Bach’s mysterious Sarabande

While learning and learning about the Prelude to Bach’s G minor Lute Suite, I also came into contact with the suite’s Sarabande. This piece is famous among music theorists, because while it’s only forty measures long, those forty measures are action-packed, harmonically speaking. Here’s a performance by Evangelina Mascardi.

I appreciate that Mascardi doesn’t play it with too much melodrama or rubato. Note that, like most lutenists, she’s using Baroque tuning, so it sounds like she’s playing in F-sharp minor.

This is pretty weird! Let’s look just at the first measure, the opening five notes. There’s D and B-flat, the fifth and third in G minor. There’s an F-sharp, the leading tone in G minor, which now makes you realize you might have just heard a D augmented chord. There’s a G, which resolves the tension of the Daug or whatever it was. And then the longest and lowest note in the measure, which feels like its conclusion, is… an E-flat? What? This note does actually make sense when you hear the next measure, which outlines the rest of a C minor chord, but still. The whole piece is like this, full of tense notes on unexpected beats. It does all make sense within the rules of Western tonal theory, but you sometimes have to make some imaginative effort to understand how.

The G Minor Lute Suite is adapted from the C Minor Cello Suite. The cello version is even weirder and more ambiguous. While the lute version has some skeletal harmony, the cello version is just single lines. Here’s Mstislav Rostropovich’s performance, which was my first exposure to the piece.

None of Bach’s original tempo markings for the Cello Suites survive (if he even wrote any to begin with.) Sarabandes are customarily slow, but “slow” means a lot of different things. Rostropovich plays so slowly that I lose track of the meter entirely in some places. This is probably the desired effect. Rostropovich wants the piece to sound abstract, and he approvingly compares it to contemporary classical music.

So what is a sarabande anyway? In the Grove Dictionary, Richard Hudson and Meredith Ellis Little explain:

One of the most popular of Baroque instrumental dances and a standard movement, along with the allemande, courante and gigue, of the suite. It originated during the 16th century as a sung dance in Latin America and Spain.

Here’s a fun fact:

The zarabanda was banned in Spain in 1583 for its extraordinary obscenity, but literary references to it continued throughout the early 17th century in the works of such writers as Cervantes and Lope de Vega. From about 1580 to 1610 it seems to have been the most popular of the wild and energetic Spanish bailes, superseded finally by the chacona, with which it is frequently mentioned. The dance was accompanied by the guitar, castanets and possibly other percussion instruments, and by a text with refrain.

My old friend the chacona!

J.S. Bach composed more sarabandes than any other dance type. His 39 surviving sarabandes are all virtuoso pieces in suites for a solo instrument (keyboard, cello, flute, violin or lute) except for the one in the Orchestral Suite in B minor BWV1067. They display a rich variety of techniques and styles, including variations or written-out doubles (BWV808, 811 and 1002), elaborate, dramatic italianate flourishes (BWV806, 828 and 1007), entrée grave style (BWV829 and 1010) and even strict canon at the 12th (BWV1067). Sarabandes sometimes occur, though untitled, in other works, such as his chorale prelude An Wasserflüssen Babylon (BWV653), the aria to the Goldberg Variations (BWV988) and the final chorus of the St Matthew Passion (BWV244).

It is difficult for me to hear a connection between the sarabande’s origins as a wild (and wildly obscene) dance genre and the intense solemnity of the Sarabande from Bach’s Lute Suite. There are lots of videos online of people dancing Baroque sarabandes, and their vibe is always tightly constrained.

Back when the sarabande had more of a groove, what might it have sounded like? Jon Silpayamanant disputes the Grove Dictionary’s assertion that it had a guitar-and-castanets flamenco sound.

Yeah, you're probably not going to find those because they likely don't exist. Castanets weren't really used in the New World during the zarabanada dance era. They were more of a Spanish thing & Spain banned zarabandas in 1583 (not that banning music really ever stopped anyone).

— Jon Silpayamanant โจนาทาน ศิลปยามานันท์ | Mae Mai (@Silpayamanant) August 21, 2022

In the absence of any hard data, I’m forced to speculate. Here’s a triple-meter groove on castanets by Larry Salzman; I have no idea whether it sounds like a historical sarabande, but it sounds great.

I used Salzman’s castanet groove as the basis for my remix of Thomas Dunford’s lute recording. I don’t know whether this has any historical authenticity at all, but I like how it sounds, and it certainly clarifies my hearing of the piece.

The form of the Sarabande is a simple AABB: an eight bar A section that repeats, and then a twelve-bar B section that repeats. Sometimes people add embellishments and curlicues on the repeats, and sometimes they just skip them. Here’s my analysis.

The first half of the A section is in G minor. Its melody forms a distinctive pattern: a descending three-note arpeggio that lands on a leading tone, then the leading tone’s resolution, then a long bass note that points to the harmony in the next measure. The second half of the A section, mostly in B-flat major, takes the basic melodic shape from the first half, turns it upside down, and adds an extra note to the end.

The B section follows the same basic melodic plan as the A section, though the harmony is more complicated. The first four bars of the B section use the same melodic shape as the first half of the A section. The second four bars of the B section use the same melodic shape as the second half of the A section. The final four bars of the B section combines ideas from the first and second four bars.

Daniel Ketter’s analysis is much more historically accurate than mine, with nice graphical analysis of the melodic motives.

Harmonically, the weirdest thing that happens in the Sarabande comes in measure 23. The first two notes are a G and an E-flat over a B-flat in the bass. This isn’t mysterious; they sound like Gm/Bb and Eb/Bb respectively. The last two notes in the bar are F-sharp and G, implying D7 and Gm, also not very mysterious. But the middle two notes are C-sharp and D over an E-flat in the bass. These two implied chords are setting up the D7, but what the heck are they? In my analysis I labeled them in a simple-mindedly literal way as Eb7 and Ebmaj7, but that makes no sense. I guess Eb7 is a predominant augmented sixth chord? That would make sense with its spelling. Maybe the D is just anticipating the coming D chord? It sure does rub hard against the E-flat! If you are a Baroque music specialist with a legitimate explanation of what is happening here, please enlighten me in the comments.

The second weirdest thing in the piece is in measure 17, the first bar of the B section. The bar starts innocently enough: an arpeggiated F chord resolving to Bb. But the next note is an A-flat an entire ninth below. Is this meant to be a Bb7 chord with its seventh in the bass? Such a thing is not unheard of in Baroque music, but people don’t usually leave so much empty space. If the chord really is Bb7/Ab, the logical thing to come next would be Eb/G, but there are no more Eb chords in entire rest of the piece that I can detect, and there isn’t a G in the bass until the end of measure 19. The A-flat in the bass is probably supposed to be part of the B°7 arpeggio in measure 18. (I labeled it G7b9/B, but Bach would have thought of it as a diminished chord.) This kind of ambiguity is a big part of my attraction to the Sarabande.

People are starting to quote my blog in legitimate academic sources, so I want to be crystal clear that my analysis of Bach’s harmony is not historically accurate, or accurate period. When Baroque music specialists like Daniel Ketter do legitimate analyses of the Sarabande, they hear many fewer chords than I do. Bach wasn’t thinking about chords at all, really; he was writing linear contrapuntal melodies, and the vertical harmonies were just whatever happened to emerge out of those melodies’ interactions. Bach was probably thinking in general terms about tonic and dominant, but he certainly would not have approved of my labeling every single note in the Sarabande as a freestanding harmonic entity. My analysis is what I imagine myself to be hearing, not what Bach intended me to be hearing.

I analyze Bach with jazz chord symbols because it’s an effective memory aid for me as I struggle to play the notes on the page. For better or for worse, the foundation of my harmonic understanding comes from block chords on the guitar, and my understanding of the linear relationships in counterpoint is limited mainly to vibes. So please do not cite this post in your term paper or anything. My intention here is not to explain how Bach’s music “really works”. My intention is to look for ideas and inspiration for my own jazz-centric concept of harmony. This approach has apparently been useful and interesting to several people other than me, and I’m glad of that. But if you want to know what Bach was thinking, look elsewhere.