New! 1-page of the goldberg variations New!

Part 1

3 The Analysis of Music @

New! 1-page of the goldberg variations New!

Here, I am going to analyze the Goldberg Variations BWV988 as "Music of Intellect" from the score and from performances.

(1) The Analysis of the Score @

(a) The Score @
the name of the original (in German)

Clavier-Ubung bestehend in einer Aria mit verschiedenen Veraenderungen vors Clavicimbal mit 2 Manualen. Denen Liebhabern zur Gemuths-Ergetzung verfertiget von Johann Sebastian Bach, Konigl, Pohl. und Churfl. Saechs. Hof-Compositeur, Capellmeister u. Directore Chori Musici in Leipzig. Nurnberg in Verlegung Balthasar Schmids.
Keyboard practice or clavier etude consisting of an Aria with different variations for the harpsichord with two manuals. Prepared for the comfort and enjoyment of music lovers by the Court Composer for the king of Poland and Selected Governor of Saechs, Director of Music in Leipzig von Johann Sebastian Bach. Published by Verlegung Balthasar Schmids in Nurnberg.

The year of composition

In the Leipzig era, in 1741 or 1742 of Bach's matured age.

In the same era, the Well-Tempered Clavier volume 2 was published. But this volume No.2 contains works which were composed some more years before, so they were not completed in a limited term with concentration. Moreover, at that time " the Art of Fugue" was been under composition parallelly with the Goldberg. It means more.

The version of scores for the analysis of music

The version by Ralph Kirkpatrick
The scorebook with a note "on September 15th, 1934 in Salzburg" on the cover was translated into Japanese and published by "Zennon" a Japanese publisher. This score contains detailed descriptions and explanations by Ralph Kirkpatrick himself, and of course the original score and with the second scores with note ornament's descriptions "how to play actually these ornaments". So, amateurs can trace the way to the actual performance. Without this scorebook, I could not have imagined how to analyze "the Goldberg" by myself.

The version of Brightcop; the version of Bach's Association

The association which was established in 1850 to publish the works of J.S.Bach is The Bach's Association. It had published 46 volumes until 1900 with the help of a publisher Brightcop & Heltel. This is called "Bach's Complete Works".

Other versions

In ordinary scores for piano lessons, original parts are eliminated and there are wrong interpretations concerning ornaments as if these were genuine. We must be careful about it.

Scores resembled to the original may have difficulties. Because there are only original styles of ornaments. Novices cannot study using these scores for they don't know what mean ornaments.

(b) Instruments and tuning @

The original score specifies the cembalo with two manuals. As the score indicates you are going to play, you need a cembalo with two manuals. There are some parts where two hands are physically overlapping. Glenn Gould Played the Goldberg on Video program in 1981. He played piano but his two hands were moving on the keyboard freely enough like using two manuals. When piano is used for performance, some parts are usually transposed one octave higher or lower, or some notes are eliminated.

Naturally tuning was not specified to be well-tempered. There is no problem in using well-tempered piano. Other tuning may create a better taste.

(c) The expression on the score @

(Click Name or , then music will start.)

(all at once midi or mp3 )
Aria midi mp3 G major3/432
Variation 01 midi mp3 G major3/4322One
Variation 02 midi mp3 G major2/4323One
Variation 03 midi mp3 G major12/8163Onecanon all Unisono
Variation 04 midi mp3 G major3/8324One
Variation 05 midi mp3 G major3/4322One or Two
Variation 06 midi mp3 G major3/8323Onecanon alla Secunda
Variation 07 midi mp3 G major6/8322One or Two
Variation 08 midi mp3 G major3/4322Two
Variation 09 midi mp3 G major4/4163Onecanon alla Terza
Variation 10 midi mp3 G major2/2324OneFuguetta
Variation 11 midi mp3 G major12/16322Two
Variation 12 midi mp3 G major3/4323Onecanon alla Quarta
Variation 13 midi mp3 G major3/4323Two
Variation 14 midi mp3 G major3/4322Two
Variation 15 midi mp3 g minor2/4323Onecanon alla Quinta
Variation 16 midi mp3 G major2/2-3/848 OneFrench Overture
Variation 17 midi mp3 G major3/4322Two
Variation 18 midi mp3 G major2/2323Onecanon alla marcia
Variation 19 midi mp3 G major3/8323One
Variation 20 midi mp3 G major3/4322Two
Variation 21 midi mp3 g minor4/4163Onecanon alla Settima
Variation 22 midi mp3 G major2/2324OneAlla breve
Variation 23 midi mp3 G major3/4322-4Two
Variation 24 midi mp3 G major9/8323Onecanon alla Ottava
Variation 25 midi mp3 g minor3/4323TwoAdagio
Variation 26 midi mp3 G major18/16-3/4323Two
Variation 27 midi mp3 G major6/8322Twocanon alla Nona
Variation 28 midi mp3 G major3/4323Two
Variation 29 midi mp3 G major3/4323One
Variation 30 midi mp3 G major4/4324TwoQuodlibet
Aria da Capo midi mp3 G major3/432One

The Goldberg Variations Complete ----- by midi - by mp3 through TruePianos

(Send me your opinion or impression.)

(d) Key @

G Major or G minor

There are 3 variations with G minor, but aria and other variations are all with G Major. We would like to discuss about the reason why Bach chose mainly G Major for this work.

(e) Range of notes @
The highest note is 1G and the lowest is d3. The range of notes is not so wide because this work is for cembalo. And it extends only about 4 and half octaves. For your information R. Schumann's Kreisleriana uses 6 octaves.
And, because it is originally for the cembalo with two manuals, two hands are overlapping each other in middle range. Some pianists alter the octave.

(f) Composition @

[ repeat ] @
For every aria or variation, the score requires repeats, namely AABB. It is not so normal. In case of the Well-Tempered Clavier, repeats are exceptional.
On the contrary, performers usually don't obey this rule. They choose one of the basic four alternatives, like A-B-, A-BB, AAB-, AABB. The combination depends on each performer's thought. It seems uncommon as far as I know in the classic music world. For your information, [-] of A-B- means omission of a repeat.
First, let's study the performance by Glenn Gould. We can find the big difference between '1955' and '1981' related to the combination of repeats. In 1955 there were 32 A-B-s. In other words, there was no repeat. (In 1959's live performance, there were 30 A-B-s and 2 AAB-s.) In 1981 there were 13 AAB-s. So the playing time of 1981 was naturally a little longer. Anyway there were no A-BB nor AABB.
According to the score, there should be several variations in which the first play and the repeated play are different in the 16th bar. these variations are the 2nd, the 4th, the 6th, the 16th and the 25th. In 1955 Gould ignored these cases and played as A-B-. Some notes written by J. S. Bach were surprisingly omitted.
The performances can be varied depending on repeats' arragement.

[ number of bars ] @

The Aria consists of 32 bars. This number '32' is equal to the number of this total work. The Aria consists of two parts with 16 bars each. Each variation basically consists of the first part with 16 bars and the second part with 16 bars also. Exceptionally the 3rd, the 9th, the 21st and the 30th consist of 16 bars in total, and the 16th consists of the first part with 16 bars and the second part with 32 bars, in total 48 bars.

[ beats ] @

The basic pattern is the Aria's 3/4 beats. There are 15 of 3/4 beats. 3 of 4/4, 2 of 2/4, 3 of 3/3, 3 of 2/2, 2 of 6/8, 1 of 12/8, 1 of 12/8, 1 of 12/16, 1 of 9/8, 1 of 18/16 and 3/4. In case of the 26th variation, the left hand plays 18/16 beats and the right hand plays 3/4. It is very special.

[ canon ] @

There are nine canons at intervals successively from the unison to the ninth, those at the fourth (the 12th variation) and fifth (the 15th variation) in contrary motion, that at ninth without any independent third voice, such as accompanies the others.
Canon is a form of counterpoint method, in which one melody is duplicated and followed by its own clone strictly. The first means that the follower is on the same height, in other words unison canon. So, the eighth canon has a follower melody of one octave high or low. The composition of canon is a highly sophisticated intellectual work.
For your information, in the year 1974, 14 canons (BWV 1087) based on the chord of the Goldberg Variations were discovered in Strasbourg.
Of course, canon is not only of the era of Bach but also the era of Mozart, Beethoven or Schumann, who used often the canon method. Especially Robert Schumann wanted to get mental stability by studying and practicing canons. Schumann thought, "It is very useful to analyze canons and fugues. It will give us the humanistic and moral strength. The works of Bach are never shrinked or morbid. They were composed to have immortal lives." But Schumann did not mention much about the Goldberg Variations. He talked frequently about the Well-Tempered Clavier and other works. If he loved the Goldberg Variations much more and Clara played the arias and variations for his husband, I suppose, he would not have had so painful days suffering from his mental disease. But in the fantasy novel 'Kreisleriana' written by E.T.A. Hoffmann in 1838, a conductor-composer named Johannes Kreisler played a work which might be the Goldberg Variations at its opening. This novel was said to inspire Robert Schumann's piano pieces 'Kreisleriana'.
Arnold Schoenberg, who is credited with the invention of serialism, or 12-tone music, is known as a master of canon. The 17th and 18th of 'Pierrot Lunaire' are the good examples of canon.

(2) The Analysis of performance @

Let's analyze the Goldberg Variations.

(a) Who played ? @

The first and monumental recorded performance was that of Ms. Wanda Landowska in the year 1934. This record was so famous that there was no necessity to have another record of the Goldberg Variations for a long time. In 1955 Glenn Gould told to the CBS Record that he intended to record the Goldberg Variations for his debut album, but one of the executives of CBS opposed this plan, saying, "You don't want to take a goddamned risk to throw away your novice recording pianist career by choosing such an unfavorable, reluctant and intricate work." or "As to the Goldberg Variations, everybody knows that dreadful Wanda Landowska, and she played and recorded with her dreadful harpsichord. Among music freaks, the Goldberg means Landowska."
Who performed the Goldberg? According to the list of CDs which I can get now, in the recording year order, Wanda Landowska, Karl Richter, Gustav Leonhardt, Glenn Gould (re-recording), Ton Koopman, Daniel Barenboim and Mari Kumamoto from Japan recorded the Goldberg Variations. Glenn Gould recorded in Salzburg in 1957. Yerk Demus also recorded but I don't know when. In addition, Tatiana Nikolayeva and Andrei Gavrilov made recordings.
According to the list of concerts of Glenn Gould, he played the Goldberg Variations for the audience 27 times all over the world (except Japan: He never visited Japan.). The first stage performance of the Goldberg Variations was in Ottawa at age of 23. The last and final stage performance of the Goldberg was in Los Angeles in 1961. In those days people could enjoy the Goldberg on stage by Glenn Gould if they wanted.

(b) When did they do recording ? @

Wanda Landowska did the oldest recording of the Goldberg Variations, as far as I know, in 1934. The next oldest recording was surprisingly of Glenn Gould in 1955. The third oldest might be Glenn Gould's 1959 live recording which he intended or not intended, I'm not sure. The next should be Karl Richter's in 1970, 10 years or so passed. After several years passed, Gustav Leonhardt did the recording in 1976. These two great masters did great and characteristic achievements with noble appearance which will remain in history. After a while in 1981 Glenn Gould did astonishing re-recording.
After that, in 1987 Ton Koopmann, in 1989 Daniel Barenboim. But the strong influence of Glenn Gould's Goldberg affected those recordings down. The recent recording of Miss Mari Kumamoto in 1993 gives us a good and natural impression because she had no rivalry towards Gould at all. Miss Mari Kumamoto worships Glenn Gould so much and she is proud of her own experience that she met Gould himself and was given a message from him when she was a student.

(c) Piano or Cembalo ? That is the question ? @

Who played the piano are Glenn Gould, Daniel Barenboim, Tatiana Nikolayeva, Andrei Gavrilov, Mari Kumamoto, as far as I know. Who played the cembalo are Karl Richter, Gustav Leonhardt, Ton Koopmann, Yerk Demus, Wanda Landowska and additionally Keith Jarrett. The choice of instruments totally depends on performers' intention. Glenn Gould chose his favorite one of many Steinway & Sons pianos (C174) for his debut recording. For re-recordig of the Goldberg Variations Gould used a Yamaha's secondhanded piano which he encountered by chance in New York City because his preferably using Steinway & sons (CD318; made in 1945) was damaged during transport and he could not use it.
Gustav Leonhardt used a cembalo made by William Dought in 1975 based on the Branche model cembalo made in Paris in 1730s.

(d) Is it true that the tempo on which Gould recorded in 1955 was terribly fast ? @

Definitely the duration of Gould's performance of the whole of the Goldberg Variations in 1955 is very short and actually 38min 17sec. However he played without repeats, so it cannot be compared simply. If Glenn Gould played with repeats for all, namely, not A-B- but AABB, the duration must be doubled. Then it will be 76min 34sec. Daniel Barenboim required 82min 38sec. (Barenboim played the last Aria with A-B-, all others with repeats. So 2min 46sec was added to 79min 52sec as repeat portion of the last Aria.) There is only about 8% difference. Not great?. Speed of Performance cannot be calculated only by the length of time.
Speed of Performance

(e) How to deal with repeats ? @

Glenn Gould played A-B- only in 1955. In 1981, the number of A-B- is 19 and that of AAB- is 13. There was a great change between 2 versions. In the live performance in 1959, only two variations, the 4th and the 30th, were of AAB-. How to deal with repeats depends on performers.
Karl Richter and Daniel Barenboim played all AABBs except the last Aria's A-B-. Yerk Demus played basically AABBs except the first and last Arias, the 15th, the 21st and the 25th variation which are of slow tempo. Ton Koopmann played strangely all AAB-s except the Aria da Capo's A-B-. Gustav Leonhardt played all A-B-s against anticipation that he should have been somehow stubborn. Miss Mari Kumamoto played freely as a whole. She applied pliantly variety of combinations, and her performance was very attractive.
Wanda Landowska was eccentric. She recorded in 1934, in the era of SPs, so whole variations needed 3 or 4 disks. And at that time there was only one recording. There was no other records, so her performance had exactly a great influential power. But there is something strange. She played basically A-B-s. I thought that it was necessary for her to shorten the playing time. Suddenly at the 4th variation she used AABB, then next variation, the 5th, she played A-B(A/2) a little surprisingly. A-B(A/2) means that after playing the latter half's 16 bars and succeedingly the first half 8 bars of the first 16 bars. This kind of thing happened in the 7th and 18th. Wanda Landowska was seemed to have added her individual touch. I don't think it was reasonable at all because no other performer except her played at that time, and moreover her record was the only the Goldberg Variations for a long time.

Sviatoslav Richter went to hear Zuzana Ruzickova's Goldberg Variations on cembalo in 1975 and took note about it. "It was the first experience for me to listen to this great work by Bach on cembalo. I have been to the concert of it by Glenn Gould and also listened to his record. Someday in future I wish I could play it for myself.... if I could go through to the end. This performer born in Prague played it sincerely. Thankfully she played all the repeats. (You had better not play it at all without designated repeats.)*"
In 1972, after listening to the record of Bach's Partitas by Glenn Gould, Richter wrote, "Gould is the greatest performer of Bach." But he added, "He plays a little bit too brilliantly. In other words, his play is somehow superficial. Moreover he dares not play any repeats. I can not tolerate it*."
After all Richiter never recorded or even played the Goldberg Varitions on any occasion in his life.
(*Bruno Monsaingeon: Richter, Ecrits, Conversations 1998 Editions Van de Velde)

How to deal with repeats

(f) Tempo Allocation @

- Aria
There are more or less two ways of playing the Aria like 'ton -ton -toh -toh -toon' in a regular speed and toon -toon -too -too -tooooon' in a relaxed speed. It can be said that two versions of the Goldberg Variations by Glenn Gould are the typical examples of these two ways. 1 min 53sec for '1955'. 3min 5sec for '1981'. There are substantial difference between two Gould's versions in relation with tempo. These two are of course both remarkably splendid. Almost of all tempos by other performers are between '1951' and '1981'. The version '1981' by Glenn Gould is as perfect as to be described 'filled with tension excelling the feeling of comfort or looseness.'
Only comparing the durations of play, 1 min 53 sec of Gould's '1955' is about half of 4 min 16 sec of Mari Kumamoto's '1933'. In case of Gould's '1955', the Aria will start and end while saying 'Ah'.

fast------------( converted on the basis of 'without repeats' )

----1'53" Glenn Gould '1955'
----2'08" Mari Kumamoto '1993' Hoowever AABB requires 4 min 16 sec
----2'17" Wanda Landowska '1934'<
----2'39" FujitaSMF '1992'

----3'05" Glenn Gould '1955'
---(4'16") Mari Kumamoto '1993' ( 4 min 16 sec )


*About the 25th variation

Gould's '1981' (6min 3sec) was a little longer in duration than that of '1981' ( 6 min 29 sec ). Later Gould confessed that he played the 25th variation too sentimentally like playing F. Chopin's works.

To be continued...