
Is there a discernable pattern in Grieg’s use of tempo modification? Broadening
of the tempo seems to be used primarily to delineate between sections and to
mark the closure to a composition. Broadening and hastening were both appar-
ently used to enhance dramatic effect, particularly accompanying an increase in
dynamic level. Hastening also increases the excitement of the music, giving a pre-
cipitous or tumbling effect to the overall character. Hastening toward a phrase
climax and broadening afterward gives a particular shape to the phrase.
Prolongation of certain notes and rests produces a marked emphasis, bringing
these into the foreground of the texture. The manner in which Grieg modifies
tempo is in stark contrast with present-day practice. Moreover, accounts of
Grieg’s playing do not convey very important and intrinsic elements of his playing
style. A Grieg-style performance based on the written evidence alone would
undoubtedly differ markedly from the style preserved in his recordings.
Pugno—a pianist of the generation after Brahms, Reinecke, and Leschetizky—
made recordings in 1903 that afford direct comparison with his verbal advice.
Looking, for example, at his thoughts on Chopin’s Nocturne Op. 15 No. 2, he evi-
dently practiced some of what he preached. But there are many features of his
playing that are not mentioned and some that sound quite different from his own
description. Pugno states that the mood of the entire first part of the Nocturne is
of peacefulness and resignation and should therefore “be played with absolute
tranquillity.”47 Indeed, the first four bars in his performance remain at a constant
tempo and achieve these effects. However, there is precipitation of tempo through
the four semiquavers at the beginning of bar 5 (Fig. 5.18)48 and a compensatory
slowing toward the end, in such a manner that the overall length of the bar
remains practically unaltered (Audio Ex. 5.35 ). This occurs despite his warning
that, throughout the piece, one must “never curtail these demisemiquavers [sic]
[semiquavers], nor the fourth quaver in any bars.”49 Such modification causes a
definite disturbance to the original atmosphere. But an even greater disturbance

Pugno, The Lessons, 66.
Pugno, The Lessons, 67.
Pugno, The Lessons, 66.
280 off the record

Figure 5.18 Chopin Nocturne Op. 15 No. 2, bar 5, ed. by Pugno.

is felt at bar 11 (Fig. 5.19). In his advice, Pugno is quite specific about the manner
of playing the right-hand grace notes:

Play this figure with precisely the division indicated by the lines. To
obtain all the fluidity essential to its rendering, play it smoothly without
shading, and with a slight retard only on the four or five concluding
notes. Prolong the fourth quaver in the bass so as to let the right hand
play the last twelve notes of the bar without hurry.50

Apart from the lengthening of the fourth quaver, Pugno’s description and nota-
tion gives no indication that the left-hand accompaniment figure is to be altered.
In his performance, however, he makes quite a dramatic alteration to the tempo
(Audio Ex. 5.36 ). The first three quavers in the left hand are played almost as
semiquavers, and the fourth quaver is lengthened to make up the time of the bar.
The effect is a sudden, dramatic, and unprepared più mosso followed by a ritar-
dando. An approximate notation of this is cited in Figure 5.20. Certainly, “peace-
fulness and resignation” have momentarily disappeared. Pugno’s indulgence in

Figure 5.19 Chopin Nocturne Op. 15 No. 2, bar 11, ed. by Pugno.

Pugno, The Lessons, 67.
Te mpo Modif i cati on 281

Figure 5.20 Chopin Nocturne Op. 15 No. 2, bars 10 and 11, Pugno, recorded in 1905
(Audio Ex. 5.36 ).

tempo modification is carried to an extreme in bars 13 and 14 (Figure 5.21 ),

where the tempo is practically doubled without restitution.51 For these two bars,
Pugno advises the player to “develop the tone colour and warmth.”52 In his perfor-
mance, this translates to a sudden increase in tempo that corresponds with the
poco crescendo and the con forza indication (Audio Ex. 5.37 ). Nowadays, such an
effect would hardly be associated with the development of tone color and warmth,
which suggests something more akin to dynamic shading. In the return of the
calm opening section, Pugno makes a very similar acceleration during bars 52 and
53, confirming that this was indeed an intended effect. This exaggerated hasten-
ing is heard as a reinforcement of the passage leading to the marking con forza.
Another curious practice in Pugno’s performance is the elongation of trills. At
bar 7 (Fig. 5.22 ),53 he instructs the player that the trill should (apparently for
expressive reasons) be rather long, and shaded with a decrescendo.54 However, his
description gives no clues as to the time lapse involved. In reality, here and in
other similar places, Pugno lengthens trills by a significant amount and well
beyond what might be expected today (Audio Ex. 5.35 ). In this place, the
lengthening of the trill effectively adds a whole beat to the bar making five quaver
beats instead of four. Further to this, at bar 15, Pugno states that the trill should

Pugno, The Lessons, 67.
Pugno, The Lessons, 67.
Pugno, The Lessons, 68.
Pugno, The Lessons, 67.
282 off the record

be prolonged like the first time.55 On this occasion, he makes an extraordinary

lengthening that includes not only the trill but also the end of the bar forming the
close of the first part of the A section (Audio Ex. 5.37 ). This prolongation
extends the bar to approximately seven quaver beats: the trill and its termination
have three quaver beats, and the following melodic pattern A–G is elongated into
a dotted quaver and semiquaver respectively. The extent of this tempo modifica-
tion is simply not conveyed in Pugno’s verbal description. In the following section,
Pugno’s instructions contradict the nuances indicated by Chopin, but in a manner
that could scarcely be predicted from his own verbal advice (Fig. 5.23 )56
At the upbeat to bar 17, he states that the playing should be “clearer here, but
unchangingly expressive.” And during bar 17, he advises that the two first Es
should be “very much stressed” while the third E (the crotchet in the following
bar) should be elongated a little and played very softly.57 In his performance,
Pugno slightly increases the tempo at bar 17, imbuing the phrase with forward
momentum. At face value, his instruction implies that the two melody notes E
should be emphasized or accented in some way. In fact, he gives them emphasis by
broadening the tempo considerably and making a corresponding decrescendo to
the third E, which is dynamically the softest. After this, the overall tempo is
increased, but the grace-note roulades are played with expressive ritardandi for
each successive phrase until bar 22, at which point the sense of con forza is clearly
felt (Audio Ex. 5.38 ). Again, none of these vital aspects of Pugno’s performance
could have been extrapolated directly from his written comments or notation.
Several other examples serve to demonstrate the ambiguity of Pugno’s written
texts. At bar 22, Pugno suggests “a passionate rubato movement,” giving the
impression to us now that some very noticeable tempo modification ought to take
place.58 This is not the case. Although there is a slight accelerando during bar 21,
the increase of passion is effectively created by introducing metrical rubato altera-
tions. These alterations are also made in bar 22, where Pugno plays with a fuller
sound and slightly broadens the tempo (Audio Ex. 5.38 ). And for the “Doppio
movimento” section of the work from bars 25 to 48 (Fig. 5.24 ),59 Pugno advises
the player to “double the pace. Begin by playing very smoothly the groups of semi-
quavers. Despite the half-light, the undulation, the imprecision of these first eight
bars, it is necessary to make the melody stand out.”60 On his recording, Pugno
does in fact bring out the melody notes by accenting them. However, what is
not conveyed by his advice is the accelerando he makes from the beginning of the

Pugno, The Lessons, 68.
Pugno, The Lessons, 68.
Pugno, The Lessons, 68.
Pugno, The Lessons, 68.
Pugno, The Lessons, 68.
Pugno, The Lessons, 68.
Te mpo Modif i cati on 283

section at bar 25 until bar 42. Having commenced the section at approximately
crotchet = 80MM (double the speed of the opening), the tempo accelerates to
108MM by the climax of the section at bar 39 (Audio Ex. 5.39 ). The only refer-
ence Pugno makes to an increase of drama during this section is at bar 33, where
he advises that there should be “a great crescendo lasting until the fortissimo, which
is the culminating point of this very impassioned section.”61 Quite obviously,
Pugno felt no need to mention the increase of tempo, probably taking it for
granted that this was implied by the increase of passion.
In addition to accelerations and decelerations, Pugno employed expressive
agogic lengthenings, particularly noticeable at the beginning of the work. Here,
in comparison with the tempo of the first full bar, the anacrusis melody notes,
A and B, originally notated as a dotted semiquaver followed by a demisemiquaver,
are approximately twice the length, thus sounding like a dotted quaver followed
by a semiquaver (Audio Ex. 5.35 ). In his written advice, Pugno makes no
mention of this. At the upbeat to bar 9—a decorated version of the opening
(Fig. 5.25)—Pugno makes almost exactly the same lengthening on the melody
note A. At this point he advises the player to “spread out the chord very broadly
from the first note in the bass to the A which begins the melody again.”62 He does
not, however, provide any indication of the lengthening of the melody note A.
Quite clearly, Pugno’s verbal advice does not convey many of the important
tempo modifications he employed in Chopin’s Nocturne Op. 15 No. 2. The fea-
tures and the frequency of these could not have been deduced from his written
texts alone. They were, however, integral to his expressive palette and gave an
individual poignancy to his playing.
Like Pugno, the opinions of Paderewski concerning tempo modification
are particularly valuable since they too afford direct comparison with his own

Figure 5.25 Chopin Nocturne Op. 15 No. 2, bar 9 with upbeat, ed. by Pugno.

Pugno, The Lessons, 69.
Pugno, The Lessons, 67.
284 off the record

recordings. On several occasions, Paderewski reasoned that flexibility of tempo

was essential to artistic piano playing. In “Tempo Rubato” (1909), he explains that
soloists in particular must be given freedom and “discretional power” to use
tempo rubato.63 This he defines as “a more or less important slackening or quick-
ening of the time or rate of movement.” In solo playing, he denounces any notion
of compensation, explaining that “the value of notes diminished in one period
through an accelerando, cannot always be restored in another by a ritardando.
What is lost is lost.” When playing with an orchestra, on the other hand, compen-
sation is both “natural” and necessary.64 In this, he appears to be describing
metrical rubato.
Comparing tempo modification with the art of oration, Paderewski summa-
rizes eloquently some of the effects that may be achieved by its use:

[Tempo modification] emphasizes the expression, introduces variety,

infuses life into mechanical execution. It softens the sharpness of lines,
blunts the structural angles without ruining them, because its action is
not destructive: it intensifies, subtilizes, idealizes the rhythm. . . . [I]t
converts energy into languor, crispness into elasticity, steadiness
into capriciousness. It gives music, already possessed of the metric and
rhythmic accents, a third accent, emotional, individual.65

Further to this, he suggests that—for all composers—musical expressions such

as expressivo, con molto sentiment, con passione, and teneramente demand a degree
of emotion that is successfully expressed through tempo modification. The simi-
larity between this and Corri’s association of various terms with arpeggiation, for
example, is obvious. To safeguard against exaggeration, Paderewski underlines
the need for a “real knowledge of different styles, a cultured musical taste, and a
well-balanced sense of vivid rhythm.”66 This is sage advice to be sure. But can we
appreciate now what he or other musicians of his era considered tasteful or well
Paderewski’s support of tempo modification as an expressive device is clear.
But he was certainly disapproving of its inappropriate use. Around 1895, he
remarked that

only too many think that they display a vast deal of feeling if they make
frequent ritardandi and long pauses on single notes. I would call this
over-sentimentalism simply the abuse of rhythm. The only way to avoid

Paderewski, “Tempo Rubato,” 27.
Paderewski, “Tempo Rubato,” 30–31.
Paderewski, “Tempo Rubato,” 30.
Paderewski, “Tempo Rubato,” 32.
Te mpo Modif i cati on 285

this is to keep strictly as possible to the rhythm and the tempo. Nothing
is to be gained by such affectation but distortion of the composer’s ideas.
Under the same head comes the exaggeration of the rubato, so deplorably
frequent in the playing of Chopin. This springs from the same mistaken
notion that it adds feeling and character. The only remedy of the fault is
to stick closely to both rhythm and tempo.67

Paderewski’s displeasure is clear. Yet his words do not clearly convey what he was
railing against: what frequency of ritardandi and what length of pauses caused
Paderewski’s recordings made between 1911 and 1930 reveal that he used
tempo modification both more frequently than indicated by the composer’s
notation and much more than might be extrapolated from his own advice.
According to current canons of good taste, his tempo modifications often seem
erratic and exaggerated, giving the impression that his playing was somewhat
uncontrolled. However, close scrutiny reveals the pattern in these modifications:
they were not simply aberrations or moments of extreme fancy. Paderewski’s
tempo modifications occur in several different ways. Often they consist of no
more than the lengthening of or lingering over a single note or moment in a
phrase. At other times, they consist of a hastening or slackening of the tempo as
a means of emphasizing the climax of a phrase or to mark its close. At yet other
moments, the subtle but noticeable modification of the tempo of an entire
passage is used to emphasize its expressive effect.
In his 1912 recording of three movements from Schumann’s Fantasiestücke
Op. 12, Paderewski makes several significant tempo modifications. In Des Abends
Op. 12 No. 1, the overall tempo is approximately quaver = 60MM. In the first sec-
tion from bars 1 to 16 and during its repeat, he noticeably lingers and therefore
stretches the tempo at particular moments (Fig. 5.26 ; Audio Ex. 5.40 ).68
He lingers over the upbeat to and the downbeat of bar 5 in order, it seems, to mark
the repetition of the musical thought. Additionally, there is a particularly notice-
able lengthening of the upbeat to and downbeat of bar 12, enhancing the expres-
sion at the culmination of the phrase. The tempo is immediately resumed at bar
13. To finish the section, Paderewski broadens the tempo very noticeably during
bars 15 and 16. During the following section, Paderewski uses tempo modifica-
tion to emphasize the effect of a poignant inner part in Schumann’s texture
(Fig. 5.27 ; Audio Ex. 5.41 ).69 He commences bar 17 slightly faster than
quaver = 60MM and, by making a ritardando at the end of bar 20, prepares a
slower tempo of about quaver = 54–56MM for the section from bar 21 to bar 24.

Paderewski, “The Best Way to Study the Piano,” vii.
Robert Schumann, “Des Abends Op. 12 No. 1,” ed. Clara Schumann, 80.
Schumann, “Des Abends Op. 12 No. 1,” ed. Clara Schumann, 80.
286 off the record

A similar pattern of tempo alteration occurs between bars 25 and 31 containing

similar musical material. Clearly, on this occasion, Paderewski’s tempo modifica-
tions were carefully planned. Following this, between bars 32 and 35, the tempo
is accelerated slightly as if to compensate for the previous broadening. In bar 36,
there is a significant retardation of the tempo and the link section (bars 37 and
38) is played with a certain freedom of tempo. A further significant tempo
alteration occurs in the coda section from bars 77 to 88, heightening the senti-
mental and nostalgic effect of the music (Fig. 5.28 ).70 Here, by making a notice-
able retardation in the link passage at bar 76, the tempo of the coda—about
quaver = 54MM—is well prepared. And at bars 82, 84, and elsewhere within the
coda, he makes little retardations that heighten the expressiveness of the section.
Quite obviously, Paderewski made many unauthorized and noticeable tempo
modifications to Schumann’s original text.
Paderewski’s rendition of Schumann’s Aufschwung Op. 12 No. 2, provides fur-
ther fascinating examples of his use of tempo modification. Having been so criti-
cal of those who made use of frequent ritardandi, he appears to fall prey to the
same temptation. In addition, he uses accelerando very frequently in order, it
seems, to enliven the spirit of particular passages. Keeping in mind that Schumann
indicated ritardandi only four times at bars 31, 70, 83, and 137, Paderewski adds
approximately thirteen further large-scale ritardandi and several smaller-scale
lingerings, as well as many subtle inflections which are almost impossible to
notate. In addition, he often extends ritardandi, starting them earlier than
notated by Schumann. A good example of this is where Schumann marks ritar-
dando in bar 70 but Paderewski dramatically slows down during bars 68 and 69
(Fig. 5.29 ; Audio Ex. 5.42 ).71 At other moments, ritardandos appear to mark
the end of a phrase, such as at bars 7, 27, 38, and similar places. Sometimes the
ritardandos take an extreme form, bringing the music almost to a standstill.
Between bars 50 and 52, a ritardando marks the division of one section from
another—that is, it distinguishes between the stormy theme and a quieter, more
lyrical episode (Fig. 5.29). Elsewhere, ritardando appears to have a dual purpose,
both marking the end of the phrase and making a compensation for a previous
acceleration. This is particularly noticeable between bars 53 and 64, where a pat-
tern of accelerando followed by ritardando occurs twice in succession (Fig. 5.29).
Large-scale tempo modification also appears to enhance the expressive effect
of an entire section in Paderewski’s performance of this movement. In general,
lyrical episodes are played in a somewhat slower tempo, often with extra expres-
sive lingerings at phrase climaxes. For example, the episode that commences
halfway through bar 16 is played at a slower tempo and the highest points in
the phrase are prolonged. At other times, a slower tempo seems to help to

Schumann, “Des Abends Op. 12 No. 1,” ed. Clara Schumann, 81.
Schumann, “Aufschwung Op. 12 No. 2,” ed. Clara Schumann, 83.
Te mpo Modif i cati on 287

emphasize an inner voice, an example of which occurs between bars 20 and

24 (Fig. 5.30 ).72
No less significant are Paderewski’s numerous accelerandos. Sometimes these
act as a transition between sections of varying character and tempi. At other
moments, accelerando enhances the excitement of a particular passage. The most
noticeable of these occur between bars 71 and 82, and bars 105 and 114. During
the latter, the tempo varies between approximately dotted crotchet = 104MM and
120MM—a large variation in tempo by modern standards (Fig. 5.31 ; Audio
Ex. 5.43 ).73 From this evidence, it is abundantly clear that Paderewski varied
the tempo frequently and widely.
Paderewski’s recording of Schumann’s Warum? Op. 12 No. 3 provides other
notable examples of his tempo modification (Fig. 5.32 ).74 It reveals that he
made certain large changes of tempo to color the musical character. Paderewski’s
performance starts at about crotchet = 48MM, but the tempo is hardly ever con-
stant and fluctuates in almost every bar. For example, there is a marked broaden-
ing toward the high F in bar 3, after which the tempo is further broadened. In
bar 7, there is a very sudden acceleration toward the highest note of the phrase.
This is coupled with an unexpected decrease in volume, creating an unusually
poignant effect. The same technique is noticeable for the climax in bar 12, after
which Schumann’s notated ritenuto is observed. The second section from bar
17 commences at approximately 48MM and is followed by noticeable forward
surges that enhance the passionate nature of the music (Audio Ex. 5.44 ). From
bars 21 to 23, Paderewski increases the tempo to approximately 69MM, and from
bars 25 to 27, there is a further increase to 72MM. This constitutes a considerable
divergence from the initial tempo. Further impact is achieved by broadening the
tempo between bars 27 and 30 to the vicinity of the high 30s MM. So, although
Schumann notates a rit. at the end of bar 30, Paderewski commences it four bars
early. The section from bar 31 to the end is played nostalgically at a tempo
approaching 38MM, markedly slower than the original tempo. And within this,
there are fluctuations, particularly in the middle of bars 35 and 39, where the
tempo increases slightly toward the highest note. Paderewski finishes the section
with a broadening that brings the piece almost to a standstill, in spite of the fact
that the repeat of the B section is yet to come.
The examination of a few of Paderewski’s recordings reveals that many types of
tempo modification were intrinsic to his expressive piano technique. He used
tempo modification in a manner that enhances phrase shapes and delineates
musical structures. Most important, the recordings preserve the characteristic

Schumann, “Aufschwung Op. 12 No. 2,” ed. Clara Schumann, 82.
Schumann, “Aufschwung Op. 12 No. 2,” ed. Clara Schumann, 84.
Schumann, “Warum? Op. 12 No. 3,” ed. Clara Schumann, 86.
288 off the record

features and the frequency of his tempo modifications that were simply not
conveyed in his verbal advice.

The Hidden Meanings in Written Texts

Late-nineteenth-century written texts imply that tempo modification of various
forms was considered an essential aspect of any musically satisfying performance;
they also document the controversy that surrounded so-called unauthorized
changes of tempo. In 1877, Pauer advised that such expressive means as “to hurry
or accelerate the movement (accelerando or stringendo), or to lessen and decrease
the movement (ritenuto or rallentando)” should not be applied arbitrarily. For him,
the “real beauty and effect” of tempo modification was governed by general laws
of interpretation such as its “well-defined and carefully-weighed gradations,”
“regulated growth and decline,” and “increasing animation, and almost impercep-
tible return to calmness and quiet.”75 Clearly, however, Pauer’s verbal expressions
do not convey the type and frequency of tempo modifications that were the
hallmark of many late-nineteenth-century pianists.
Other late-nineteenth-century musicians expressed concern for the type of
tempo modification that led to exaggeration. In his Katechismus des Klavierspiels
(1888), translated as Catechism of Pianoforte Playing (1892) Hugo Riemann criti-
cized conceited pianists who, by departing altogether from convention, produced
“incredible interpretations.” His first concern was about the agogic lengthening
of angular notes particularly when they appear as they most regularly do in the
“crescendo-part” of a phrase, because the stringendo effect that naturally accompa-
nies it becomes distorted. He considered this effect to be very charming when
used sparingly, but disagreeable when misused. Riemann’s other main concern
was for “the disdainful hurrying over the diminuendo-part, the careless, frivolous
rushing over ground attained, conquered.”76 Although it is impossible to tell from
this what frequency or degree of lingering on single notes Riemann considered an
abuse, it is worth noting that the practice itself was apparently already well estab-
lished. For example, Türk had already described in the late eighteenth century
“the possibility of lingering somewhat longer on a very important note than on
one less important.” He advised that a long note should “not be lengthened more
than half of its value.” He further explained: “Holding a note for a longer or shorter
time depends also on the length of the note and its relationship to the others. . . .
[O]ne can linger longer on a quarter note than on a sixteenth.”77

Pauer, Art of Pianoforte Playing, 67.
Riemann, Katechismus des Klavierspiels (Leipzig, Germany: 1888); trans. as Catechism of
Pianoforte Playing (London: Augener, 1892) 79.
Türk, Klavierschule, 328.
Te mpo Modif i cati on 289

It appears, too, that incredible interpretations of one sort or another had

already infiltrated piano playing by the first half of the nineteenth century. In
1805, Adam complained that certain pianists made it fashionable not to play
in time and “to execute all pieces of music like a fantasy, prelude or caprice.”78 In
1828, Hummel described this supposed malady as “a capricious dragging or slack-
ening of the time (tempo rubato), introduced at every instant and to satiety.”79
And, in 1846, Czerny stated that the strict keeping of time had been “almost
entirely forgotten” and that arbitrary changes of tempo—both acceleration and
deceleration—were then “often employed even to caricature.”80
By the second half of the nineteenth century, a certain degree of tempo modi-
fication was most likely considered indispensable in any effective piano playing.
In 1855, Reinagle refers to a compensating style of tempo modification: “Tempo
rubato implies robbing one bar, or part of a bar, for the sake of enriching another
which is considered of greater importance.”81 In 1861, Kullak hails it as “a feature
essential to the beauty of musical expression.” But he advises that, of all expres-
sive nuances, tempo modification must be used more sparingly if the melodic
material is to retain its rhythmic form and “poetic meaning.”82 But just how spar-
ingly such modifications were employed in reality is open to debate. Indeed,
Liszt thought of tempo modification among other devices as essential in orches-
tral performance. In the “Preface” to his Symphonic Poem Prometheus (1856), he
recognizes the difficulty of making these sensible in the notation:

It is not enough for a composition to be regularly beaten and mechani-

cally performed with more or less correctness, to make its author proud
of this way of the propagation of his Work, nor for him to recognise in it
a truthful interpretation of his thoughts. . . . I have endeavoured to make
my intentions about dynamics, accelerations and delays of speed, etc. as
sensible as possible by providing exact signs and marks of expression;
nevertheless it would be an illusion to believe that the beauty and the
character of a performance could be set forth on paper.83

Adam, Méthode, 160.
Hummel, Art of Playing the Pianoforte, 40.
Czerny, Art of Playing, 29.
Reinagle, A Few Words, 242.
Kullak, The Aesthetics, 281.
Franz Liszt, “Préface” to Prométhée, in Symphonischen Dichtungen für groses Orchester (Leipzig,
Germany: Breitkopf & Härtel, 1856), unpaginated 4: “Il ne suffit pas qu’une composition soit régu-
lièrement bâtonnée et machinalement exécutée avec plus ou moins de correction pour que l’auteur ait
à se louer de cette façon de propagation de son œuvre, et puisse y reconnaître une fidèle interprétation
da sa pensée. . . . Je me suis attaché à rendre mes intentions par rapport aux nuances, à l’accélération
et au retard des mouvements, etc. aussi sensibles que possible par un emploi détaillé des signes et des
expressions usitées; néanmoins ce serait une illusion de croire qu’on puisse fixer sur le papier ce qui fait
la beauté et le caractère de l’exécution.”
290 off the record

Other references reveal significantly less tolerance for tempo modification in

some circles. In 1872, the critic Hanslick berated Wagner’s conducting style for its
“intolerable arbitrariness” in terms of tempo modification. His overwhelming
concern was that orchestral performances would become infected by “that musi-
cal sea-sickness which so afflicts the performances of many singers and
instrumentalists.”84 Yet, a decade or so later in 1884, Hanslick confirmed that
metronomic strictness—if it ever really existed—had “been disavowed by all
modern conductors.” Describing a performance of the Meiningen Court Orchestra,
he shows positive acceptance for the style of tempo modification produced under
the direction of Bülow:

Bülow conducts the orchestra as if it were a little bell in his hand. The
most admirable discipline has transformed it into an instrument upon
which he plays with utter freedom and from which he produces nuances
possible only with a discipline to which larger orchestras would not
ordinarily submit. Since he can achieve these nuances securely, it is
understandable that he applies them at those places where they would
seem appropriate to him if he were playing the same piece on the piano.
It would be unjust to call these tempo changes “liberties,” since conscien-
tious adherence to the score is a primary and inviable rule with Bülow.
It is hard to draw the line. Opinion will vary according to individual
taste and the character of specific passages.85

An obvious anomaly presents itself here. How did Bülow make unauthorized
tempo changes while conscientiously adhering to the score? Perhaps in late-nine-
teenth-century parlance such an adherence incorporated a degree of freedom that
it no longer does. And, if it was difficult for Hanslick to draw the line between
what was tasteful and overdone, how much more difficult is it for us to appreciate
what he meant, over 100 years later?
Indeed, some sense of Bülow’s attitude to tempo modification may be gained
from his annotations to Beethoven’s Piano Sonatas. For example, for the powerful
passage from bars 117 to 135 in the first movement of Beethoven’s Sonata Op. 57
Appassionata, Bülow explains that a simultaneous increase in tempo (he marked
stringendo ad lib.) with an increase in dynamic is “decidedly allowable at this place
and the effect of a ‘heaven-storming’ climax aesthetically justified.”86 During the
climactic coda (from bar 325) of the third movement of the same Sonata, Bülow
again sanctions an accelerando explaining that “if one can quicken from here to
the end, the acceleration will well accord with the continuous urging towards

Hanslick, “Richard Wagner’s Concert (1872),” Music Criticisms, 106.
Hanslick, “The Meiningen Court Orchestra (1884),” Music Criticisms, 234–35.
Beethoven, “Sonata Op. 57,” Sonatas and Other Works, ed. Bülow, part 3, vol. 4, 62.
Te mpo Modif i cati on 291

the close.”87 For Bülow, increasing the tempo could also be used to enhance unusual
modulation. Referring to the passage from bar 240 to 246 in Beethoven’s Fantasie
Op. 77, he explains that “in the original stands ‘tempo primo.’ But if the former
tempo re-enters here, the peculiarly roguish charm of the modulation into C-major
disappears, and the crescendo in modulating back into B loses the splendor
imparted to it, acquired through the connection with an accelerando.”88 In other
instances, however, he cautioned against too much tempo fluctuation. Concerning
the passage commencing at bar 93 in the first movement of the Sonata Op. 81a
Les Adieux, he advises that “one fundamental tempo, one fundamental mood
should be maintained throughout.” He warns, however, that this “should never
rule with tyrannical rigidity,” but should adapt itself elastically “to the various
emotional sensations.” But in the end, he felt that such matters came down to
“the ‘undefinable’ educated artistic taste” and the individual temperament of
the artist.89 Though indefinable, it was, for Bülow and his circle, an educated
late-nineteenth-century artistic taste that dictated the acceptable boundaries of
tempo modification. If he was unable to define such things, how can we hope to
understand from his texts the subtleties that characterized this indispensable
Whereas some were enamored with Bülow’s style of tempo modification, others
were not so. For example, a concert review in the Musical Times (1884) slates his
performance of Sterndale Bennett’s Piano Sonata Op. 46 The Maid of Orleans for
its “exaggerated expression and an over-indulgence in the rubato-style.”90 It is a
shame that no recorded evidence of Bülow’s piano playing survives. Without
this, it is impossible to assess his playing style. It is likely though that Bülow’s
verbal advice regarding tempo modification gives an impression to us now that is
probably quite different from his actual practice.
The use of tempo modification remained a controversial subject during the late
nineteenth century, and, in some circles, conservatism appears to have reigned.
A review in the Musical Times (1885) of the pianist Mademoiselle Kleeburg—
probably Clotilde Kleeburg (1866–1909)—protests against a growing tendency
“to distort the works of classical masters.” The reviewer remarks that only few
composers of the highest rank, including Beethoven, remain safe, but the works
of others like Weber “suffer terribly at the hand of editors and executants.”
Kleeburg is accused of indulging “in tricks of style, especially unauthorised
changes of tempo” in her rendition of Weber’s Piano Sonata No. 2 in A flat Op. 39.
For this, insists the reviewer pedantically, “no excuse could be pleaded.”91

Beethoven, “Sonata Op. 57,” Sonatas and Other Works, ed. Bülow, part 3, vol. 4, 90.
Beethoven, “Fantasie Op. 77,” Sonatas and Other Works, ed. Bülow, part 3, vol. 4, 113.
Beethoven, “Sonatas Op. 81a,” Sonatas and Other Works, ed. Bülow, part 3, vol. 4, 145.
Anon., “Dr. Hans Von Bülow’s Recital,” The Musical Times, vol. 25 (1884): 337.
Anon., “Mdlle. Kleeburg’s Second Recital,” The Musical Times vol. 26 (1885): 402.
292 off the record

Other references give the impression that tempo modification was perfectly
acceptable to enhance the intended effect. Concerning the passage at the letter o
in the example from Mendelssohn’s Rondo Capriccioso Op. 14 (Fig. 5.33), the editor
of the Instructive Edition (1899) advised that “In case the tempo should have been
excusably accelerated during the foregoing passionate bars, the original speed of
the Principal Subject must be reasserted abruptly at this place,—possibly even a
trifle more moderate.”92 Indeed, in their 1891 edition, Lebert and Faisst promoted
the use of accelerando with notated crescendos in the third movement of
Beethoven’s Sonata Op. 27 No. 2 Moonlight.93
Indeed, tempo modification was sometimes considered particularly appropri-
ate for the expression of certain passionate feelings. For example, for the fifth
study marked “Allegretto agitato con passione” from Moscheles’s Studies Op. 70,
the editor Saunders advises that its character “admits of, and even requires, a
frequent deviation from the regular time.” The points at which such accelerations
and retardations should be introduced “is left to the taste and feeling of the
Other references attest to the widespread use of tempo modification. In “The
Training of a Chorus” (1900), British conductor Henry Coward (1849–1944) is
very positive about it as a means of “proper phrasing.” Citing various examples, he
explains that

one has only to recall the brightening effect of the accelerando with its
subsequent rallentando to the normal tempo of the choral part of “I waited
for the Lord” [Mendelssohn’s Hymn of Praise], from bar 71 to the re-entry
of the solo voices; the exquisite effect of the ritardando and pianissimo at
the close of [the chorus] “O pure in heart” (“Golden Legend”) [an Opera/
Oratorio by Sir Arthur Sullivan]; the imposing dignity of the broadening
out of the final of the Prologue and the overpowering majesty of the
swell at the 13th bar from the end of the Epilogue of the same work.

To those who might object to such changes because they are not marked in the
score, Coward retorts that composers—not always realizing the best way to
express their own music—give only general directions, leaving the rest to the
performers. He adds that it may be assumed that composers reasonably expect
conductors to mix their ideas with their methods of preparing the music.95

Felix Mendelssohn-Bartholdy, “Rondo Capriccioso Op. 14,” Instruktive Ausgabe Klassischer
Klavierwerke, ed. Percy Goetschius (Stuttgart, Germany: Cotta, 1899), 83.
Beethoven, “Sonata Op. 27 No. 2,” Sonatas and Other Works, ed. Lebert and Faisst, part 3,
vol. 2, 68.
Ignaz Moscheles, Studies for the Pianoforte Op. 70, ed. Saunders, 23.
Henry Coward, “The Training of a Chorus: Some Practical Hints,” The Musical Times vol. 41
(1900): 450.
Te mpo Modif i cati on 293

Figure 5.33 Mendelssohn Rondo Capriccioso Op. 14, bars 97 to 105, ed. Goetschius.

Whether Coward’s attitude was something new or unusual is difficult to assess.

Writers may use almost exactly the same words but imagine something com-
pletely different. Certainly, earlier in the nineteenth century, Czerny had advo-
cated tempo modification in piano playing for the enhancement of certain
emotional states. Opining that an experienced player would immediately intuit
where these “may be conveniently expressed,” he stated that,

such general emotions or feelings may be: gentle persuasion; a slight

degree of doubt, or wavering hesitation; tender complaining; tranquil
assent; transition from a state of excitement to a more tranquil one;
refusal on reflection; sighing and grief; whispering a secret; taking leave,
and innumerable other sentiments of this sort. . . . [I]n such cases, a
slight holding back in the time (calando, smorzando, &c.) may generally be
introduced to advantage, since it would be contrary to good sense to
employ in such cases any acceleration or hurrying onward in the speed of
the movement.
Other passages, on the contrary indicate: sudden cheerfulness; hasty
or curious interrogations; impatience; incipient anger; fixed and power-
ful resolution; unwilling reproach; pride and ill temper; timid flight, tran-
sition from a state of tranquility to one of excitement, &c. . . . In such
passages the hurrying onwards and acceleration of the time is natural
(accelerando, stringendo, &c.), and in its proper place.96

Earlier, at the end of the eighteenth century, Türk recommended tempo modifica-
tion to produce apparently similar effects “when one is playing alone or with a
very attentive accompanist.” In certain passages in sonatas, concertos, and so on,

Czerny, Pianoforte School, vol. 3, 31–32.
294 off the record

“the more important notes must . . . be played slower and louder, and the less
important notes more quickly and softer,” emulating a good singer or orator. He
admits the difficulty of specifying all the places where such modifications might
take place, but goes on to cite a few:

In compositions whose character is vehemence, anger, rage, fury, and the

like, the most forceful passages can be played with a somewhat hastened
(accelerando) motion. Also, certain thoughts which are repeated in a more
intensified manner (generally higher) require that the speed be increased
to some extent. Sometimes, when gentle feelings are interrupted by a
lively passage, the latter can be played somewhat more rapidly. A hasten-
ing of the tempo may also take place in a passage where a vehement effect
is unexpectedly to be aroused.
For extraordinarily tender, longing, or melancholy passages, in which
the emotion, as it were, is concentrated in one point, the effect can be
very much intensified by an increasing hesitation (Anhalten, tardando).
The tempo is also taken gradually slower for tones before certain ferma-
tas as if their powers were gradually being exhausted. The passages
towards the end of a composition (or part of a composition) which are
marked diminuendo, diluendo, smorzando, and the like, can also be
played in a somewhat more lingering manner.97

Although illuminating, Türk’s, Czerny’s, and Coward’s descriptions do not indi-

cate the extent to which tempo was modified in a way that makes their meaning
absolutely clear today.
For Mathis Lussy, who produced one of the most detailed pedagogical texts of
the late nineteenth century, tempo modification was certainly an indispensable
aspect of musical expression. Writing in 1874, he opined that the forces acting
within a composition “all combine to develop a greater stimulus, a crescendo of
sound and accelerando of tempo, which again is naturally followed by a gradual
decrease of sound and slackening of pace.”98 Lussy describes two diametrically
opposed schools of playing. He abhors the first school, which “demands a uniform
rate of time, without accelerando or ritardando” and regards very highly regular-
ity and mechanical precision. He favors the opposing school, which accelerates
and decelerates with every rhythm and every change and does “not feel anything
objectionable in the consequent irregularity.”99 At least in theory, the debate about
the use of tempo modification continued on.

Türk, School of Clavier Playing, 360.
Lussy, Musical Expression, 9–10.
Lussy, Musical Expression, 163.
Te mpo Modif i cati on 295

By the end of the nineteenth century, a more restrained use of tempo modifica-
tion was being promoted in some circles. In 1897, Taylor explained that its
employment was not restricted to the places indicated by the composer. Both
accelerando and decelerando aid expression and “much may be gained by their
judicious introduction.” But Taylor insists that this must not lead to eccentric or
capricious effects but should “attract the attention of the listener to any particular
phrase by emphasising its legitimate expression.”100 Further to this, any tempo
modification, agogic lengthening or added pause in silence “must not be long
enough to seriously disturb the rhythm.”101 Taylor’s advice may well represent a
reaction against much more exaggerated modifications, though of course it is
hard to gauge what he really meant.
Clearly, there was something of a schism between apparent conservatism and
artistic freedom concerning the use of tempo modification in the late nineteenth
century. But there is little doubt that it was generally established as an intrinsic
part of musical performance.

Detailed Codifications
During the second half of the nineteenth century, certain pedagogues produced
highly detailed descriptions and notational symbols in an attempt to codify,
among other things, tempo modification. In Der Ausdruck in der Musik (1878), Hugo
Riemann suggests that each bar or phrase has one dynamic high point toward
which there is a crescendo and after which there is a diminuendo. This dynamic
high point may occur either at the beginning or first note (“on-emphasized”),
a middle note (“in-emphasized”), or the final note (“down-emphasized”).102 The
increase of dynamic toward the high point, whichever it may be, must be accom-
panied by an increase of liveliness (accelerando) and subsequent decrease in
dynamic liveliness (ritardando). He justifies such a theory, explaining that

ascending pitches, crescendo and stringendo are increasing positive

forms of development; descending pitches, diminuendo and ritardando
are diminutions, negative forms of development: so it is thoroughly
natural, that the first three as well as the last three (named) are needed

Taylor, Technique and Expression, 71.
Taylor, Technique and Expression, 72.
The terms “anbetonte,” “inbetonte,” and “abbetonte,” for which there were no previous equiva-
lents, were coined by Riemann in Der Ausdruck, 51.
296 off the record

for the expression of the same motion of the soul, for the same feeling,
for the reinforced intensity of their interplay.103

Riemann also describes some of the important places where tempo modification
might be applied:

A slight increasing of the tempo is appropriate for the first development

of a musical theme in the same way as the increasing of the pitch and the
crescendo is characteristic of it; in reverse a slight slowing down is appro-
priate for the final melody (phrase, tune), for the dying out of the theme,
whilst (the slowing down) is often so considerable that it can no longer
be ignored in the musical text, but requires a real ritardando [marked] in
the notation; but equally as well for the end of the theme, is the charac-
teristic falling of the pitch and the decrescendo. A final stretto-like ritar-
dando at the beginning of a tone-piece is a more rare exception and of
course has expressly to be indicated.104

The other form of tempo nuance discussed by Riemann applies to localized

moments in the phrase that require lengthening. Here, “the dynamic main note of
the phrase as well as certain harmonically important tones (suspension tones)
have to be slightly lengthened.” According to him, this is a time-honored means of
making a performance impressive. As noted earlier, Türk mentioned similar prac-
tices in the late eighteenth century. Riemann also advises the use of localized
lengthening to delineate between phrases, but makes it clear that this is some-
thing less understood and less employed. He feels that the end of each motive
“demands a very small addition to the time value.” This might take the form of a

Riemann, Der Ausdruck, 47: “Wir können gleich als drittes noch die agogischen Schattirungen,
d.h. die kleinen vorübergehenden Veränderungen des Tempos, das Stringendo und Ritardando hin-
zunehmen, welche sich der Wahrnehmung direct als vermehrte und verminderte Lebendigkeit darstel-
len. Steigende Tonhöhe, Crescendo und Stringendo sind Steigerungen, positive Entwickelungsformen;
fallende Tonhöhe, Diminuendo und Ritardando sind Verminderungen, negative Entwickelungsformen:
es ist daher durchaus natürlich, dass die drei ersteren wie die drei letzteren zum Ausdruck derselben
Seelenbewegung, derselben Empfindung gebraucht werden und deren Intensität verstärkend zusam-
Riemann, Der Ausdruck, 47: “Ein geringes Antreiben des Tempos eignet der ersten Entwickelung
eines musikalischen Themas ebenso, wie ihr das Steigen der Tonhöhe und das Crescendo eigenthümlich
ist; umgekehrt eignet der Wendung zum Schluss, dem Ausleben des Themas ein geringes Verlangsamen,
das oft genug so bedeutend ist, dass es die Notenschrift nicht mehr ignoriert, sondern als wirkliches
Ritardando fordert; gleichermassen ist aber auch das Fallen der Tonhöhe und das Decrescendo für den
Schluss charakteristisch. Eine abschliessende Stretta oder ein Ritardando zu Anfang eines Tonstücks
sind seltenere Ausnahmefälle und natürlich stets ausdrücklich zu verlangen.”
Te mpo Modif i cati on 297

“small lengthening of the final note or an added short rest.” In his opinion, the
effect of too precise an entry of the new motive “blurs the drawing.”105
Later, Riemann notes that all suspensions demand “a considerable lengthen-
ing.” To play these strictly in their notated length is “an offence against the
demands of expressive playing” and is to be reprimanded especially where terms
such as cantabile or con espressione are indicated. Referring to an example
(Fig. 5.34 ), he adds that because the suspension is not the final note of the
motive, there will in such cases sometimes be “two lengthened notes in
Evidently such advice had strong historical precedents. For example, in the
eighteenth century, C. P. E. Bach recommended that in both fast and slow move-
ments, and particularly in playing affetuoso playing, “certain notes and rests
should be extended beyond their written length, for affective reasons.” Regarding
his appended examples (Fig. 5.35 ), he explains that sometimes these exten-
sions are notated as “broadened values.” At other times, they are marked by a
“small cross.”107
Later in 1888, Riemann discusses agogic lengthening of special melody notes
(suspensions and so on) where accompaniments are of short values and figurate
in texture. He explains, giving an example (Fig. 5.36), that it is in fact the semi-
quaver e in the left hand that must be played with “a gentle lingering” if the treble
d of the right hand is to be given warmth. As to how much to linger Riemann
states that this “cannot be determined by an invariable rule; one may say that
every lengthening that strikes us as such is too great; it should make itself felt
only as living expression.”108
The face-value implication here is that any noticeable prolongation is an
exaggeration. Given the recorded evidence presented earlier, however, it is likely
that what Riemann considered barely noticeable might strike a modern listener
as extreme. As we have seen, in the playing of some pianists—for example,

Riemann in Der Ausdruck, 53: “weniger bekannt ist dagegen wohl, dass das Ende jedes Taktmotivs
eine ganz geringe Zugabe zum Zeitwerth des Motivs verträgt und oft genug gebieterisch fordert. Diese
Zugabe ist entweder eine geringe Verlängerung der Schlussnote oder eine eingeschaltete kurze Pause.
Der allzupräcise Einsass der Anfangsnote des neuen Motivs verwischt die scharfe Zeichnung; das
gilt besonders bei sehr langsamen Tempo, wo die mögliche Zugabe schon ein allenfalls definirbarer
Zeitwerth ist, während bei schneller Bewegung die Gliederung in der Regel durch die Bässe, überhaupt
die Begleitstimmen deutlich genug gemacht werden wird.”
Riemann in Der Ausdruck, 55: “Eine solche Note beansprucht stets eine erhebliche Verlängerung;
ein streng taktmässiges Hinweggehen über dieselbe erscheint als Verstoss gegen die Forderungen
eines ausdrucksvollen Spiels und ist wenigstens da immer zu rügen, wo cantabile oder con espressione
vorgeschrieben ist. Da die Vorhaltsnote nicht selbst die Schlussnote des Motivs sein kann, so erhalten
wir in solchen Fällen gelegentlich zwei verlängerte Noten nach einander:”
Bach, Versuch, part 1, 160–62.
Riemann, Catechism, 64–65.
298 off the record

Figure 5.36 Riemann, annotations for the lengthening of special melody notes.

Pugno and Paderewski—localized lengthenings sound fairly exaggerated by

present-day standards.
In an attempt to be as scientific as possible, Riemann quantified the proportion
of lengthening of single notes. Referring to the musical example (Fig. 5.37), he
provided the following mathematical description:109

Here f is perhaps longer than g as the ratio 3:2, and the rest [demisemi-
quaver rest] is likewise 3:2 as long as e, yet e itself is not as short as g (the
whole passage is ritardando). This highly important means of explaining
the rhythmical nature of a motive (precisely through the “agogic” nuances)
was formerly much too little considered.

Such proportions provide a sense of the type of hierarchy that in principle may
have governed the use of agogic lengthening. In spite of this, however, it is impos-
sible to know the intended aural effect. Riemann concludes that the “healthy
foundation of expression” lies in the employment of rallentando and accelerando,
but adds that in the past such modifications were left to instinct: “theory had no
rules for their use.”110
Although Riemann provided an unusual level of detail, his explanations could
not indicate clearly degrees of temporal change: the reader is no closer to under-
standing how much or how little was considered tasteful. It appears, too, that
Riemann is unjustified in saying that no rules about tempo modification had

Figure 5.37 Riemann, annotation for lengthened notes.

Riemann, Catechism, 65–66.
Riemann, Catechism, 65–66.
Te mpo Modif i cati on 299

previously been formulated. Earlier in the nineteenth century, Czerny had sum-
marized some of the presumably innumerable instances requiring tempo modifi-
cation, particularly retardation. Such modifications were intended to delineate
between musical structures and enhance the expression of melodic material:

The Ritardando, according to the generally established Rule, is much

more frequently employed than the Accelerando, because the former is
less likely to disfigure the character of the piece, than too frequent hur-
rying on in the speed of the movement. We may retard the time most

a. In those passages which contain the return to the principal subject.

b. In those passages, which lead to some separate member of a melody.
c. In those long and sustained notes which are to be struck with
particular emphasis, and after which quicker notes are to follow.
d. At the transition into another species of time, or into another
movement, different in speed from that which preceded it.
e. Immediately after a pause.
f. At the Diminuendo of a preceding very lively passage; as also in
brilliant passages, when there suddenly occurs a trait of melody to
be played piano and with much delicacy.
g. In embellishments, consisting of very many quick notes, which we
are unable to force into the degree of movement first chosen.
h. Occasionally also, in the chief crescendo of a strongly marked
sentence, leading to an important passage or to the close.
i. In very humorous, capricious, and fantastic passages, in order to
heighten the character so much the more.
j. Lastly, almost always where the Composer has indicated an
espressivo; as also diminuendo.

NB, It is of course understood, that here, under the term Ritardando, we

mean to comprehend all other equivalent expressions, which indicate a
more or less marked slackening of the original degree of movement,
as for Example: rallent, ritenuto, smorzando, calando, &.c; as they are
only distinguished from each other by the more or less degree of

Lussy acknowledged Czerny’s contribution to the codifi cation of tempo modifica-

tion practices. But he remarked that he could find no other text that “made a
single practical observation on the emotional element, nuances, and changes

Czerny, Theoretical and Practical, vol. 3, 33–34.
300 off the record

of time.” Although this is not entirely true, it is somewhat surprising that these
were not more substantially codified before the era of Lussy and Riemann. For
Lussy, the emotional element embraces the irregularities of tempo and included
accelerandos caused by passion and excitement, and ritardandos caused by
exhaustion and fatigue after passion, as well as “the presence of a sudden and
unexpected obstacle in a complicated structure.”112
Lussy states that in quick movements such as Prestos, Allegros, Galops, and
Valses, it is natural to keep a uniform rate of tempo; slowing down occurs only
occasionally. In slow expressive pieces, such as Nocturnes, Rondos, Rêveries,
Andantes, Adagios, and Romances, it is natural to modify the tempo. Accelerandos
and rallentandos are made according to every change of feeling and particularly
when the “expressive structure of the phrases, or their motion up or down,”
requires them. Lussy felt strongly that the distinguishing characteristics and the
poetry of a work would be destroyed if such pieces—full of rhythmical, harmonic,
and expressive changes—were played in a uniform tempo.113
Presenting various analogies, Lussy concludes that an increase in tempo should
occur under the following conditions:

1. Where several expressive notes follow one another consecutively, or where a

single note of exceptional length occurs at the beginning or in the middle of a
2. Where several notes, or groups of similar notes, occur exceptionally after an
ascending or descending progression
3. In exceptional passages, which introduce agitation or passion into the middle
of an Andante or Adagio

On the other hand, a decrease in tempo should occur under these conditions:

1. Where several consecutive expressive notes appear suddenly at the beginning

of a rhythm without there being the proper time to give them the necessary
2. Where the force expended on ascending or descending series or progressions
of notes produces fatigue and exhaustion
3. In exceptional passages, as in the middle of an Allegro, where a more compli-
cated or expressive structure occurs with a change to calmness, gravity, or
4. On expressive notes or passages, reiterated notes, and higher auxiliary notes
at the end of a phrase

Lussy, Musical Expression, 164.
Lussy, Musical Expression, 164–66.
Te mpo Modif i cati on 301

Lussy warns that before attempting the application of the foregoing principles, it
is important to consider that the tempo changes which expressive notes inspire
depend on the following:

1. Their position. Thus when a note is by exception repeated several times

consecutively, it is necessary to make an accelerando at the beginning of the
rhythm and a rallentendo at the end. (The higher auxiliary note requires a
rallentando both at the beginning and end of the rhythm.)
2. The general structure of the piece. It has been said that passages with
exceptional ascending or descending sequences require accelerandos and
rallentandos; but if the general structure of the piece is of an ascending or
descending nature, there must be no rallentando or accelerando. Nevertheless,
in a piece of an ascending structure, if the first rhythm has a more animated
accompaniment at the repeat than it had the first time, it must be accelerated.
3. The number of voices or instruments employed in the composition. A solo
player can allow himself modifications in the normal tempo that are not
permissible for an orchestra. In the orchestra, every performer must efface
himself for the sake of the general result and must sacrifice any emotional
element that may exist in his particular part.
4. The sense of words (in vocal music). Words expressing sadness or melancholy
must be sung more slowly than those expressing joy, happiness, or triumph.114

Lussy elaborates on these precepts with further details and musical illustrations
from popular works. He advises that accelerandos are indispensable in the follow-
ing instances:

• On long notes occurring exceptionally

• On a note repeated several times at the beginning or middle of a phrase
• On long or repeated notes when they are presented as syncopations or
when the accompaniment is in ascending or descending motion, but not if the
accompaniment is stationary
• On a repeated higher auxiliary note at the beginning of a phrase
• On modulations at the beginning or end of a phrase
• On descending figures of short notes particularly when followed by a longer
and/or higher note; in straightforward codettas in short notes
• For notes in ascending motion
• On ascending or descending sequences at the beginning of a phrase, using a
pivotal stationary note where melody and bass move in contrary motion at the
beginning of a phrase

Lussy, Musical Expression, 166–67.
302 off the record

• Where small rhythmical groups are repeated in ascending or descending

• For repeated small groups of notes at the end of fast movements where the
bass is stationary
• In ascending or descending runs at the end of phrases in fast tempos
• In playful passages of short or uniform groups of notes in the middle of
expressive phrases
• In phrases accompanied by chords struck together succeeding a phrase with
spread chords or where harmonies follow a regular progression
• In syncopated passages broken by rests, in passages with syncopated struc-
tures, or in phrases accompanied by ascending or descending arpeggios115

For music of slow or moderate tempo where one or two expressive notes occur
at the beginning of a phrase, Lussy advises that rallentandos are required in the
following situations:

• On a rest which follows the first note of a staccato figure, particularly if it is the
highest note followed by the next note directly below of equal value
• On the first note of a legato passage if it is by exception a high note or succeeds
passages that have begun with low notes
• On a rest following the first note of a figure, particularly if it is a repeated high
note reached by a wide ascending skip and followed by a lower note
• On the rest following the highest note of an ascending progression followed by
a lower note
• On the higher auxiliary note that begins a group of introductory notes
• On the first notes of a figure marking a distant modulation, change of charac-
ter or mood and, so on
• On a high note forming the pivot to a low one
• On a note replacing a higher or lower one, thus changing the direction of the
following figuration
• At the end of an ascending or descending progression particularly if the design
• On a succession of high notes, gradually rising, and suddenly interrupted by a
low note, or especially when the high note is preceded by a rest or by exception
is repeated
• On descending figures immediately following ascending ones
• On a group of low notes following high ones
• In expressive melodious passages in which long notes and rich harmonies
occur, especially in an Allegro composed of short notes

Lussy, Musical Expression, 168–77.
Te mpo Modif i cati on 303

• In expressive or dreamy passages introduced into the middle of fast

• In passages that, after being given in the major, are presented in the minor
• At the end of soft and expressive phrases
• On a long note preceding the final note, especially if it includes a trill
• On notes exceptionally repeated several times
• On the highest note at the end of the penultimate bar, especially if it is synco-
pated, prolonged, or chromatic
• On the higher auxiliary note at the end of the penultimate bar; on reiterated
notes at the end of the penultimate bar
• On the penultimate note if the last one is a reiterated note
• On the repetition of short figures in the penultimate bar, especially if it
contains higher auxiliary notes
• On crotchets occurring by exception in the penultimate bar; on short notes
occurring by exception and containing reiterated or higher auxiliary notes
at the end of a phrase except if the passage is a florid one replacing a long
penultimate note
• On the last notes of a phrase proceeding in consecutive descending motion
and following a high note
• On descending notes at the end of a penultimate bar
• On the last notes of a suspended rhythm
• At the end of a phrase that is, by exception, polyphonic and contrapuntal and
contains complicated harmony, resolved discords, or suspensions, and so on
• On the last notes of a codetta containing a higher auxiliary note, reiterations,
and stationary notes
• On pedal points, syncopations, and long notes at the end of an ornamental
group in a cadenza ad libitum
• On rhythmical repetitions such as groups of notes repeated several times con-
secutively at the end of expressive phrases
• On final chords separated by rests116

Clearly, Lussy endeavored to be as comprehensive as possible about the situations

requiring tempo modification. But as in so many other cases, his descriptions do
not qualify what extremes of modification caused by ritardando and accelerando
might be achieved or expected. In this respect, Michael Green’s research, in which
he compared various performances of the opening song—“Seit ich ihn gesehen”—
from Schumann’s Frauenliebe und-leben Op. 42, with precepts laid down by Lussy,
is telling. His findings corroborate Lussy’s thesis “when it comes to local detail.”
Artists appear to respond to the same surface events in the music. But the feature
that makes one performance different from another is the “degree of expressive

Lussy, Musical Expression, 177–95.
304 off the record

response to a particular event.” Green explains that “whether a performer slows

minutely or excessively to an event is of critical importance to how a performance
integrates local details with their role in shaping the larger musical structure.”117
It is this degree of response, the how much, that we cannot discover from Lussy’s
words. Thus, although Lussy’s theories correspond with the practices of many
performers heard in recordings, in the end they do not inform us of what rate
of change of speed or of what tempo extremes were appropriate or tasteful in
his time.
To what extent late-nineteenth-century codification preserves the actual fea-
tures of contemporaneous practice remains unclear. Aware of the proliferation of
such works, Christiani opined in 1885 that these provide enough information to
appreciate “the leading principles” that govern the application of accelerando and
ritardando. But he hoped that students would separate those rules “that are prac-
tical and generally fitting” from those “that are only casually fitting, or based on
personal taste.” The latter should be taken for “what they are worth.”118 Christiani
highlights a significant dilemma: namely, that not all such detailed texts preserved
an entirely realistic and objective view of current practices.

The Early-Twentieth-Century Attitude

During the first half of the twentieth century, some authors advocated a more
judicious use of tempo modification. In 1909, Hofmann held the principles of bal-
ance and compensation as most important. He explains that “what you shorten of
the time in one phrase or part of a phrase you must add at the first opportunity to
another in order that the time ‘stolen’ (rubato) in one place may be restituted in
another.” Any modification of tempo must take place in such a way that the length
of the work when “played in the strictest time” is preserved.119 One is reminded
here of Reinagle’s definition in 1855 of a compensating style of tempo rubato.
Hofmann appears to be expressing an ideal that combats excesses in tempo
modification. In having to make compensation for any change, one is perhaps less
likely to deviate too far from the prevailing tempo. Such advice may well have
been the catalyst for the mid-century style exemplified in the playing of pianists
such as Solomon, cited earlier in Tables 5.1, 5.2, and 5.9. However, if Hofmann’s
advice gives the impression that his tempo modification would have sounded bal-
anced, his playing shows that he retained some late-nineteenth-century traits. In
his 1912 recording of Schumann’s Warum? Op. 12 No. 3, for example, he makes

Michael D. Green, “Mathis Lussy’s Traité de l’expression musicale as a Window into Performance
Practice,” Music Theory Spectrum: The Journal of the Society for Music Theory vol. 16 (1994): 216.
Christiani, The Principles of Expression, 296.
Hofmann, Piano Questions Answered, 100.
Te mpo Modif i cati on 305

tempo modifications that do not sound proportioned by present standards

(Audio Ex. 5.45 ). In other examples, such as his 1903 recording of Chopin’s
Nocturne Op. 9 No. 2, there is frequent modification of tempo for expressive
effect (Audio Ex. 5.46 ). Although a sense of proportion and balance is notice-
able, this is not in a manner that might have been envisaged from his description.
There is some process of compensation evident, but the principle is not applied
strictly or pedantically.
As the twentieth century progressed, a stricter adherence to musical notation
was increasingly encouraged. In 1932, Gieseking professed that the most highly
regarded musicians of the day are very accurate in their interpretations and
“reject all that is contrary to the intentions of the composer.” He adds sternly that
“absolutely correct execution of a composition is the only foundation upon which
a really excellent interpretation can be built.” But he also advised musicians to
develop a keen sense of musical aesthetics, to “know and feel where a slight
accelerando or ritardando is permitted or relevant.”120 In truth, however, many
distinguished and renowned musicians of the era were far less particular about
accuracy or indeed adherence to the score.
Evidently, Gieseking adopted a similar approach to Riemann and promoted a
compensating style of tempo modification that would help preserve unity and
balance within a phrase:

It is a well-known fact that every phrase has its climax, to reach which a
slight hurrying of pace, or a slight increase of sound is permitted, whilst
the reverse should take place from the climax to the end of the phrase. If
these fine points, therefore, are executed in the right manner, that is to
say, in natural proportion, they will doubtlessly serve to vitalize the
phrase, will correspond to natural musical feeling, and will increase

Further to this, he warned that “care must be taken not to overdo these changes
of tempo,” adding that “inaccurate and disproportionate interpretations of cre-
scendo, diminuendo, ritardando, accelerando . . . take away from the naturalness
of the interpretation.”121 What is significant here is that the words are so similar
to those of nineteenth- and early-twentieth-century writers but their meaning is
so different.
Gieseking’s remarks equate with a stricter more text-faithful style of perfor-
mance. And his recordings confirm that he introduced tempo fluctuations within
narrower limits than the preceding generation of pianists. In his 1939 recording
of Brahms’s Intermezzo Op. 117 No. 2, for example, it is evident that tempo

Gieseking and Leimer, Pianistic Perfection, 43.
Gieseking and Leimer, Pianistic Perfection, 43–44.
306 off the record

modifications are made only where specifically notated (Audio Ex. 5.47 ). By
contrast, Freund’s 1953 recording of the work shows much more variation in
tempo and agogic lengthening, in the style she undoubtedly adopted from Brahms
himself (Audio Ex. 5.48 ). Still, in some performances, such as Brahms’s
Capriccio Op. 76 No. 1 recorded in 1956, Gieseking is less strict with tempo
(Audio Ex. 5.49 ). But in her 1950 recording of the same work, Freund is more
rhythmically capricious than Gieseking. Notably, she makes agogic lengthenings
where Brahms indicates “< >” in bars 3, 5, and other places (Audio Ex. 5.50 ).
According to Fanny Davies, this type of lengthening—outlawed by Gieseking—
emanated directly from Brahms himself. At least in this work, Gieseking’s playing
is markedly devoid of such nuances.
Other pianists associated with Brahms, for example Friedberg and Eibenschütz,
use noticeable tempo modification to enhance expression in Brahms’s Ballade
Op. 118 No. 3. Crutchfield gives a great account of this:

Eibenschütz starts the ballade at a tempo of half-note = 85 or 90 and

speeds up. Friedberg starts at about 85 and slows down. Taking an aver-
age from tapped beats shows him in the mid-60s for much of the “A”
section—though that is something of an exaggeration, since almost
every bar is lengthened by rhetorical hesitation before attacking the next
upbeat. Eibenschütz always accelerates with harmonic tension and
retards with cadences.

This, of course, represents markedly individual approaches to the same work. But
the style is of one era. Crutchfield’s eloquent conclusion really brings the point
home: “Friedberg’s performance and Eibenschütz’s are as different as night and
day, but night and day in the same city.”122
The reaction against exaggerated tempo modification is particularly evident in
the comments of the English musician Ernest Walker (1870–1949). In “Some
Questions of Tempo” (1930), he avows that “no performance worth anything, of
any music, remains mathematically level, either in time or tone, for more than a
very limited period.” However, he is highly critical of eminent soloists whose
apparently degenerate musical palates prevent them from distinguishing “what is
reasonably flexible” from “what is downright bad.” He blames tradition—presum-
ably a late-nineteenth-century tradition—for certain unauthorized tempi and
asks rhetorically what else might account for this? Such practices include

the ruinously pompous slowings-down at the end of any Handel air, and
the train-catching hurries on the ends of the “Études symphoniques”
[Schumann] or Chopin’s A flat polonaise? And how often do we hear

Crutchfield, “Brahms, by Those Who Knew Him,” 18.
Te mpo Modif i cati on 307

Chopin’s D flat Valse as he wrote it, in unchanging dance-tempo and

ending with twenty-four even grace-notes (no bar-lines in the right
hand), and no rit. at all—not the dreadful 9+9+6+molto rit. of some edi-
tions and most pianists?123

Walker’s attitude to tempo modification appears to be very different from that of

even thirty years earlier, when noticeable accelerations and decelerations were
freely used to enhance the mood and character of the music. His words and those
of many others heralded a move toward a limited modification of tempo that came
to be generally accepted during the second half of the twentieth century and that,
by and large, has taken a strong hold on performance today.
Written texts therefore confirm that tempo modification was used as an
expressive device around the turn of the twentieth century. Some of these texts,
varying in detail and scope, provide a catalogue of situations requiring modifica-
tion. In comparison with early recordings, however, it is clear that these texts
did not preserve many important features requisite for a fuller understanding of
the use of tempo modification. In addition, a modern interpretation of the descrip-
tive language used does not accord with their original meaning. It is difficult,
therefore, to assess from the texts alone what was acceptable to late-nineteenth-
century musicians. For this, the legacy of early recordings is essential.
When I was at university some twenty-five years ago, I was taught to keep
fairly strictly in time in most repertoires. Small deviations were permitted espe-
cially in Romantic music, but anything too noticeable was frowned upon or met
with stern objection. I have since come to appreciate the expressive quality and
the structural delineation that can be achieved through the use of tempo modifi-
cation. My own taste has changed. Whereas I was at one time shocked by the
sudden forward surges and the halting ritardandos of early recordings, I now find
these totally acceptable. In my own performances, I am trying to emulate what
I have heard on these recordings. A good example of this comes from a live perfor-
mance of Brahms’s Sonata in G Major Op. 78.124 My violinist colleague and I com-
menced the exposition of the first movement at a tempo that we felt reflected
Brahms’s vivace ma non troppo indication. In bar 7, we lingered (beyond what might
normally be considered comfortable) on the second dotted minim to make some-
thing special of Brahms’s hairpin. For the sempre p e tranquillo at bar 11, we agreed
to play immediately slower to create a sense of tranquillity, recovering the tempo
where the opening theme returns at bar 21. Further on in the movement, we
markedly broadened the tempo to bring out the feeling of sostenuto at bar 48 as

Ernest Walker, Free Thought and the Musician and Other Essays (London: Oxford University
Press, 1946), 138.
Live unedited performance with violinist Robin Wilson, recorded November 19, 2010, in Recital
Hall East, Sydney Conservatorium of Music.
308 off the record

requested by Brahms. But we contrasted this dramatically by surging forward

during bars 50 and 51 to bring out the extremely passionate flavor of the music
(Audio Ex. 5.51 ). These changes of tempo certainly helped us to clarify struc-
tural elements and to make our expressive intention bold and clear.
We may never know just how much Brahms, Chopin, or Mozart really modified
tempo in their performances. What is certain though is that they did so and in
very individual ways that could not be encapsulated in their scores.

Dislocation, arpeggiation, metrical rubato, and various forms of rhythmic altera-

tion as well as tempo modification, were indispensable performing practices in
piano playing around the turn of the twentieth century. The abundance of recorded
examples examined in this book reveals that many important pianists—including
some revered virtuosi and pedagogues of the late Romantic era—made use of
these techniques. But they did so in a manner that does not accord with current
notions of tasteful interpretation. Yet historical evidence has convinced me that
the characteristics of such practices derived from long-established traditions.
Recordings also demonstrate the progressive decline of these practices during the
twentieth century.
The comparison between early recordings and contemporaneous written texts
has exposed striking contradictions time and time again. In many cases, signifi-
cant features heard on the recordings are not conveyed in the texts. Sometimes
the verbal advice of particular pianists appears to conflict with their own record-
ings. Clearly, such advice was intended only for a specific time or place. Or the
descriptive language—assuming knowledge of prevalent practices—had implica-
tions that may be irretrievably lost. To add to the complexity, many notational
symbols and musical terms appear to have indicated something quite different
from the meaning that they now convey. The recordings therefore represent not
only an audible key to understanding written descriptions and clarifying details
that would otherwise have remained hidden, but are also a unique source of
evidence in their own right.
Approached with open-mindedness, early recordings allow us to experience
and appreciate the expressive style of important late-nineteenth-century pianists.
Astonishingly, in so many of the cases presented, this style could not have been
envisaged from a face-value interpretation of their written advice. In fact, a style
of performance based on the advice alone would seldom approach the style of the
recordings. Here the gulf between theory and practice is most noticeable. We
cannot, therefore, assume that written texts convey clearly or meaningfully the
practices that in previous eras were considered essential to artistic performance.

310 Conclusion

The implications here are manifold. In conjunction with written texts, early
recordings provide vital clues to the performance styles of pianists who did not
record. It is more than likely that Schumann—who greatly appreciated Reinecke’s
understanding of his music—would whole-heartedly have approved of his style of
playing Warum? And Leschetizky’s and Saint-Säens’s rhythmic alterations in
Chopin’s music may well reflect Chopin’s own practices as described by his stu-
dents. Without the possibility of time travel, we will never really know. The fact is
that many of the devices employed by these earlier pianists, and the manner in
which they employed them, will not be clearly discernible from contemporaneous
written texts, nor indeed from their scores. And if the musical traditions of the
late-nineteenth-century pianists captured on recordings seem surprising to us
now, it is more than likely that earlier styles—particularly of hallowed pianists—
would be even more so.
Although many of the practices preserved on early piano recordings seem alien
today, it is clear that these were integral to late Romantic pianism as exemplified
in the playing of acknowledged masters of the period. To come to an understand-
ing about what musical notation meant to composers and performers of the late
nineteenth century, we must accept that the most admired musicians of the era
approached the aesthetics of performance from a very different perspective than
musicians today. Such knowledge suggests that a historically informed style of
performance for any repertoire, time, or place requires more than just playing by
the book. It requires a great deal of imagination and reading between the lines.
And it cannot be achieved simply by the adoption of appropriate instruments or
the application of only those practices that do not challenge current notions of
good taste or that do not take us out of our comfort zone. Indeed, it may be argued
that it cannot be achieved at all. However, a willingness to push the boundaries of
accepted taste, coupled with guidance from historical sources, will undoubtedly
lead to fresh, insightful, and inspired interpretations of music of the old masters.

Adam, Louis. Méthode du piano du conservatoire. Paris: Conservatoire impéral de musique, 1804/5.
Anon. “Altmeister Karl Reinecke und das Pianola.” Zeitschrift für Instrumentenbau (Sept. 11, 1904):
Anon. “Busoni: [by] One of His Pupils.” The Pianola Journal: The Journal of the Pianola Institute
no. 10 (1998): 61–63.
Anon. “Dr. Hans Von Bülow’s Recital.” The Musical Times vol. 25 (1884): 337–38.
Anon. Le Pianiste vol. 1, no. 5 (Mar. 1834): 78.
Anon. “Mdlle. Kleeburg’s Second Recital.” The Musical Times vol. 26 (1885): 402.
Anon. “The Medium of a New Musical Art: An Interview with Harold Bauer and the Duo-Art
Pianola” [Advertisement]. The New York Sun (March 26, 1916).
Anon. “On Musical Colouring.” The Harmonicon vol. 1, no. 11 (1823): 161–62.
“Player-Piano Notes.” The Musical Times vol. 70 (1929): 135, 905.
Anon. “Touch—What Rubinstein Says about It.” Magazine of Music Pictorial: Pianoforte Tutor
(1891): part 2, 172.
Babitz, Sol. “Concerning the Length of Time that Every Note Must Be Held.” The Music Review
vol. 28 (1967): 21–37.
———. “Style and the Performer.” Early Music vol. 4, no. 4 (Oct. 1976): 491–93.
Bach, Carl P. E. Versuch über die wahre Art das Clavier zu spielen. Vol. i (Berlin: 1753; rev. 2nd ed.
1787); vol. ii (Berlin: 1762; rev. 2nd ed. 1797); repr. of 1st ed., incl. revs. of 1787 as a separate
section (Leipzig , Germany: 1787); trans. and ed. William J. Mitchell as Essay on the True Art of
Playing Keyboard Instruments. New York: Norton, 1949.
Bach, Johann S. French Suites. Arr. Feruccio Busoni. Ed. Egon Petri. London: British & Continental
Music Agencies, 1941.
———. “Two-Part Inventions BWV 772–786.” Klavierwerke Busoni-Ausgabe iv Zweistimmige
Inventionen (BWV 772–786), ed. Feruccio Busoni, trans. L. Elson. Leipzig, Germany: Breitkopf
& Härtel, c. 1930–50.
Badura-Skoda, Paul. “Kommentar.” Carl Czerny: Uber den richtigen Vortrag der sämtlichen
Beethoven’schen Klavierwerke. 1963. Trans. as “Commentary.” On the Proper Performance of all
Beethoven’s Works for the Piano. Vienna: Universal, 1970.
Baillot, Pierre. L’Art du violon: nouvelle méthode (Paris: 1834). Facs. reprint, Méthodes et Traités:
Violon Vol. 3. Courlay, France: Fuzeau, 2001.
Barth, George. The Pianist as Orator. Ithaca, N.Y.: Cornell University Press, 1992.
Batten, Joe. Joe Batten’s Book: The Story of Sound Recording. London: Rockliff, 1956.
Bauer, Harold. Harold Bauer: His Book. New York: Norton, 1948.
Beaumont, A . “Sleeve Notes.” Ferruccio Busoni: Complete Recordings, GEMM CD 9347 (1989).
Beethoven, Ludwig van. “Adagio from ‘Sonata Pathétique.’” Magazine of Music: Pictorial Pianoforte
Tutor (1891): part 2, 130.

312 Bibli og raphy

———. “From the Moonlight Sonata.” Magazine of Music: Pictorial Pianoforte Tutor (1891):
part 2, 129.
———. “Marcia Funebre sulla morte d’un eroe.” Magazine of Music: Pictorial Pianoforte Tutor (1891):
part 2, 131.
———. Moonlight Sonata Op. 27 No. 1, ed. Cipriani Potter. London: Mills, 1854.
———. Sonata Pathétique for the Piano Forte, ed. Cipriani Potter. London: Mills, 1854.
———. Sonatas for the Pianoforte, ed. Charles Hallé. London: Chappell, c. 1880.
———. The Sonatas of Beethoven for the Pianoforte, ed. William Dorrell. 2 vols. London: Boosey,
———. Sonaten für das Pianoforte Solo, ed. Franz Liszt. Wolfenbüttel, Germany: Halle, c. 1845.
———. Sonaten für Pianoforte, ed. Carl Reinecke. 2 vols. Leipzig, Germany: Breitkopf & Härtel,
———. Sonaten und andere Werke für das Pianoforte. 5 vols. Stuttgart, Germany: Cotta, 1871.
Reprinted with German and English text. Trans. as Sonatas and Other Works for the Pianoforte
. . . With the Cooperation of Hans von Bülow, Immanuel von Faisst, Ignaz Lachner, Franz von Liszt,
founded by Sigmund Lebert. English trans. of Instructive Text by J. H. Cornell. 5 vols. Stuttgart,
Germany: Cotta, 1891; Vol. 2: 10 Sonatas (Opp. 22–49). With the cooperation of Immanuel
von Faisst elaborated by Sigmund Lebert; Vol. 3: Variations, Rondos, Bagatelles, Andante
in F major (Opp. 33–51 and without number of opus). With the cooperation of Immanuel
von Faisst, elaborated by Sigmund Lebert; Vol. 4: Sonatas (Opp. 53–90; Variations Op. 76;
Fantasie Op. 77; Sonatina Op. 79; Polonaise Op. 89) with explanatory remarks for teachers
and pupils by Hans von Bülow; Vol. 5: 5 Sonatas (Opp. 101–11; Bagatelles Op. 119; Diabelli
Variations Op. 120; Bagatelles Op. 126; Rondo a Capriccio Op. 129) with explanatory remarks
for teachers and pupils by Hans von Bülow.
Benko, Gregor. “More on Piano Rolls.” International Classical Record Collector vol. 2, no. 7 (1996):
Berger, Jonathan, and Nicols, Charles. “Brahms at the Piano: An Analysis of Data from the Brahms
Cylinder.” Leonardo Music Journal vol. 4 (1994): 23–30.
Bériot, Charles de. Méthode de violon. 3 parts (Paris: 1858). Facs. reprint Méthodes et Traités: Violon,
Vol. 6. Courlay, France: Fuzeau, 2001.
Bernard, Elizabeth, and Timbrell, Charles. “Louis Diémer.” The New Grove Dictionary of Music and
Musicians, 2nd ed., ed. S. Sadie. London: Macmillan, 2001, vol. 7, 328.
Bertini, Henri. Twenty Five Preparatory Studies Op. 32 for the Pianoforte, ed. Gordon Saunders.
London: Hammond, 1902.
Bird, John. “Grainger’s Anecdotes.” Percy Grainger. Oxford: Oxford University Press, 1999.
Blume, F., ed. Die Musik in Geschichte und Gegenwart—Allgemeine Enzyklopädie der Musik. Kassel,
Germany: Bärenreiter, 1963, Band I.
Borwick, John. “Electrical Recording 70 Years Old This Year.” ICRC (Nov. 1995): 83–90.
———. “Sound Recording, Transmission and Reproduction.” The New Grove Dictionary of Music and
Musicians, ed. S. Sadie. London: Macmillan, 1980, vol. 17, 567–90.
Bose, Fritz von Bose. “Carl Reinecke: An Appreciation.” The Musical Times vol. 51 (1910): 302.
Bourligueux, Guy. “Raoul Pugno.” The New Grove Dictionary of Music and Musicians, 2nd ed.,
ed. S. Sadie. London: Macmillan, 2001, vol. 20, 592–93.
Brahms, Johannes. Hungarian Dance No. 1. Hungarian Dances, Book 1. New York: Schirmer,
———. Ungarische Tänze No. 1. Klavier-Werke. Leipzig, Germany: Breitkopf & Härtel, c. 1890.
Brée, Malwine. Die Grundlage der Methode Leschetizky. Trans. Dr. Theodor H. Baker as The Groundwork
of the Leschetizky Method. New York: Schirmer, 1902.
Breithaupt, Rudolf M. Die natürliche Klaviertechnik, Band 1 (Handbuch der modernen Methodik und
Spielpraxis für Künstler und Berufsspieler, Lehrer und Lehrerinnen, Konservatorien und Institute,
Seminare und Schulen). Leipzig, Germany, C. F. Kahnt Nachfolger: 1905. 2nd ed., Leipzig,
Germany: Kahnt, 1912.
Brossard, Sébastien de. Dictionnaire de la Musique. Paris: 1701. 3rd ed., Amsterdam: c. 1708. Reprint,
Geneva: Minkoff, 1992.
Bibli og raphy 313

Brown, Clive. Classical and Romantic Performing Practice 1750–1900. Oxford: Oxford University
Press, 1999.
Bülow, Hans von, ed. See Ludwig van Beethoven.
Burden, Michael. Performing the Music of Henry Purcell. Oxford: Clarendon, 1996.
Busoni, Ferrucio. Letters to His Wife. Trans. Rosamund Ley. London: Arnold, 1938.
———. “Vortrag” to Johann S. Bach, Das wohltemperierte Klavier, erster Teil, bearbeitet und
erläutert, mit daran anknüpfenden Beispielen und Anweisungen für das Studium der mod-
ernen Klavierspieltechnik von Ferrucio Busoni, vol. 1 (Leipzig , Germany: c. 1894). Trans. as
“Interpretation” in “Corrections and Augmentations to Books I and II” of The Well Tempered
Clavichord Revised, Annotated, and Provided with Parallel Examples and Suggestions for the Study
of Modern Pianoforte-Technique by Ferrucio B. Busoni. New York: Schirmer, 1899.
———, ed. See Johann S. Bach.
Butt, John. Playing with History. Cambridge: Cambridge University Press, 2002.
Catalogue of Music Rolls for the Duo-Art Reproducing Piano. New York: Aeolian Company, 1924.
Cauchie, Maurice. “Respect for Rhythm.” The Musical Times vol. 70 (1929): 890–92.
Chan, Mary, and Kassler, Jamie C. Roger North’s “Of Sounds” and “Prendcourt Tracts.” Digests
and Editions with an Analytical Index by Janet D. Hine. Kensington, Australia: University of
New South Wales, 2000.
Chasins, Abram. Speaking of Pianists. New York: Knopf, 1957. 3rd ed., New York: Da Capo, 1981.
Chopin, Frédéric. “Nocturne Op. 27 No. 2.” Repertoire Leschetizky: 14 ausgewählte Stücke für
Pianoforte, ed. Theodor Leschetizky. Leipzig , Germany: Rahter, 1880.
———. Oeuvres complètes: revues, doigtées et soigneusement corrigées d’apres les éditions de Paris,
Londres, Bruxelles et Leipsic [sic], ed. Karl Klindworth. Berlin: Bote & Bock, c. 1890.
———. Pianoforte-Werke, ed. Carl Mikuli. 14 vols. Leipzig, Germany: Kistner, 1879.
Christiani, Adolphe. The Principles of Expression in Pianoforte Playing. New York: Harper, 1885.
Cinti-Damoreau, Laure. Méthode de chant, composée pour ses classes du conservatoire. Paris: 1849.
Clark, Ronald W. Edison: The Man Who Made the Future. London: Macdonald and Jane’s, 1977.
Clérambault, Louis-Nicolas. Premier livre de pièces de clavecin. Paris: 1704. Reprint, Geneva: Minkoff,
Cone, John F. Adelina Patti: Queen of Hearts. Portland, Ore.: Amadeus, 1993.
Copeland, Peter. Sound Recording. London: British Library Board, 1991.
Corri, Domenico. A Select Collection of the Most Admired Songs, Duetts &c. 3 vols. Edinburgh:
c. 1779.
———. The Singer’s Preceptor, or Corri’s Treatise on Vocal Music. 2 vols. London: London: 1810.
Corri, Philip. A. L’anima di musica. London: 1810.
———. Original System of Preluding. London: Chappel, 1813.
Couperin, François. L’Art de toucher le clavecin. Paris: 1717. Reprint, Geneva: Minkoff, 1986.
———. Premier livre de pièces de clavecin. Paris: 1713. Reprint, Courlay, France: Fuzeau, 1988.
Couperin, Louis. “Prelude in C Major” Bayn Manuscript (c. 1658), ed. and typeset Steve Wiberg. Due
West Editions, 2009.
Coward, Henry. “The Training of a Chorus: Some Practical Hints.” The Musical Times vol. 41 (1900):
Cramer, Johann B. Instructions for the Pianoforte. London: Chappell, 1812.
———. 50 Selected Studies by J.B. Cramer, Systematically Arranged and the Fingering and Marks of
Execution Critically Revised with Instructive Notes by Dr. Hans von Bülow. Trans. L. Rothfeld.
Munich: Jos. Aibl., 1880.
———. Sixty Selected Pianoforte-Studies, with Preface, Fingering, Marks of Execution and Explanatory
Notes by Hans von Bülow, 3rd ed. Trans. Constance Bache. Munich: Aibl, 1889.
Crelle, August L . Einiges über musikalischen Ausdruck und Vortrag. Berlin: 1823.
Crutchfield, Will. “Brahms, by Those Who Knew Him.” Opus vol. 2, no. 5 (1986): 12–21 and 60.
———. “Fashion, Conviction, and Performance Style.” Authenticity and Early Music: A Symposium,
ed. Nicholas Kenyon. Oxford: Oxford University Press, 1988, 19–26.
Czerny, Carl. The Art of Playing the Ancient and Modern Piano Forte Works . . . Being a Supplement to
the Royal Pianoforte School Op. 500. London: Cocks, 1846.
314 Bibli og raphy

———. Vollständige theoretische-practische Pianoforte-Schule Op. 500. 3 vols. Vienna: 1839. Trans. as
Theoretical and Practical Pianoforte School Op. 500. London: Cocks, [1839].
Dahl, Per. “Contemporary Evaluations of Grieg as a Pianist.” Edvard Grieg: The Piano Music in Historic
Interpretations. Trans. William H. Halverson. Simax PSC 1809 (1992).
Damm, Gustav. Klavierschule und Melodienschtaz für die Jugend (Leipzig , Germany: 1868), 21st ed.,
rev. with English trans. 2 vols. Leipzig, Germany: Steingräber, 1879.
Daquin, Louis-Claude. “Avertissement.” Premier livre de pièces de clavecin. Paris: 1735. Reprint,
Geneva: Minkoff, 1982.
David, Ernest, and Lussy, Mathis. Histoire de la notation musicale depuis ses origins. Paris: 1882.
Davies, Fanny. “Some Personal Recollections of Brahms as Pianist and Interpreter.” Cobbett’s
Cyclopedic Survey of Chamber Music, 2nd ed., compiled and ed. Walter W. Cobbett, with supple-
mentary material ed. Colin Mason. 2 vols. London: Oxford University Press, 1963, vol. 1,
Day, Timothy. A Century of Recorded Music: Listening to Musical History. New Haven, Conn.: Yale
University Press, 2000.
Dearling , Robert, Dearling , Celia, and Rust, Brian. The Guinness Book of Recorded Sound. London:
Guinness, 1984.
Dolmetsch, Arnold. The Interpretation of the Music of the Seventeenth and Eighteenth Centuries. 1915.
2nd ed., London: Novello, 1946.
Donington, Robert. The Interpretation of Early Music. New rev. ed. New York: W. W. Norton, 1992.
———. “The Present Position of Authenticity.” Performance Practice Review vol. 2, no. 2 (1989):
———. “Rubato.” The New Grove Dictionary of Music and Musicians, ed. Stanley Sadie. London:
Macmillan, 1980, vol. 16, 292.
Duphly, Jacques. Troisième livre de pièces de clavecin. Paris: 1756. Reprint, Geneva: Minkoff, 1987.
Eargle, John. The Microphone Book. Woburn, Mass.: Focal Press, 2001. 2nd ed., Oxford, U.K .: Focal
Press, 2004.
———. Sound Recording. New York: Van Nostrand Rheingold, 1976.
Ehrenfechter, C. A . Technical Study in the Art of Pianoforte-Playing (Deppe’s Principles). c. 1889.
3rd ed., London: Reeves, 1891.
Eigeldinger, Jean-Jacques. Chopin vu par ses élèves. Neuchâtel, France, Baconnierre: 1970. Trans.
Naomi Shohet, Kyrisa Osostowicz, and Roy Howat as Chopin: Pianist and Teacher—As Seen by
His Pupils, 3rd ed., ed. Roy Howat. Cambridge: Cambridge University Press, 1986.
Encyclopedia Britannica. Available:
Evans, A . “Moriz Rosenthal.” The New Grove Dictionary of Music and Musicians, 2nd ed., ed. S. Sadie.
London: Macmillan, 2001, vol. 21, 703.
Farrenc, Aristide, and Farrenc, Louise, eds. Le Trésor des pianists. 23 vols. Paris: 1861–72. Reprint,
New York: Da Capo, 1977.
Faure, Jean-Baptiste. La Voix et le chant: traité pratique. Paris: 1886.
Field, John. “Nocturne in B Flat.” Stücke aus dem Repertoire Essipoff-Leschetizky, ed. Theodor
Leschetizky. Hamburg and Leipzig, Germany: Rahter, 1883.
Finck, Henry T. Grieg and His Music. New York: Lane, 1910.
Flesch, Carl. The Memoirs of Carl Flesch. Eds. Hans Keller and Carl Flesch. Trans. Hans Keller.
London: Rockliff, 1957.
Ford, Peter. “History of Sound Recording.” Recorded Sound: The Journal of the British Institute of
Recorded Sound vol. 1, no. 7 (1962): 221–29.
Forqueray, Jean-Baptiste. Pièces de viole mises en pièces de clavecin. Paris: 1747. Reprint, Geneva:
Minkoff, 1987.
Foucquet, Pierre-Claude. Premiere et deuxième livre de pièces de clavecin. Paris: 1750–51. Reprint,
Geneva: Minkoff, 1982.
Frescobaldi, Girolamo. Il primo libro di capricci fatti sopra diversi soggetti et arie in partitura. Rome:
———. “Al Lettore.” Toccate e partite d’intavolatura di cimbalo et organo . . . libro primo. Rome: 1615.
Reprint, Florence: SPES, 1980.
Bibli og raphy 315

Gaines, James R . Evenings in the Palace of Reason: Bach Meets Fredrick the Great in the Age of
Enlightenment. 2005. Reprint, New York: HarperCollins, 2006.
Gaisberg , Frederick W. The Music Goes Round. New York: Macmillan, 1942.
García, M. García’s New Treatise on the Art of Singing: A Compendious Method of Instruction. London:
Beale & Chappell, 1857.
———. Traité complet de l’art du chant. Paris: 1847.
Gasparini, Francesco. Larmonico Pratico al Cimbalo (Venice: 1708). Trans. as The Practical Harmonist
at the Harpsichord by Frank S. Stillings, ed. David L. Burrows. New York: Da Capo Press, 1980.
Gatty, Reginald. “Tempo Rubato.” The Musical Times vol. 53, no. 829 (Mar. 1912): 160–62.
Gaultier, Denis, and Gaultier, Ennemond. Livre de tablature des pieces de luth. c. 1672. Reprint,
Geneva: Minkoff, 1975.
Gelatt, Roland. The Fabulous Phonograph 1877–1977, 2nd rev. ed. London: Cassell, 1977.
Gerhard, Anselm. “Willkürliches Arpeggieren: Ein Selbstverständliches Ausdrucksmittel in der
Klassische-Romantischen Klaviermusik und seine Tabuisierung im 20. Jahrhundert,” Basler
Jahrbuch für Historische Musikpraxis vol. 27 (2003): 123–34.
Gerig , Reginald. Famous Pianists and Their Technique. Washington, D.C.: Luce, 1974.
Germer, Heinrich. Wie spielt man klavier? Op. 30, Fünf didaktische Abhandlungen über Tonbildung,
Accentuation, Dynamik, Tempo und Vortrag. Leipzig, Germany: Leede, 1881.
Gieseking , Walter, and Leimer, Carl. Modernes Klavierspiel nach Leimer-Gieseking (1930). Trans. as
The Shortest Way to Pianistic Perfection (1932). Reprint in Piano Technique. New York: Dover,
Givens, Larry. Re-Enacting the Artist . . . a Story of the Ampico Reproducing Piano. New York: Vestal
Press, 1970.
Goetschius, P., ed. See F. Mendelssohn-Bartholdy.
Grainger, Percy. “Collecting with the Phonograph.” Journal of the Folk-Song Society vol. 3, no. 12
(1908): 147–242.
———. “‘Grieg’s “Norwegian Bridal Procession’—A Master Lesson by Percy Grainger.” The Etude
vol. 38, no. 11 (1920): 741–45.
———. “Personal Recollections of Grieg.” The Musical Times vol. 48, no. 777 (1907): 720.
Green, Michael D. “Mathis Lussy’s Traité de l’expression musicale as a Window into Performance
Practice.” Music Theory Spectrum: The Journal of the Society for Music Theory vol. 16 (1994):
Grieg , Edvard. Diaries, Articles, Speeches. Ed. and trans. Finn Benestad and William H. Halverson.
Columbus, Ohio: Peer Gynt Press, 2001.
Grimaldi, Maria L . The Art of Piano Playing and Teaching. London: Reeves, c. 1895.
Grove, George, and Allan, Jean M. “Franklin Taylor.” The New Grove Dictionary of Music and Musicians,
ed. S. Sadie (London, 1980), vol. 18, 605.
Grunsky, Karl. Die Technik des Klavierausrugs. Etwickelt am dritten Akt von Wagners Tristan. Leipzig ,
Germany: 1911.
Gurewitsch, M. “True Colors.” Smithsonian Magazine, July 2008. Available: http://www.
Hall, Denis. “Duo-Art Rolls: A Description of Their Production and an Assessment of Their
Performance.” The Pianola Journal: The Journal of the Pianola Institute vol. 10 (1998): 5–53.
———. “Recording Welte-Mignon Piano Rolls in Germany.” The Pianola Journal: The Journal of the
Pianola Institute vol. 16 (2005): 49–52.
———. “The Reproducing Piano—What Can It Really Do?” The Pianola Journal: The Journal of the
Pianola Institute vol. 14 (2001): 3–26.
———. “A Window in Time—a Response.” The Pianola Journal: The Journal of the Pianola Institute
12 (1999): 10–12.
Hallé, Charles, ed. See Ludwig van Beethoven.
Hambourg , Mark. From Piano to Forte: A Thousand and One Notes. London: Cassell, 1931.
———. How to Become a Pianist. London: Pearson, 1922.
Hamer, Mark. “It Don’t Mean a Thing if It Ain’t Got That Swing. But What Is Swing?” New Scientist
vol. 168, nos. 2270/2271 (December 2000): 48–51.
316 Bibli og raphy

Hamilton, Kenneth. After the Golden Age: Romantic Pianism and Modern Performance. Oxford: Oxford
University Press, 2008.
———. “The Virtuoso Tradition.” The Cambridge Companion to the Piano, ed. David Rowland.
Cambridge: Cambridge University Press, 1998, 57–74.
Hanslick, Eduard. Music Criticisms 1846–99, rev. ed., trans. and ed. Henry Pleasants. London:
Penguin Books, 1963.
Harvith, J., and Edwards Harvith, S., eds. Edison, Musicians, and the Phonograph: A Century in
Retrospect. New York and London: Greenwood Press, 1987.
Hass, F. Zwischen Brahms und Wagner: Der Dirigent Hermann Levi. Zurich, Switzerland, and Mainz,
Germany, 1995.
Haynes, Bruce. The End of Early Music: A Period Performer’s History of Music for the Twenty-First
Century. New York: Oxford University Press, 2007.
Hefling , Stephen E. Rhythmic Alteration in Seventeenth- and Eighteenth-Century Music: Notes Inégales
and Overdotting. New York: Schirmer, 1993.
Herz, Henri. Méthode complète de piano Op. 100. Mainz, Germany: Schott, [1838].
Hofmann, Joseph. Piano Playing with Piano Questions Answered (1909). Facs. reprint of 1920 ed.
New York: Dover, 1976.
Holliday, Kent A . Reproducing Pianos Past and Present—Studies in the History and Interpretation of
Music, vol. 13. New York: Edwin Mellen Press, 1989.
Howard, Geoffrey. [Sleeve Notes]. Pupils of Theodore Leschetizky. Pearl Opal CD 9839 (1988).
Howat, Roy. “Review.” The Pianola Journal: The Journal of the Pianola Institute vol. 13 (2000): 39.
Hudson, Richard. “Rubato.” The New Grove Dictionary of Music and Musicians, 2nd ed., ed. Stanley
Sadie. London: Macmillan, 2001, vol. 21, 832–35.
———. Stolen Time: The History of Tempo Rubato. Oxford: Oxford University Press, 1994.
Hummel, Johann N. A Complete Theoretical and Practical Course of Instructions on the Art of Playing
the Piano Forte. London: Boosey, 1828.
Huray, Peter Le. Authenticity in Performance: Eighteenth-Century Case Studies. Cambridge: Cambridge
University Press, 1990.
Joachim, Joseph, and Moser, Andreas. Violinschule. 3 vols. Trans. Alfred Moffat. Berlin: Simrock, 1905.
Jollage, Charles-Alexandre. Premier livre de pièces de clavecin. Paris: 1738. Reprint, Geneva: Minkoff,
Kalkbrenner, Frédéric. Méthode pour apprendre le piano-forte à l’aide du guide-mains, Op. 108 (Paris:
1830). 2nd ed. Paris: Pleyel, c. 1831.
Kipnis, Igor, ed. The Harpsichord and Clavichord: An Encyclopedia. New York: Routledge, 2007.
Kitchen, John. “Harpsichord Music from England and Germany.” Early Music vol. 33, no. 3
(Aug. 2005): 530–32.
Kleczyński, Jan. O wykonywaniu dziel Szopena (Warsaw: 1879). Trans. Alfred Whittingham as How
to Play Chopin: The Works of Frederic Chopin and Their Proper Interpretation, 4th ed. London:
Reeves, 1882.
König , Werner. “The Welte-Mignon Reproducing Piano and Its Place in the History of Music.”
The Pianola Journal: The Journal of the Pianola Institute vol. 18 (2007): 52–62.
Kroll, Mark. Playing the Harpsichord Expressively: A Practical and Historical Guide. Lanham, Md.:
Scarecrow Press, 2004.
Kullak, Adolphe. Die Ästhetik des Klavierspiels (Berlin: 1861). Trans. of the 3rd ed. by Dr. Theodor H.
Baker as The Aesthetics of Pianoforte Playing, ed. Dr. Hans Bischoff. New York: 1893. Reprint,
New York: Da Capo Press, 1972.
Lawson, Rex. “Cleaning the Windows of Time.” The Pianola Journal: The Journal of the Pianola
Institute vol. 12 (1999): 13–44.
Lebert, Sigmund, and Stark, Ludwig. Grosse theoretische-praktische Klavierschule für systematischen
Unterricht nach allen Richtungen des Klavierspiels vom ersten Anfang bis zur höchsten Ausbilding,
3 parts. Stuttgart, Germany: 1858.
Lemare, Edwin H. “The Art of Organ Playing.” The Musical Educator vol. 4 (c. 1900): xiii–xviii.
Lenz, W. von. “Die grossen Pianoforte-Virtuosen unserer Zeit aus persönlicher Bekanntschaft:
Liszt, Chopin, Tausig, Henselt.” Neue Berliner Musikzeitung (1868), 22/37–39 (1868), 291–93,
Bibli og raphy 317

299–302, and 307–10. Republished as Die grossen Pianoforte-Virtuosen unserer Zeit aus persön-
licher Bekanntschaft: Liszt, Chopin, Tausig, Henselt (Berlin: 1872). Trans. M. Baker as The Great
Piano Virtuosos of Our Time: Liszt, Chopin, Tausig, Henselt (New York: 1899). Modern ed.,
ed. Philip Reder. London: Kahn and Averril, 1983.
Leschetizky, Theodor, ed. See John Field.
———. See Frédéric Chopin.
Library of Welte-Mignon (Licensee) Music Records [Catalogue]. New York: De Luxe Reproducing Roll
Corporation, 1927.
Lichtenthal, Pietro. Dizionario e bibliografia della musica. Milan, Italy: Fontana, 1826.
Liszt, Franz, Frédéric Chopin (1852); 2nd ed. (1879). Trans. Frederick Niecks as Chopin as a Man and
Musician. 2 vols. London: Novello, 1888.
———. “Préface” to Prométhée, in Symphonischen Dichtungen für groses Orchester. Leipzig, Germany:
Breitkopf & Härtel, 1856.
Lussy, Mathis. Traité de l’expression musicale: Accents, nuances et mouvements dans la musique vocale et
instrumentale. Paris: 1874; trans. of 6th ed. in 1892 by M. E. von Glehn as Musical Expression.
London: Novello, Ewer, [c. 1885].
Lyle, Wilson. A Dictionary of Pianists. London: Hale, 1985.
MacDowell, Edward. “Starlight Op. 55 No. 4.” Sea Pieces. Jung, 1898; Schmidt, 1899; Marian
MacDowell, 1926.
Manuscrit Barbe: Pièces de luth de différents auteurs en tablature française. Paris: c. 1690. Reprint,
Geneva: Minkoff, 1985.
Marcailhou, Gatien. L’Art de composer et d’exécuter la musique légére. London: Schott, 1854.
Marpurg , Friedrich W. Principes du clavecin. Berlin: 1756. Reprint, Geneva: Minkoff, 1974.
Marx, Adolf B. “Seb. Bach’s chromatische Fantasie—Einige Bemerkungen.” Allgemeine musikalische
Zeitung vol. 50, no. 3 (Jan. 19, 1848): 33–41.
Mason, Daniel G. “Postscript after Twenty-Five Years.” From Grieg to Brahms: Studies of Some Modern
Composers and Their Art (1902), 2nd ed. New York: Macmillan, 1927.
Mathias, G. “Preface” to I. Philipp, Exercise quotidiens tirés des oevres de Chopin. Paris: 1882.
Maunder, Richard. Domeninco Corri’s Treatises on Singing. 4 vols. Ed. Richard Maunder. New York:
Garland, 1993.
May, Florence. The Life of Johannes Brahms. 2 vols. London: Arnold, 1905.
Mayer Brown, Howard. “Pedantry or Liberation? A Sketch of the Historical Performance Movement.”
Authenticity and Early Music: A Symposium, ed. Nicholas Kenyon. Oxford: Oxford University
Press, 1988, 27–56.
McCallum, Peter. “When Seeking a Good Tan, Turn before You Burn.” Sydney Morning Herald,
Nov. 23, 1998, 15.
McEwen, J. B. Tempo Rubato or Time Variation in Musical Performance. London: Oxford University
Press, 1928.
McWilliams, Jerry. The Preservation and Restoration of Sound Recordings. Nashville, Tenn.: American
Association for State and Local History, 1979.
Mendelssohn-Bartholdy, Felix. “Rondo Capriccioso Op. 14.” Instruktive Ausgabe Klassischer
Klavierwerke, ed. Percy Goetschius. Stuttgart, Germany: Cotta, 1899.
Merrick, Frank. “Memories of Leschetizky.” Recorded Sound: The Journal of the British Institute of
Recorded Sound vol. 18 (1965): 335–39.
———. Practising the Piano. London: Rockliff, 1958.
Methuen-Campbell, James. “Theodor Leschetizky.” The New Grove Dictionary of Music and Musicians,
2nd ed., ed. Stanley Sadie. London: Macmillan, 2001, vol. 14, 584–85.
Mikuli, Karol. “Introductory Notes” to Frederic Chopin’s Complete Works for the Piano. New York:
Schirmer, 1895.
———. “Vorwort” to Frederic Chopin’s Pianoforte-Werke. 17 vols. Ed. Karol Mikuli. Leipzig, Germany:
Kistner, 1880.
Milchmeyer, Johann P. Die Wahre Art Das Pianoforte Zu Spielen (Dresden, Germany: 1797). Trans.
as The True Art of Playing the Pianoforte by Robert Rhein. DMA diss., University of Nebraska,
318 Bibli og raphy

Mitchell, Mark. Vladimir de Pachmann: A Piano Virtuoso’s Life and Art. Bloomington: Indiana
University Press, 2002.
Mitchell, Mark, and Evans, Allan, eds. Moriz Rosenthal in Words and Music. Bloomington: Indiana
University Press, 2006.
Moiseiwitsch: Opinions on Player Piano Rolls, interviewed by Denys Guerolt (Oct. 26, 1962), BBC
Archives LP 28654 (12” 33 1/3 rpm) mono disc.
Moore, Gerald. Am I Too Loud? (1962, 1966). Reprint, Harmondsworth, U.K.: Penguin, 1968.
———. Furthermoore: Interludes in an Accompanist’s Life. 1983. Reprinted in Collected Memoires.
Middlesex, U.K.: Penguin, 1986.
Moore, Jerrold N. [Sleeve Notes]. Raoul Pugno: His Complete Published Piano Solos. Pearl, Opal CD
9836 (1989).
Moroney, Davitt. “Couperin, Marpurg and Roeser: A Germanic Art de Toucher le Clavecin, or a
French Wahre Art?” The