Play a G7, then play a D♭7. Two chords whose roots are as far apart as two roots can get, and they can stand in for each other in the middle of a ii-V-I without the progression losing its way.
The usual one-line explanation is that they share a tritone. That is true, and it is not quite the interesting part.
Which two notes, and what happens to them
G7 is G, B, D, F. Its third is B and its seventh is F.
D♭7 is D♭, F, A♭, C♭. Its third is F and its seventh is C♭ — which is the same key on the piano as B.
So the two chords do not merely both contain a tritone. They contain the same two keys, and the roles are swapped: what is the third in one is the seventh in the other. That symmetry is exactly why the substitution holds up. The pair of notes that makes a dominant chord sound like a dominant, and that leans hard towards a resolution, is sitting there unchanged. Only the note underneath it has been replaced.
It is also why a tritone is the one interval this works with. A tritone splits the octave exactly in half, so inverting it gives you another tritone — the interval is its own mirror. No other interval in the system does that, and no other substitution gets this for free.
What actually changes
If the guide tones do not move, something else has to be doing the work. Two things are.
The bass line. A ii-V-I in C runs D, G, C — two leaps of a fourth. Substitute and it runs D, D♭, C: three notes descending by semitone. It is the most audible consequence of the swap, and Levine notes that bass players are fond of exactly this chromatic motion. A descending semitone line is a strong shape, and a tune can acquire one for free.
What the melody note becomes. This is the underrated half. Hold a G in the melody over an F7 and it is the ninth: warm, unremarkable, exactly where you would expect a ninth to sit. Now substitute B7 for the F7 — the tritone partner — and that same, unmoved G is the flat thirteenth of the new chord. (Respelled as F𝄪 it is B7’s augmented fifth; you will see it written both ways and the sound is identical.) You did not touch the melody. Its meaning changed underneath it.
That reinterpretation is a large part of what makes the device worth having when you are arranging. It lets a melody note sound like a decision rather than an inevitability, without rewriting a single note of the tune.
The hand that does not have to move
Here is the practical consequence, and it is the reason this technique feels different at the piano than it looks on paper.
Take the rootless A voicing of G7 — the third, the thirteenth, the flat seventh and the ninth, which in G is B, E, F, A, with no G anywhere in your hands because the bass player has that covered. Now leave your hands exactly where they are and have the bass play D♭ instead of G.
Count what those same four notes are doing over the new root. Respell them the way the new key wants them — B becomes C♭, E stays E, F stays F, A becomes B𝄫 — and they are the flat seventh, the sharp ninth, the third and the flat thirteenth of D♭7. You are playing an altered D♭7, an idiomatic and fully specified altered dominant, and you have not moved a finger.
The respelling is the same trick as C♭ and B at the top of this piece, and it is not bookkeeping for its own sake. It is the whole mechanism: one set of keys, two sets of letter names, two chords.
That is also the most useful way to hold the technique in your head at the piano. Not as a chord change, but as a bass note that reinterprets a voicing you were going to play anyway — which is why altered dominants and tritone substitutes so often turn out to be the same idea approached from either end.
One tritone, two homes
There is a consequence of the symmetry that is easy to miss and worth having.
If B and F belong equally to G7 and to D♭7, then that pair of notes is not pointing at one destination. It is pointing at two. G7 resolves to C: the B rises a semitone to C and the F falls a semitone to E. D♭7 resolves to G♭: the F rises a semitone to G♭ and the C♭ — the same key as the B — falls a semitone to B♭. Same two keys under your fingers, resolving in opposite directions, arriving at two tonics a tritone apart.
Which is to say the tension does not know where it is going until you tell it. The diminished family has the same property in a more extreme form, and for the same reason: an interval that divides the octave evenly cannot single out one key to belong to.
That is why the substitution is also a modulation device and not only a decoration. Set up the tension, then choose the other exit.
The bebop version: substitute the whole pair
The simple form leaves the ii chord alone and swaps only the V. D-7, G7, Cmaj7 becomes D-7, D♭7, Cmaj7.
The bebop generation went further and gave the substitute V a ii chord of its own. Schematically, D-7, G7, Cmaj7 becomes A♭-7, D♭7, Cmaj7: a complete ii-V a tritone away from the original, resolving to the same home chord.
It can also arrive as an insertion rather than a replacement, and the difference is worth seeing. Charlie Parker, Bud Powell and Thelonious Monk all wrote this way, and Powell does it in “Dance Of The Infidels”: where a listener expects G-7 into C7, he plays G-7, then D♭-7, then G♭7. The original G-7 is still there. The G♭7 is the tritone substitute for the C7, and the D♭-7 is a new ii chord built to introduce it — inserted in front, not swapped in.
Be precise about what survives that move, though: only the dominants keep their guide tones. C7 and G♭7 share their third and seventh, enharmonically, exactly as G7 and D♭7 do. G-7 and D♭-7 share nothing at all — not one common tone. The ii chord is not substituting on the strength of a shared interval; it is there because a dominant sounds better with its own ii in front of it, and the new dominant wants a new one.
Which is worth knowing before you use it, because it means the two halves of the substitution are doing different jobs. The dominant swap is the sleight of hand. The minor chord in front of it is just good manners.
Two ways it goes wrong
The substitution always works arithmetically, and that is the trap.
It can wreck a bass line. Not every tune wants a chromatic descent at that particular bar; some have a bass figure doing something specific, and dropping a semitone slide into it flattens the thing that made the passage recognisable.
And it can collide with the melody. The reinterpretation that turns a ninth into a flat thirteenth is a gift when the melody note is passing through. When the melody is sitting on that note, held, at the top of a phrase, you have just turned the most exposed note in the bar into a sharper altered tension than it was under the original harmony.
Neither is a reason to avoid it. Both are reasons to check before using it, and checking takes about two seconds once you know what you are checking for.
What to practise
Play the ii-V-I in C twice — once as D-7, G7, Cmaj7, once as D-7, D♭7, Cmaj7 — and do not analyse. Just listen to the bass. Then do it in three more keys.
Then the important exercise: hold a rootless G7 voicing and move only the root beneath it, G to D♭ and back, until the reinterpretation stops being a fact you know and becomes something you hear. That is the drill that turns this from a substitution you apply into a relationship you can see on the keyboard, and once it is there it never goes away.
Only after that should you try the full ii-V version. It is a stronger effect and a bigger commitment, and it is much easier to place well once you can already hear what a single substituted bass note does.
Questions people actually ask
- Can I use a tritone substitution on any dominant chord?
- Mechanically, yes — every dominant seventh has a partner a tritone away and the guide tones will always line up. Musically, no. Levine's own caution is simply to use taste rather than reaching for it automatically, and the two things worth checking first are these. The bass line: where the dominant was resolving down a fifth, the substitution turns that leap into a half step, which is usually an improvement and occasionally destroys a bass figure the tune depends on. Where it was going somewhere else, work out what the new bass note actually does before committing to it. Second the melody: the note that was a comfortable ninth over the original chord may become something much sharper over the substitute, which is sometimes exactly what you want and sometimes a collision.
- Is the tritone substitute the same chord as an altered dominant?
- They are very close, and on the piano they are often the same handful of notes. Play a rootless voicing of G7 — the third, fifth, seventh and ninth, with no G in it — and then have a bass player put a D♭ underneath instead of a G. Those same four notes read as the seventh, flat ninth, third and flat thirteenth of D♭7: an altered dominant. The two ideas are one relationship seen from either end.
- Why is it also called a flat-five substitution?
- Because the substitute chord's root is a tritone above the original's, and a tritone above any note is also the flattened fifth of that note. D♭ is the tritone from G and it is also G's flat five. The two names describe the same move from different directions.