Goldberg Tempo Variations

I like charts. And I like Bach’s Goldberg Variations. Here is a chart of tempos taken from six prominent recordings of the Goldberg: Glenn Gould’s landmark 1955 recording, Gould’s 1959 live recording, Gould’s 1982 recording, Murray Perahia’s 2000 recording, András Schiff’s 2001 recording, and Angela Hewitt’s 2000 recording.

Click to embiggen

The detailed data is given in the table below. Naturally, performance tempos are rarely maintained with any kind of precision throughout an entire movement. These tempo readings, therefore, reflect simply the first steady tempo established in each movement, the tempo that is generally maintained through a good portion of its opening. For those movements which are too slow to effectively determine a tempo and/or which are played with enough rubato to make calculation difficult, tempo readings were determined by dividing the number of beats played by the track’s duration. Asterisks (*) in the chart below mark tempos derived in such a way. Tempos in bold indicate the quickest pace for each movement; markings in italics, the broadest. For the bipartite 16th variation, the Overture, two tempo markings were registered.

Movement	Gould 1955	Gould 1957	Gould 1982	Perahia 2000	Schiff 2001	Hewitt 2000
Aria	50*	53*	31*	48*	54*	47*
Var. 1	137	121	83	104	103	114
Var. 2	111	95	79	81	96	79
Var. 3 (Canon 1)	72	60	63	68	66	57
Var. 4	69	61	58	59	62	62
Var. 5	173	154	166	144	131	129
Var. 6 (Canon 2)	67	73	73	48	50	38
Var. 7	63	59	55	75	86	64
Var. 8	140	124	111	109	118	105
Var. 9 (Canon 3)	104	96	97	60	89	71
Var. 10 (Fughetta)	99	99	94	86	91	85
Var. 11	77	67	74	78	67	63
Var. 12 (Canon 4)	115	105	92	90	104	77
Var. 13	52	55	42	42	46*	43*
Var. 14	110	99	96	92	103	93
Var. 15 (Canon 5)	29*	26*	19*	30*	34*	30*
Var. 16 (Part 1)	33	34	26	40	37*	28
Var. 16 (Part 2)	93	82	69	76	72	71
Var. 17	118	111	109	121	92	90
Var. 18 (Canon 6)	92	96	94	96	104	94
Var. 19	146	130	90	130	150	122
Var. 20	141	118	118	107	117	109
Var. 21 (Canon 7)	43	49	47	52	71	50
Var. 22	102	96	93	90	85	94
Var. 23	111	103	103	105	94	95
Var. 24 (Canon 8)	108	103	89	82	90	68
Var. 25	15*	23*	16*	26*	28*	25*
Var. 26	114	111	112	99	94	97
Var. 27 (Canon 9)	85	75	73	80	86	68
Var. 28	94	98	92	93	68*	87
Var. 29	102	103	94	86	108*	85*
Var. 30 (Quodlibet)	91	80	73	82	108	70
Aria	45*	48*	26*	44*	54*	41*

In 13 of the 33 movements (counting Variation 16 as two separate movements), Angela Hewitt provided the broadest tempo among the six performances. Eliminate Gould ’82 from the chart and that number rises to 18. In fact, Hewitt’s rendition as a whole is easily broader than Gould’s notoriously broad ’82 recording. Averaging out all the tempo readings in Hewitt produces an overall average of 74.3 bpm compared to a 77.5 bpm average in the Gould.

It may come as little surprise to those who know Gould’s ’55 recording that it represents the briskest recording of the six with an average of 90.6 bpm, registering the quickest tempo in 16 of the 33 movements. (Hewitt provided the briskest tempo in none of the movements.) What I did find a bit surprising is that while I expected Gould to provide the most metronomic performances (and thus the easiest to do a tempo calculation), this was not at all the case. Easily the most metronomic of these performances is by Hewitt. One almost gets the sense that she practices hours on end with metronome on. Natural as it may be that András Schiff’s ultra-dynamic pianism made tempo calculation in his performances most difficult, I was still surprised at just how elastic his tempi can be.

Calculating tempo averages from quickest to broadest puts Gould ’59 just under Gould ’55 at 85.1 bpm followed by Schiff at 83.6 bpm, Perahia at 79.5, Gould ’85 and Hewitt.

I could go into further mundane details about the chart, but I will refrain and let the reader explore those details. Readers interested in hearing more about tempo choices in Gould’s 1982 recording might want to listen to the pianist’s (entirely scripted) interview with Tim Page conducted at the time of the recording’s release. The interview took place just months before Gould’s death.