Duke Ellington (1899-1974) knew something about elegance in the social sense. But one of his signature tunes, Satin Doll—perhaps partly (or even entirely) the work of long-time Ellington collaborator Billy Strayhorn—also exhibits great musical elegance in the economy and ingenuity of its musical structure. Let’s take a closer look at an eight-bar excerpt. Here it is, notated somewhat schematically:
It’s not hard to see that the first two measures are repeated, transposed up a whole step, in measures three and four, nor that the harmonic rhythm—the rate at which chords change—speeds up in the second line.
Or does it? Looking again, the chords continue to change every two beats, so at the surface level, the harmonic rhythm does not actually change. Yet the harmonic rhythm feels as though it accelerates. Tabulating the changes can make the reason for this perception clearer. Let’s make our table in two-measure segments:
Mm 1-2: Dm7 G7 (2x)
Mm 3-4: Em7 A7 (2x)
Mm 5-6: Am7b5 G7 Abm7 Db7
Mm. 7-8: C (plus turnaround, not shown)
In bars 5 and 6, there is twice as much harmonic information, since the repetition of the first two chords is eliminated and two new chords substituted.
Moreover, there’s a clear logic governing the specific chords chosen. To understand it, we need to understand the concept of the “ii-V” progression in jazz. (For readers who don’t know this numerical terminology for chords, the “ii” denotes a chord built upon the second degree of a scale and the “V” a chord built upon the fifth degree. For example, in A major “ii” would be a B minor chord, “B” forming the second degree of the A major scale, and “V” would be an E major chord, since “E” is the fifth scale degree.)
The “ii-V” progression is very common in all “common-practice” music, as easy to find in Bach as in Ellington. But in jazz it is particularly useful because it so clearly defines a key center, and creates such a strong sense of harmonic motion. In fact, it has become a stock element, used ad lib, for example, as an insertion at the end of a repeated section to lead back to the beginning of that section—a “turnaround.” This ‘stock element’ becomes the basic unit used to construct Satin Doll. Here’s our table of chords, redone in terms of the key to which each of the “ii-V” progressions belongs:
Mm 1-2: ii-V, key of C (2x)
Mm 3-4: ii-V, key of D (2x)
Mm 5-6: ii-V, key of G(minor); ii-V, key of Gb
Mm. 7-8: NA (cadence chord in C)
Ellington—or Strayhorn—begins with a “ii-V” that defines the key as C major. We aren’t given the tonic triad, C major, which helps to sustain the harmonic energy at the beginning of the phrase—a tonic chord always tends to act as a ‘ground’ for that energy when heard, and we don’t want the phrase ‘short-circuited.’
The upward transposition in mm. 3-4—we noted that above—then suggests a secondary key area of D. (Tonicization of “ii” (or “II”) is a strategy that Franz Schubert was rather fond of, by the way; it can be observed in the ‘Unfinished’ Symphony, among other works.) Again, there is no tonic function actually heard.
Measures 5-6 give us a new implied tonality, G. Notice that this makes a larger scale “ii-V?” The implied “D” of mm. 3-4 would be “ii’’ in the home key of C, while the “G” of 5-6 would be “V.”
The “minor seven flat five” chord at the beginning of measure 5—also known as a half-diminished seventh—suggests a minor, not major key, since it fits naturally into the minor mode, unlike the minor seventh chords which heard in previous versions of the “ii-V.” I suspect, though, that the chord quality has more to do with preparing what we hear next than with the niceties of chordal implication.
For what we get next is tonally surprising. As the table shows, the final “ii-V” seems to belong to the tonally remote key of Gb major. Following the established tonal logic, we might have expected a “ii-V” in the tonic key. It would have made a logical, convincing cadence. Indeed, it’s not hard to rewrite the tune to do that:
But it would be a disappointing ending, at least in comparison with the ‘real thing.’
It’s easy to understand why this is a musical ‘plot twist’—suddenly we find ourselves in a distant tonal area. It’s perhaps a bit less obvious why it still makes sense.
Jazz, as an art for which improvisation is highly important, and in which individual expression is prized, has developed many techniques by which common (as in ‘shared’) melodies and lines can be expressively inflected by improvisors and arrangers. One is the concept of chord “substitution.” As I’ve explained elsewhere, a common form of substitution involves using third-related chords—for example, an A minor triad might be used in place of a C major triad. This has a certain logic in that two of three tones are common to both chords, yet the effect can be surprisingly fresh.
A bit more exotic is the “tritone substitution.” It, too, goes back to earlier antecedents—it can be found, for example, in Chopin. It consists of substituting for a dominant seventh chord—say, G7—the dominant seventh a tritone away—say, Db7. This may seem arbitrary at first blush, but makes sense upon reflection: the two most active tones in the chord, the third and seventh, are actually common to both chords (if we neglect a little enharmonic respelling.) For example, G7 has for a third “B” and for seventh, “F.” For Db7, the tones switch: “F” is now the third, while “B”—respelled as Cb—becomes the seventh.
In Satin Doll, then, Ellington has applied the tritone substitution to the entire “ii-V” progression, not just the dominant seventh itself. It’s the perfect musical plot twist, as logical as it is unexpected.
The ‘modular construction’ we’ve been looking at also reminds me in a way of Andy Warhol’s pop-art work: everyday objects—for the “ii-V” is as everyday as a soup can—arranged artfully enough become art themselves once again, momentarily liberated from the banal patina of overexposure.
But Ellington did it first, did it less obviously, and for my money had more fun with it—with Ellington, it’s ingenuity, not irony.
Update: The text above has been reflected to correct an error–originally, the last ii-V was identified as being in Db. But, of course, Db7 can’t be its own dominant, and the correct key is Gb–a tritone from the home key of C. That’s the error I should have caught and corrected in my original reply to Jaypee, below!