8 The recorded document: Interpretation

and discography
SIMON TREZISE

To understand what we hear from recordings we must first understand

them as sources of evidence. Although initially the usefulness of recording
was unclear, in the twentieth century the industry became profitable
through focusing on entertainment. Providing consumers were enter-
tained they were content not to enquire too closely into the distortions
and illusions that recording created, especially as sound quality improved
and relative prices decreased. Thus throughout the past 110 years records
have been widely if naively accepted as surrogate accounts of live perfor-
mance. Nevertheless, reading through the medium reveals the extent to
which recording transmutes music-making. Even when production and
record seem at one, as in much rock music from the 1960s when albums
were created over long periods in the studio, records contain secrets that
challenge preconceptions. Outlining the most significant is one purpose of
this chapter. First, though, we need to know how to find recordings, and
how to date and place them: then we shall be in a better position to ask
about the sounds they encode.

Discographies and information trails

There are many discographies and other useful materials available via the
internet, but it is often impossible to estimate how accurate the informa-
tion is for there is less obvious editorial control or review than in printed
publications.1 Nevertheless, a comprehensive online discography for a sin-
gle performer, Eugene Ormandy for example,2 is often reliable, and when
an institution like the London Symphony Orchestra puts up discographical
information one has good reason to trust it.3 Digging into the frustratingly
awkward catalogues of the great sound archives, such as those at the
British Library, Library of Congress and Bibliothèque nationale, can often
yield useful information, but each is riddled with quirks and uncertainties.4
Much helpful information is given by internet retailers, such as the vast
American store ArkivMusic.com. Their excellent search engine enables
one to search by composer, performer, conductor, ensemble and label,
[186] so for example if one wishes to know how many performances conducted

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 187 The recorded document: Interpretation and discography

by Thomas Beecham are available from the store, one has only to drill
through the list of conductors to his name; then one has detailed informa-
tion by composer, orchestra and label. Unfortunately, Naxos’s excellent
historical series is not included, so there are gaps. More unpredictable but
staggeringly numerous results can be obtained by doing artist and composer
searches at allmusic.com. A search under the Beatles, for example, yields
much historical and biographical information, an extensive discography, a
list of some of the bootleg CDs, DVDs, videos and more. There are pictures
and a peculiar but possibly helpful list of moods encompassed in the songs.5
Even with so much online, the primary source of discographical
information is the library, and it takes time and practice to find what
one needs (plus a well-funded music section). There are two essential
publications, which are often the first place to look for information on
classical music on record during the 78 and early LP eras. One is WERM,
The World’s Encyclopedia of Record Music, which first appeared in 1952
and was followed by supplements; it sought to list all available recordings
of ‘permanent music’ (which in practice meant broadly canonical classical
music), plus selected ‘historical’ issues, by composer.6 Similar in structure
is The Gramophone Shop Encyclopaedia of Recorded Music, first published
in 1932 and revised in 1936 and 1948.7 Jazz collectors find extensive
information in three seminal works: Rust, Jepsen and Bruyninckx.8
Hundreds of discographies have been published since then, all falling
into one of three main categories, namely subject, performer and label.
Some appear separately as books; others appear in journals, as appendices
in books, and elsewhere.9 The last bibliography of discographies able to
satisfy a wide range of questions appeared in 1988: Michael Gray, Classical
Music Discographies, 1976–88, which formed a supplement to Gray’s
Bibliography of Discographies of 1977.10 Various information resources
can help to locate discographies, such as RILM, IIMP and Google Scholar,
as can some library catalogues, but some queries take a great deal of work.
Finding 78s and early LPs should become easier with the publication of the
CHARM online discography in 2009. Failing all these, in desperate straits
one can consult the annual catalogues issued by each record company,
many of which are available in the British Library.
The format and quality of discographies varies immensely, as does
their reliability. For example, a ‘subject’ who has fared well is Mahler,
culminating in the splendid Mahler Discography by Péter Fülöp.11 As well
as detail on sources and background, coverage is by work, with recordings
given in chronological order; there are performer indexes. In a summary
list of recordings of each work, timings for every movement and song are
given. In addition to details of ensemble and conductor, recording dates
and location are provided; there is a list of first releases on LP in various

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 188 Simon Trezise

countries, each with release date, the first CD release, the author of insert
notes, and reviews in journals such as Records and Recording. Entries are
stacked: the discography ‘presents each entry in a uniform stack of lines,
with each piece of information always appearing on the same line’.12
The ‘performer’ discography is the most common form, especially in the
documentation of singers. An example shows some of the pitfalls. John
Hunt has privately produced numerous discographies, many of them indis-
pensable. However, Giants of the Keyboard: Kempf, Gieseking, Fischer,
Haskil, Backhaus, Schnabel is problematic.13 Entries are arranged by com-
poser, but work order is inconsistent. Month and year are given but not the
day, which is usually known. The locale is given, typically the city but not
the venue although it is also known in many instances. Matrix numbers
(of which more shortly) are not given. The author fails to distinguish
between live and studio recordings. Concerning Gieseking, for example,
Donald Manildi writes:

Gieseking recorded Debussy’s Preludes, Children’s Corner, and Suite

bergamesque twice in the early 1950s … [As they] appeared in the US on
Columbia and Angel some collectors have assumed them to be identical,
but they are not. Hunt lumps them together … and makes no distinction
between them.14

Other recordings are omitted altogether.

‘Label’ discographies are the workhorses of discography. Once they have
been tackled much else becomes possible. His Master’s Voice: The German
Catalogue: A Complete Numerical Catalogue of German Gramophone
Recordings Made from 1898 to 1929 in Germany, Austria, and Elsewhere
by the Gramophone Company Ltd = Die Stimme seines Herrn, compiled by
Alan Kelly with the cooperation of the EMI Music Archive, London is a
prime example.15 The volume provides detailed history and explanation of
often highly complex matrix and catalogue-number sequences. Orchestral
records were given catalogue numbers 40500 to 40999, so the listing can be
done by type of recording, orchestra records for example, and catalogue
numbers. Entries are in columnar format with the information reading
from left to right, thus: catalogue number, matrix number, performers,
work and section details, other issues and couplings. Discographies of this
quality and extent are formidable and generous acts of scholarship.

What the discs tell us

Surface markings and surrounding documentation can tell us a great deal
about a recording. Most shellac-disc labels identify the composer, work,

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 189 The recorded document: Interpretation and discography

performers, and sometimes even the venue. To take an example at ran-

dom, DB 1555 is the second record in a three-disc HMV set of Borodin’s
Second Symphony. The following information is displayed on the label:
Company of origin: Gramophone Company, His Master’s Voice
Composer: Borodin
Work and movement(s): Symphony No. 2 in B Minor, third
movement – Andante (first record)
Orchestra and conductor: London Symphony Orchestra, conducted by
Albert Coates

Figure 8.1a HMV DB 1555, Matrix Cc17857-IIA, label, with warning sticker added by the
BBC Gramophone Library

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 190 Simon Trezise

Figure 8.1b HMV DB 1555, Matrix Cc17857-IIA, detail of matrix number

The catalogue number was assigned according to the area of distribution,

so a sister company might issue the same recording under its own cata-
logue number. This potentially confusing situation is resolved by the most
important number (or sequence of letters and numbers), which is pre-
served in the area between the grooved area and the label, usually at
6.00 o’clock (under the label). This is the matrix number, which is the
number of the master record. It is a unique identifier which can, when
the company’s records have survived (or there is an alternative source of
information), yield the date and place of the recording, often more. Early
matrix numbers identify the ‘expert’ responsible for the recording, such as
Fred Gaisberg, who is identified by a G or FG.16 RCA-Victor also printed
the matrix number on the label. Here the matrix number is Cc17857 IIA Δ.
HMV matrix numbers in the Bb and Cc series were mostly recorded in
London, Bb being used for ten-inch discs and Cc for twelve-inch. They
have been catalogued and dated by Alan Kelly,17 and so we know that
Cc17857 was recorded in the Kingsway Hall on 6 November 1929. The
roman-numeral II indicates that the side issued was of the second useable
take, and the A shows that this matrix was cut on the second of at least
two lathes simultaneously recording the same performance: in this case
the A lathe presumably produced the better-sounding disc, and take II
the better-sounding performance (not necessarily the most musical). The
triangle was a symbol used by HMV to designate electrical recordings
made by the American Westrex system.18
In contrast, the English Decca recording of extracts from Mendelssohn’s
A Midsummer Night’s Dream, catalogue number K. 1769, is documented
on the CHARM website in one of several discographies by Michael Gray,
this one being of the British Decca 78 catalogue.19 In addition to the label

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 191 The recorded document: Interpretation and discography

information, which gives the performers as the Concertgebouw Orchestra of

Amsterdam conducted by Eduard van Beinum, Gray gives the following:
Date and place of recording: 12 September 1946, Grote Zaal, Concertgebouw
Producer: [Victor] Olof
Engineer: [Ken] Wilkinson

Apart from the matrix number, other surface markings are of less
historical interest, but typically refer to ‘the development of negative
and positive metal parts from the wax or acetate session disc and to the
sequence of those transfers’.20
The diligence of companies in marking take numbers and, on some
sequences but not others, the use of a second transcription lathe at the
recording session (by a suffix A on HMV 78s) yields fascinating insights
into the practices of recording companies. With popular issues in parti-
cular, the matrices used for manufacture might eventually wear out and
the company would resort to substitution with an alternative take from
the original session. Rachmaninov’s Piano Concerto No. 2 in C Minor was
recorded on 10 and 13 April 1929 by the composer with the Philadelphia
Orchestra conducted by Leopold Stokowski. It appeared on Victor 8148-52
with the matrices CVE 48963-72. The original takes used were:
Original issue
I: 3, 1, 1
II: 1, 3, 2, 1
III: 2, 2, 1

The substitutions were made in the Second World War and affected nine
sides.
Substitutions
I: 2, 2, 2
II: 1, 1, 3, 2
III: 1, 3, 3

The original artists’ file session sheet was amended to indicate that the ‘the
substitute takes were the ones which had always been chosen for master-
ing’. This mistake was perpetuated in the official release of the original
takes on CD: the booklet accompanying the CD wrongly lists the sub-
stitute sides even though the originals were used.21 There are musically
significant differences between takes.
When a matrix number is not provided, as in the instance of Caruso’s
two recordings of ‘Celeste Aïda’ of 1908 and 1911, confusion may arise,
for both recordings were issued under the same Victor catalogue number
88127. One only discovers this through listening or visual identification:
the 1911 version has a shorter run-out area.22

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 192 Simon Trezise

Compared with 78s, LPs, EPs and many CDs are uninformative,
especially in the early days of LP. Columbia ML 4092 is an early twelve-
inch LP of Stravinsky conducting the Philharmonic Symphony Orchestra
of New York in Le sacre du printemps. The performers and work are
clearly labelled, and it was standard practice even in the early days of
LP to give a programme note, which 78s often lacked. There is, however,
no indication that this is the 1940 recording originally on Columbia 78s
11367–70. Another Michael Gray discography on the CHARM website
gives further details, including the location of the recording, Carnegie
Hall, the 78 matrix numbers, and an early CD transfer by Pearl.23
It later became customary for LPs to provide copyright information,
in the form of the publication date, but this is at best only a rough guide to
the date of the recording and may be many years after. The number
engraved on the surface between the grooves and the label refers to ‘an
edited session tape cut onto an LP record’ and is of no interest to the
historian, though some collectors find it useful in differentiating between
issues.24
With the advent of CD the recording process and its potential interest
to the consumer was sometimes taken more seriously. One of the first
Decca releases, for example, was of Janáček’s Sinfonietta and Taras Bulba
with Charles Mackerras conducting the Vienna Philharmonic Orchestra
(410 138–2). In addition to the publication date of 1983 (which applies
only to the CD, not the LP that preceded it), the location and approximate
date of the recording were also given: Sofiensaal, Vienna, March 1980.

Finding discs
Having established that a record was indeed made one has to find ways to
get hold of it. The largest collections are national and other major archives,
though many of them are unlikely to give the user access to the original
disc. Some will play the recording from a remote point in the building,
often with a microphone link to the engineer at the other end. Depending
upon local copyright laws, archives may make copies (for a price); in the
UK this usually means that anything recorded more than fifty years ago is
potentially in the public domain, though this may change. If access to
the original is required, there are a small number of collections that will
allow this, such as the sound archive at King’s College London. Otherwise
one may be able to pick up original copies from specialist dealers, includ-
ing Princeton Record Exchange, Academy Records in New York, Jerry’s
Records in Pittsburgh, Raymond Glaspole in Oxford, Mikrokosmos,
Norbeck and Peters, Nauck’s, and Larry Holdridge;25 eBay can be an

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 193 The recorded document: Interpretation and discography

excellent source as well, though one should be aware of the extremely

optimistic terms used to describe the condition of records on eBay and
elsewhere (G, ‘good’, denoting a record that is only just playable, if that). If
on the other hand one is able to work from commercial transfers (see later
in the chapter for the significant limitations this imposes), then a great
deal of material is available and one can search for it in the catalogues of
the transfer labels or on retailers’ websites like Amazon. Amazon is useful
since many reissues are no longer available and Amazon resellers may
have them second hand (remember also to try different national Amazon
sites: the French and Japanese ones can be very productive). Transfer
labels include the originating companies, such as DG and EMI, though
their level of activity is modest compared with the ‘independents’, includ-
ing Andante, Archipel, Avid, Biddulph, Brilliant Classics, Doremi, Dutton,
Dynamic, Guild, Hardy Classic, Lys, Membran, Music & Arts, Myto, Naxos,
Nimbus, Opera d’oro, Pearl, Regis, Tahra and Testament.

Diminutions and distortions

Having found a recording, and as much as we can about its origins, our
next challenge is to understand the respects in which the technology that
produced it imposes restrictions on the presentation of music. The follow-
ing are some of the major constraints.

Frequency response
The human ear in optimum, youthful condition can detect frequencies
from c. 20 to 20,000 Hz. Until after the Second World War technology
could reproduce a small part of this spectrum, so a great deal of informa-
tion heard at a live performance was lost. In the acoustic period this meant
that tinkling, sibilance, the natural brightness of a soprano voice, the bass
notes of a piano, and much more could neither be recorded nor repro-
duced. After the introduction of electrical recording in 1925 a steady
stream of innovations raised the ceiling from around 5,000 cycles to
15,000 by 1944.

Dynamic response
Until recently recording struggled with the extremes of loudness that
the ear can effortlessly disentangle, so experts (as producers or engineers
were called in the early days) sought to contain the dynamics. In the
acoustic period this could not be done on the recording equipment, so
the performer had either to modify her position relative to the receiving
horn or modify her performance so that it used less dynamic contrast than

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 194 Simon Trezise

it would at a concert. Pianists were encouraged to play loudly, often on

upright instruments with hardened hammers and missing backs; violins
were often replaced by instruments with metal sound boxes (Stroh vio-
lins); and (for example) Adelina Patti was placed on a ‘small movable
platform’ so that she could be pulled away from the horn when a loud note
approached and towards it for a quiet one.26 With the advent of micro-
phones ‘gain riding’ was applied by an engineer guided by a score-reading
assistant: an anticipated peak was curtailed and the level reinstated when
the music returned to, say, mezzo forte; quiet passages were amplified to
bring them over the level of the surface noise. Later, electronic compres-
sors and limiters would be used together with manual gain riding, for
it was often impossible for an engineer to react to sudden changes in
dynamic. Magnetic tapes used for recording in the LP era encompassed a
dynamic range of around 60 dB before distortion set in (the human ear
can accommodate around 120 dB). Compact disc went further (around
94 dB); new formats, such as DVD-Audio and SACD, match the human
ear. In many recording situations it is hard to know the extent to which the
musician performed as he or she was accustomed to do live, because the
recording process intervenes stealthily and sometimes with considerable
skill to retain a sense of a ‘real’ performance; on the other hand, some gain
riding is crude and immediately identifiable.

Spatial information
In the acoustic period performers were confined within a small space
and the recordings often lack a sense of depth, though this aspect should
not be overstated: great sophistication evolved in the use of horns and
many recordings give the impression of a soundstage. As the electrical era
dawned some music was recorded in a realistic way with orchestras set out
on the stages of concert halls, but many spatial anomalies have remained,
such as the use of recording booths for soloists, especially in popular song.
Unfamiliar spatial arrangements even in quite modern recordings can
interfere with sightlines and in other ways disrupt musicians’ accustomed
performing arrangements.

Timbral realism
The timbral quality of an instrument or voice in a recording is dependent
upon a multitude of factors. If the upper frequencies are missing, as they
are above around 3,500 Hz in an acoustic recording, this will affect the
realism of the sound. Other factors, such as the quality of a horn or micro-
phone, its placement, unsteadiness in the cutter or tape, the condition of the
commercial 78 after many playings, and the equipment used by the pur-
chaser, will determine the quality of the sound. If one is working from a

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 195 The recorded document: Interpretation and discography

commercial transfer of a 78, for example, some aspects of the timbre are the
product of undocumented decisions by the transfer engineer.

Duration
At first compact discs played for up to c. sixty minutes and have now
been extended to over eighty; high-definition formats (DVD-Audio and
video, SACD) play for longer. The earliest recordings lasted less than two
minutes. Until the late 1940s performers and listeners usually had to
make music in three- or four-minute units, depending on record diameter.
There is occasional evidence to suggest that performers accelerated their
performances to suit the restricted playing time, but a more common
consequence of it is extensive cutting of the musical text, which dimin-
ished over the years. Twelve-inch LPs, first issued in 1948, increased each
side’s playing time to around twenty minutes at first, and commercial
tapes, first issued in the 1950s, extended this still further.

Commercial, political, racial, social, and other agendas

Race, society, commercial pressure, and other ideologies and agendas
have affected the production and distribution of gramophone records.
Racism in America explains why the early propagation of jazz on record
came late and mostly with white musicians. Commercial concerns (and
differences in practice between England and America) forced Sergei
Rachmaninov to record short works; he was not given the opportunity
to record major examples of his core repertoire such as Beethoven’s last
piano sonata.

Expediency
Expediency plays a revelatory role in the history of recording. For exam-
ple, in the early part of its existence RCA and companies in Europe found
they could record in concert-like acoustics with credible results. That they
chose not to must be partly explained by the effort and expenditure
entailed: if Lionel Mapleson was able to make sometimes quite convincing
recordings of live Metropolitan Opera performances on an amateur cylin-
der set-up in the period 1901–4, major companies might surely have done
likewise had not the small, cramped, acoustically dead studio been more
expedient.
In addition to factors restricting the recording of music, once the
recorded document entered the market place it was subject to the whims
and vicissitudes of its many differently disposed owners. Records were
themselves the subject of performance. Rituals were enacted; finely engi-
neered products were treated with a obsessive, quasi-religious concern for
their faithful reproduction by some and with a carelessness boarding on

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 196 Simon Trezise

crassness by others. These extremes are already discernible in the early

days of 78s and peaked in the age of the LP.

Playback curves
Early recordings were recorded at many speeds around 78 rpm, but
working out what they were in each case is based on knowledge, judge-
ment, taste and guesswork, so that is another constraint. Pre-war and
many post-war electrical 78s were recorded with a system of equalisation
that attenuated the mid to bass frequencies (and in some cases boosted the
treble as well), as were LPs. In order to achieve optimum playback one
seeks a near-exact reciprocal of the original ‘curve’ so that the mid-to-low
frequencies are amplified and thereby reproduced at the strength heard in
the studio. Electrical gramophones, which started to appear in the late
1920s, could do this to an extent, and from the 1950s amplifiers were
appearing that could be adjusted for the various curves used by LPs
(standardised to the LP curve in the mid-1950s).

Contingencies of playback
There is no limit to the capacity of consumers to affect playback in their
homes. Anyone old enough to remember 78s and LPs will recall discs
warped by electric fires, LPs with soup spilt on them, scratches on the
playing surface, and so on. Some record players were almost infinitely
adjustable for the enthusiast, so great variations in playback could be
achieved, including, on most machines, speed adjustment for 78s.

Case studies of recording

The following examples examine recordings that are characteristic of their
time. Under critical scrutiny they emerge as complex documents that
demand interpretation in the same way that written texts do. The studies
include consideration of the crucial role of the remastering engineer.

Patti
Adelina Patti was the most acclaimed soprano of the second half of the
nineteenth century. She made a number of recordings over two extended
sessions in 1905 and 1906 at her home in Craig-Y-Nos (South Wales). For
many they represent the epitome of golden-age singing, albeit diminished
by the singer’s age and the limitations of the acoustic gramophone. The
recording scaled her voice down chiefly by reducing the amplitude – hence
the need to roll her away from the horn at climaxes – and by removing
the upper harmonics of her voice above c. 3,500 cycles. The acoustic

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 197 The recorded document: Interpretation and discography

properties of the horn added a complex series of frequency distortions. We

cannot say how bright or mellow her voice was, not just because the high
frequencies are missing, but because we cannot be sure what pitch the
records should be played at. It is widely accepted that she transposed
Zerlina’s ‘Batti, batti, O bel Masetto’ down from F to E♭. Played back at
74 rpm we hear the song at this pitch, but the A is set to 440: the likelihood
is that her piano was tuned to a lower pitch, so some transfers are possibly
too fast. Played in F at 440, ‘Batti, batti’ sounds bright, very impressive
in the runs, despite a certain shrillness in her voice (age and wear). But
if we take the speed down below 74 rpm, her voice is gentler, less brilliant,
and the runs now seem more realistic for a singer in the twilight of her
career.

Boston Symphony Orchestra

The Boston Symphony Orchestra made its first recordings in 1917 under
the direction of Karl Muck. It was not the first time an American orchestra
had been placed in front of a horn, but it was still a major event subject to
much planning and promotion. The conductor hoped to record move-
ments of symphonies, including the finales of Tchaikovsky’s Fourth and
Beethoven’s Seventh, but RCA officials also brought a stash of short
popular works, such as the Act III Prelude from Lohengrin, which formed
the basis of most of the takes. The orchestra of 100 men travelled down to
Philadelphia from Boston in an October heat wave, arriving exhausted
and irritable in Camden, New Jersey on the morning of 2 October, to be
ushered into a studio in which two igloos with door-like openings had
been constructed. In front of the igloos was a stand for the conductor;
behind the conductor was the turntable for the master disc; according
to one account, first-desk men played outside the igloos directly into
horns of their own, but the rest of the orchestra was packed tightly inside –
one igloo for the strings, the other for the wind – in order to maximise
the sound available to the recording horns. Boaz Piller, a bassoon player
in the orchestra, recalled that some instruments were not registering and
were obliged to come out of the huts and sit right in front of the horn
during solos. The oboe was most awkward during the ‘transition section’
of the finale of Tchaikovsky’s Fourth and in the end they had him play
directly into the horn.
Each side had to be repeated many times, first to obtain the right
balance, which could only be judged by playing the wax just recorded
(thereby destroying it), and then because of mistakes caused by the
difficult playing conditions, tension and heat. The triumphant cymbal
crashes were considered the last word in high-fidelity and put down a
marker for future generations of orchestral recordings.27 Pictures of other

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 198 Simon Trezise

acoustic-recording sessions show that unusual seating arrangements were

inevitable, though igloos of this sort were unusual.
The Tchaikovsky recording is impressive as presented on compact
disc by Ward Marston.28 One can hear more of the first desks than the
whole string section: indeed, it is hard to believe that the orchestra was at
full strength. The brass section registers well too, and one can detect at
around 4:45, where the oboist had to penetrate the receiving horn, that the
balance is unnatural. In spite of conditions, the performance is incisive
and exciting. Marston even manages to uncover a deepish bass line, albeit
from a tuba player, not the string basses. Signs of the old rubato style of
playing are missing: there is little portamento and much of the playing is in
tempo. In the rhythmic emphasis on rapidly progressing to the downbeat
from the previous bar, rather than the modern manner of ‘lengthening’ the
bar line, however, the performance is distinctly of its period.

Gennett jazz recordings

Some of the earliest jazz recordings, after the pioneering work of the Original
Dixieland Jazz Band (1917), were made by a small company in Richmond
Indiana called Gennett, an offshoot of Starr Piano, a manufacturer of pianos
and phonographs. Compared with the East Coast’s giants Victor and
Columbia it was a modest operation, but historically it is crucial.29 Debate
will continue as to the impact recording had on the evolution of jazz, but it
seems likely that it has been overstated. Certainly recording confined jazz
numbers to three minutes for the ten-inch side and thereby misrepresented
the style; it forced musicians to work out their performances in precise
detail as time in recording sessions was limited, so improvisation was largely
quashed. And the original instrumentation of jazz bands had to be modified
to facilitate acoustic recording: string bass was replaced by tuba and a full
percussion kit was reduced to wood blocks and/or cymbals. Such recordings
were the means by which jazz spread throughout large parts of America and
beyond. But, equally, much jazz was distributed in notated arrangements,
and the notion of writing out what was to be played for a session was by
no means uncongenial to the players, who were paid for arrangements.
Moreover, the numbers recorded were often versions of existing songs with
verse–chorus forms, which tended to coincide with the time requirements
of recording.30
Nevertheless a glance at the early Gennett sessions reveals how much
the musicians had to learn. The studio was a metre from a secondary
railroad spur, which could interrupt sessions. The studio was 38  9
metres and adjoined a control room separated from the recording area by
a double pane of glass. Drapery deadened the wooden walls so that the
acoustic was quite dead. Players clustered around two horns, typically with

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 199 The recorded document: Interpretation and discography

the reduced percussion section close to one, next to the piano; woodwind
and plucked strings were close but to one side, and brass players several
metres away and off to the right. Local or passing bands were signed up and
their first impression of the studio (as of all studios during the acoustic era,
even if local weather conditions made this one worse than most) would be
its heat – as high as 30 degrees Celsius, in order to keep the wax soft. Dozens
of waxes might have to be cut and thrown away simply to get the balance
correct. Even then many of the balances are bizarre. Once the recording
began, a red light would flick on to indicate that 2 minutes 30 seconds had
passed and the song should soon end. Each song would be recorded three
times in order to produce one publishable copy, but many numbers were
never issued. The artists had no say on what was issued and indeed made
little money from the whole enterprise, but some received copies of the
records which they could use to promote their careers.
The King Oliver band, signed in 1923, made twenty-seven takes. The
poor Gennett sound quality sucks dynamic life out of the ensemble and
lacks a bass line; trombone and clarinet dominate at the expense of the
cornets; Louis Armstrong’s counter melodies seem backward.
Joe [Oliver] and Louis stood right next to each other as they always had, and
you couldn’t hear a note that Joe was playing … [so] they put Louis about
fifteen feet over in the corner, looking all sad.31

Lillian Hardin Armstrong’s famous story has been qualified since, but
it indicates how balance was achieved. Rick Kennedy concludes from his
discussion of these historic discs that the recordings were an ‘ensemble
effort’: New Orleans jazz ‘was not grounded in improvisation’.32
In 1925, still recording acoustically, Hoagy Carmichael and the Happy
Harmonists recorded a number entitled ‘Washboard Blues’, which was
rejected as it was twenty seconds short. Carmichael insisted and was told
to come back in ten minutes and fill in the gap with a piano solo, which the
terrified law student did magnificently.33 Several transfers of this extra-
ordinary recording have appeared, all different in the amount of surface
noise removed, equalisation, and so on.34

Vaughan Williams, Symphony No. 4

Conscientious collectors of the time of the composer’s 1937 recording had
numerous means to optimise playback, including different grades of
needle, various shapes of soundbox, weight adjusters and, for owners of
acoustic record players, the Lifebelt, a tube for ‘insertion between the tone-
arm and sound-box … to give the reproducing stylus a certain quality of
flexibility’, promoted in The Gramophone by Compton Mackenzie (the
novelist and founder of the magazine).35

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 200 Simon Trezise

Figure 8.2 Lifebelt and Weight Adjuster, from an advertisement in The Gramophone, 1926

Anyone who has heard the exquisite sound obtainable from a well-
adjusted EMG gramophone must surely admit that collectors in the early
twentieth century had all they needed to bask in ‘perfect’ reproduction
in their homes, enjoying a harmony with the media and means of
reproduction comparable to hi-fi buffs in the LP era: reproduction was

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 201 The recorded document: Interpretation and discography

perfectible, just as it had been in 1907 and was to be again in 1967. Few
people play their own 78s now, so somebody else has to do it. Transfer
engineers usually work for themselves in studios conforming to no set
format. Although many have audio-engineering backgrounds, they find
their own way to obtaining a suitable product for modern consumption.
In taking a set of 78s, such as the Vaughan Williams Fourth, and
bringing it to CD, they interpret the discs according to an intricate set
of criteria; no two transfers sound alike, as so many decisions are
founded on personal taste and experience. The following list summarises
the process:

 The record is cleaned and centred on the platter so the stylus does not swing from
left to right.
 Pitch is checked for every side in case there are deviations; most engineers
confronted with a record of this period would expect concert A to be at 440.
 A stylus that best fits the groove wall and therefore produces least surface noise in
exchange for more audio information is chosen (it may not fit all sides).
 As the records were made with the lower frequencies attenuated, a suitable
turnover curve has to be selected on a specialist amplifier; for an HMV recording
of this period the turnover is likely to be around 200–250 Hz (around middle C),
from whence there is a 6 dB attenuation every octave, though there are many
variations; the treble should be played flat.
 High-frequency broadband noise above the musical signal has to be filtered.
 Most engineers do much more equalisation, often with a view to achieving as
natural an orchestral sound as was heard in the hall in 1937 (most of the original
recording venues used no longer exist so much guesswork and experience of
other recordings is involved).
 Other processes are used to remove noise, including declicking and decrackling.
 New processes from CEDAR, for example, notably Retouch, can be used to
remove almost any unwanted noise, including coughs, creaky chairs, and cutter
rumble.36
 More touching up is possible, including the elimination of discrepancies of
acoustic, microphone set-up, level, etc. between sides.
 The sides are joined up, sometimes wrongly, as in the last two sides of this
symphony, where several transfers lop out a bar before the recapitulation.

The composer’s recording of his Symphony No. 4 was made in Abbey

Road Studio No. 1. It is a biting, incisive, disturbing recording, generally
faster than any since. In the LP period one transfer was made (c. 1969–70),
by Anthony Griffith, who used a razor blade, an analogue equaliser, a reel-
to-reel tape recorder, and probably no form of analogue noise suppression
other than some steep filtering. The Gramophone reviewer of 1970 was
impressed; he wrote, ‘The recording sounds extraordinarily good. Some
things may seem “fiddled” (e.g. very forward woodwind at times) but it is
full and still has plenty of quality.’37

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 202 Simon Trezise

Table 8.1

CD issue Transfer engineer Year

Koch 37018–2 Mark Obert-Thorn 1990

Dutton CDAX 8011 Michael Dutton 1995
Avid AMSC 599 David Wright, Dave Bennett 1998
Pearl GEMS 0062 Roger Beardsley 1999
Naxos 8.111048 Mark Obert-Thorn 2006

At least five CD transfers have been issued, as shown in Table 8.1.

The authors of the 1994 edition of the Penguin Guide to Compact Discs
argue that Koch’s ‘transfer is less full-bodied than the LP version … (the
upper strings are lacking in timbre)’.38 The Gramophone reviewer in April
1991 had been no happier:
In 1970 World Records issued an Anthony Griffith LP transfer of the Fourth
Symphony (6/70 – nla), which reproduced pretty well the blunt unvarnished
but full-bodied sound of the original 78s. This transfer was used for a fairly
recent LP reissue on EMI (2/88 – nla), but there the sound was rather
smoothed out, so that body and impact were reduced. The new Koch
International transfer, alas, is no better. Both ends of the sound spectrum
have been compressed, and the performance again loses some of its weight
and strength.39

The Koch release was presumably similar in technique to the LP: analogue
tape and equalisation without CEDAR for declicking. Comparing it with
Mark Obert-Thorn’s latest transfer for Naxos, it is clear that heavier
filtering was needed to dampen the shellac noise.
Dutton’s transfer is markedly different. CEDAR processing has been
liberally applied, to the extent that the scratch and much broadband noise
have been magically washed away. Some artificial reverberation has been
added to make up for what is lost in the processing. Also notable in this
transfer is a heavy reduction in the frequencies between around 3 and 4 kHz,
perhaps in response to a desire to ameliorate the old-sounding aspect of
this recording and make it smoother, more modern in character. Left and
right channels are slightly different, suggesting artificial-stereo processing.
Like Dutton, Avid attempts to belie the age of the recording. On the
cover the interventions are trumpeted: ‘Audiophile Quality Remastering’
and ‘In the Clarity of 3-Dimensional Sound’. The authors acknowledge in
their notes that processing is not to everybody’s taste but do not disap-
prove of it themselves: their ‘aim is to produce as natural and pure a sound
character that lacks harshness and possesses a smooth response that falls
gratefully on the ear, being comfortable and clear’. As with Dutton the
remaining background noise varies in intensity and the quieter passages,

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 203 The recorded document: Interpretation and discography

where the noise reduction has had to cut most deeply, have a quality that
might be characterised as ‘submerged’.
The Pearl issue reflects a less-interventionist philosophy than Dutton
or Avid, but more than Koch and Naxos. It has the most generous bass of
all the transfers and most vigorous attack in highly modulated passages.
CEDAR or another process has removed the scratch almost completely
but, as in the Naxos issue, no attempt is made to remove all the broadband
noise; it is only attenuated by equalisation. It seems that some reverbera-
tion has been added to compensate for the rather lack-lustre acoustic of
Abbey Road at the time.
The Naxos transfer illustrates a relatively purist approach to the process.
I can detect no added reverberation; much of the broadband noise between
around 8 and 10 kHz has been filtered out, but there is no evidence of digital
hiss reduction. Compared with the Pearl transfer, sudden dynamic changes
seem to have marginally less impact; this perhaps reflects the amplification
system.
Differences between transfers affect the way the 78s are heard and their
effect on listeners. For all the strong personality of the original 78s and the
performance, each CD has its own distinct character.

The Beatles
The first two Beatles albums were recorded on two tracks (not stereo) in
conditions similar to a live performance. The left channel was used for the
rhythm section and the right for the voice. After that EMI switched to
four-track recorders, which allowed for a more subtle mixdown to stereo,
allowing the voice to be placed centrally in the mix and melody instru-
ments to be in either the left or right channel. Although the early albums
were recorded quickly with little manipulation, in contrast to the weeks of
studio time expended on albums like Revolver (1966) and Sergeant
Pepper’s Lonely Hearts Club Band (1967), one might assume that the
LPs issued by Parlophone offered definitive versions of the material. As
exhaustive research, hundreds of bootlegs, and terabytes of internet
discourse show, there is no standard version of some Beatles songs. The
first four albums, Please Please Me, With the Beatles (1963), A Hard Day’s
Night and Beatles for Sale (1964) were simultaneously issued in mono and
stereo. At the same time, Capitol issued the music in the United States; but,
disliking the subtle, fairly dry acoustic of the originals which properly
emphasises where reverberation has been used, such as on John Lennon’s
voice, Capitol drenched the songs in reverberation. The company remixed
the stereo versions to eliminate the hole-in-the-middle effect on some of
the EMI versions, thereby creating a wider but woollier soundstage. As
twelve songs was the maximum ration allowed on their LP issues, two

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 204 Simon Trezise

songs were lopped off, and the compilations were quite different (they
included hit singles, including ‘I Want to Hold Your Hand’, which EMI had
issued on EPs and therefore disdained to issue on LP as well). Dynamic
contrast was reduced by compression and/or limiting. Recently Capitol
reissued the US LP mixes on CD. Unlike EMI, who arbitrarily constrained
the first four albums to mono and the remainder to stereo – much to the
chagrin of the informed collector – Capitol included both stereo and mono
mixes in the CD issues. In spite of this apparent concern for historical
veracity, the CD issue has tended to reduce the already modest dynamic
range present on the Capitol LPs between songs, and dynamic variety within
individual songs has also been curtailed.
That there are not always single, definitive versions of some Beatles
songs is evidenced by the manifold differences between the stereo and
mono mixes, such as the forward presence of the vocal lead. Typical of the
differences that may also arise between stereo mixes is ‘Strawberry Fields
Forever’, one of the most studio-manipulated songs the group released.
Mono remix 12 was issued by EMI in mono on EP on 17 February 1967
(Parlophone R 5570). Capitol issued a stereo mix, the first to appear,
on 27 November 1967 on the Magical Mystery Tour LP (Capitol SMAL
2835). The same stereo mix was issued on LP in the UK by EMI in 1976
(Parlophone PCTC 255). In the interim German Apple issued a Magical
Mystery Tour LP using a different stereo mix made in 1971 (SHZE 327),
which now features worldwide on CD issues (all this emphasises the
importance of mono throughout the 1960s for popular music and EMI’s
lack of prescience in issuing only stereo mixes on CD). Joseph Brennan
writes in detail about the discrepancies between these two stereo versions:
The most obvious difference in the ‘German mix’ is that it is much clearer,
with a better percussion sound, and more stereo separation. Despite the
improved reproduction, some nice touches of the earlier mix are lacking.
If there is any doubt it is a different mix and not just a cleaner version of
the same mix, listen for the first three instances of panning … On Stereo
remix 3, when we first hear take 26, the cello and trumpet sound pans
quickly from left to right, a little sleight-of-hand to distract the listener from
the edit itself to the introduction of the new instruments to the song, while
in the ‘German mix’ the cello and trumpet track is on the right to start
with [etc.].40

Other versions of the song, which do not always have the official impri-
matur of the group or EMI and Apple, but are intended to illuminate the
complex recording history of the song, are available in volume 2 of The
Beatles Anthology (Apple/EMI 8 34448 2) and numerous bootlegs.41
The sound of the CD issues of the Beatles closely reflects the LP mixes,
many of which were made on four-track tape. To free up extra tracks, four

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 205 The recorded document: Interpretation and discography

tracks could be mixed down to two. Each time this happened a generation
of sound quality was lost, and further degradation of the original song
masters would be inflicted when the tapes were prepared for EP and LP
release. The version of ‘Penny Lane’ in volume 2 of The Beatles Anthology
sounds quite different from the version on the Magical Mystery Tour CD
(EMI CDP 7 48062 2), not just because it is a composite version of several
different takes, but because the original tracks were mixed down to digital
tape, thereby preserving the very high quality of the first layer of record-
ing.42 Many hope EMI will eventually remaster the Beatles’ recordings, but
when they do the notion of a definitive version of each song will become
even more strained.

Don Giovanni
Mackenzie poured scorn on the notion of stereophony in 1926, denoun-
cing ‘merely spatial illusion [as] worse than useless’.43 Nevertheless,
engineers on both sides of the Atlantic wanted it, especially for classical
music. Tape made stereo recording straightforward, and it facilitated
highly detailed editing, manipulation of the sound, such as the addition
of reverberation, and multi-channel mastering for mixing down to mono
or stereo. Early stereo records included trains, and pianos falling from
high buildings, but tests conducted early in the history of stereo, and one’s
own experience, call into doubt the realism claimed for it. Test audiences
were unable precisely to locate correctly the real position of instruments in
an orchestra when a stereo signal was relayed live to another room, and
when the microphones were moved the instruments moved and the image
was changed. In a live concert it is usually hard to locate instruments with
one’s ears alone: the eyes have to assist. The reason for this may be that
stereo audio systems and recordings tend to emphasise the frequencies
holding much of the spatial information so the listener gets precise cues as
to the location of instruments.44 Stereo, in these circumstances, becomes
an end in itself, a great improvement on mono, to be sure, because one
hears so much more – violins placed either side of the conductor can
be heard as such – but not necessarily a realistic reproduction of the live
experience. Critics first encountering stereo nevertheless ‘agreed that the
stereophonic illusion was fascinating to hear’, as they have ever since. John
Crabbe’s verdict on home listening is worthy of quotation in this context:
There is a common assumption in the hi-fi community that our goal is
the re-creation at the listener’s ears of a sound pattern identical to that
obtained at a good seat in the concert hall … For good or ill we are in the
hands of the recording producer, and the most we can hope of our domestic
reproduction is to create the sort of balance … determined by the producer
during recording.45

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 206 Simon Trezise

For opera the attraction of stereo was that action on stage could be
transmuted into movement between the left and right channels, accom-
panied by various distance effects. The producer and musicians would
attempt to re-create the dramatic dialogue and an illusion of the real-time
sequence of events, even though in recordings this is often an illusion.
When, for example, Daniel Barenboim wished to record the Don Giovanni
he was conducting at the 1973 Edinburgh Festival,46 EMI set up a schedule,
secure in the knowledge that all the singers were ‘anchored to Edinburgh
and available for all sessions’. The singer cast as the Don, Roger Soyer, sang
sotto voce during the balancing session and it emerged that he was suffering
from laryngitis. The first scene was put aside and the Masetto and Zerlina
arias were taken in the first recording session. The remaining sessions in
Edinburgh took all the music in which the Don was not directly involved,
including the closing scene after he is dragged off to hell. Schedules were set
up in London that would involve attempting to duplicate the complicated
balance painfully achieved in Edinburgh (the Assembly Hall at the George
Watson College was being redecorated during the recording). Just before
the London sessions Helen Donath rang in sick. The producer, Suvi Raj
Grubb, toyed with postponement or a new Zerlina, but finally made the
agonising decision to record all the music minus Zerlina and superimpose
her voice when she was fit. Soyer was horrified at the prospect of singing ‘La
ci darem’ to an empty space, so a few recitatives and the duet were scheduled
for a later time. Even so, Zerlina had to be superimposed elsewhere, as in the
finale of the first act, for which session two tiny loudspeakers were placed
behind Barenboim and Donath. Donath was reacting to the recording and
came in late, so in the end she sang to silence, taking only Barenboim’s beat
for her cues.
The final master tapes comprised around half the opera on two-track
tapes and three quarters of the rest on eight-track tapes made in Kingsway
Hall; the rest had Zerlina’s voice superimposed on two of these eight.
Typical of the gruelling editing involved was joining up the recitative
preceding Leporello’s aria ‘Madamina’ – the join was to be made on the ‘s’
of ‘testimon’:

In the split second that [Geraint] Evans pronounced that ‘s’ we had,
simultaneously, to fade out the eight-track recording, fade in the two-track
recording and make all the adjustments necessary for the two sounds to
match.47

The first attempt at the edit had Leporello jumping like a ‘startled rabbit’ at
the join;48 it took over an hour to get it right.
This set may have been unusually awkward, but the process by which it
came into being is representative of tape recording generally and of stereo.

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 207 The recorded document: Interpretation and discography

The question of the extent to which so much artifice created an analogue of

one of the Edinburgh performances is beyond the scope of this chapter.
The recording is full of the dramatic detail and eventfulness that stereo
brings, but it is possible that the ‘phonograph effect’ (a valuable concept
used by Mark Katz in his dissertation and subsequent book),49 perhaps
part of which is described by Grubb in his marvellous account of the
recording sessions, is present in the final product.

A view on recording
As these case studies attempt to demonstrate, a recording does not ‘show’
a performance to us, for the performance that generated the recorded
artifact is hidden: the relationship is not mimetic; rather, we may regard a
record as forming a diegesis with and within its domestic or other envir-
onment. The record and associated equipment are telling us about a
performance, but it is not the performance itself; it is filtered through a
large number of processes and contexts with which the original performer
has nothing to do. The diegesis is dependent upon a long sequence that
runs from the muscles of the pianist, through the coils of a microphone,
down metres of wire, to a sapphire needle, past circuits amplifying some
frequencies more than others, through murky black compounds, to the
final phase, the installation of the listening event in the living room. The
positioning of the record player, the condition of the record, the attention
given to the playback equipment (such as the needle), the soundbox, and
so on are part of the process. Much of this complex chain is hidden during
the telling of the ‘story’ – playback – but unlike a live performance this
story can be repeated, and each time it is different in some ways and
alike in others, to the extent that a click in the record, a momentary drop in
pitch, or a split horn note become part of one’s knowledge of the music.
A curious slant on this comes from an unlikely source: Patti’s niece Louise.

I must say I was terribly disappointed with all the numbers except ‘Voi
che sapete’. How unsatisfactory it is that when you want a thing in a high
key it alters the tempo so that things go at a terrific speed; and to lower it
everything must drag. Still it is very wonderful, though not artistic.50

This performance of the records occurred after dinner with the diva
herself present. It seems that the records were regarded as flexible in
terms of their realisation: this exalted audience, the letter implies, experi-
enced the music in a variety of keys.
Pursuing the issue of reproduction draws us into a labyrinth of knowl-
edge that people no longer possess or have access to. The assumption

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 208 Simon Trezise

that progress automatically invalidates the listening habits of a previous

generation is bogus and does a disservice to the many dedicated listeners
who believed they had found audio perfection: for them there was no
urgent requirement that tomorrow would bring improvements, for they
had found a delicate equilibrium with the resources then available. If
we need a modern analogy we might take the user of the MPEG player:
listening on trains and buses, the listener frequently plays the music in a
degraded format (MP3, typically 10 per cent of the information on com-
pact disc); and the earphones generally used are of poor quality and do not
cancel out background noise. In spite of this, millions of people enjoy
music in this way: social conventions, lifestyle, musical taste and many
other elements affect the telling of the musical story, transmuting the
materials by, for example, removing the original sequence of numbers
by picking out favourite songs.51
The gramophone is in a sense the narrator; the playback and listening
is the narration. Narrators lie, all the time. Extra information, new asso-
ciations, may rupture the illusion. Records are complex documents. To
treat them as if they are the same as performances we attend is inadvisable.
They are products of an infinitely variable matching of medium and
reproduction, allied to external knowledge culled from inserts, other knowl-
edge, expectation, and so on. Nevertheless, at a key point in the chain that
leads to the record, real historical performers exercised their larynxes and
arms to make music: their exhalations and muscular gymnastics live
on, engraved in the grooves, metamorphosed by a hundred different move-
ments, electrical circuits, and razor blades.

Copyright and recordings

NICHOLAS COOK

Copyright in relation to recordings is something of a nightmare, though for

different reasons in different places. A single recording may be covered by
several separate copyrights. For a pop CD there will be copyright in the song
(that is, the musical work), the lyrics, and the recording itself, as well as liner
notes or illustrations. For a modern recording of Beethoven’s Ninth
Symphony there will not be copyright in Beethoven’s music or Schiller’s
words (both are in the public domain), but there will be copyright in the
recording. Even if a historic recording is out of copyright (as explained
below), a CD reissue will have copyright in the transfer.
The practical significance of this depends on whether we are talking about
the general dissemination of recordings, or their use for purposes of study
and research. We’ll take general dissemination first. In the EU, copyright
in recordings lasts at the time of writing for fifty years, after which the

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023
 209 The recorded document: Interpretation and discography

recording goes into the public domain (though the musical work and lyrics
may still be covered, since copyright on them lasts for seventy years after
the author’s death). At the time of writing, however, the record industry is
trying to persuade the EU Commission to extend copyright on recordings to
ninety-five years, citing the precedent of the USA, where most recordings
(even those dating back to the 1890s) will remain in copyright until 2067,
thereafter enjoying a term of 95 years. The result is that the bulk of the
recorded heritage is not legally accessible in the US, since copyright owners
have reissued only a very small proportion of it: the entire historical collection
of Naxos Music Library, for instance, is barred to US subscribers. Should the
EU follow the US, the outlook for the rapidly developing public interest in
early recordings will be bleak.
In terms of access to recordings for purposes of study or research the
picture is complicated but less bleak. In the USA, recordings fall under ‘fair
use’ legislation, which sets out circumstances under which you are allowed
to make copies of copyright items, and these circumstances include
scholarship, research and criticism. So it is legal to copy an otherwise
unobtainable recording for research purposes, and, if you are publishing a
web article on it, you can include a clip from it, as long as you don’t take too
much (though it has no legal standing, 5 per cent is a generally accepted
limit). The UK has ‘fair dealing’ legislation which is broadly comparable to
American ‘fair use’, except that sound recordings fall only partly within it:
you can copy recordings for purposes of ‘criticism and review’, so the web
article example is covered, but not for research purposes. This is clearly
anomalous, and in 2007 the Gowers Review of Intellectual Property
recommended that copying of recordings for research purposes should be
allowed; legislation is in train. This would of course be particularly
important if the EU were to extend the copyright term.
Copyright is always complicated, especially as regards sound recordings,
and the above is not a complete or authoritative statement of the law. For
further information check the guidance available in your country.

Downloaded from https://www.cambridge.org/core. Royal Holloway, University of London, on 17 Mar 2019 at 20:34:18, subject to the Cambridge Core terms of use,
Cambridge Companions
available at https://www.cambridge.org/core/terms. Online © Cambridge University Press, 2011
https://doi.org/10.1017/CCOL9780521865821.023