I mean Western music, classical music, circle of fifths and all that. Why can’t I have a key with, for example, three sharps -- though not the usual F, C, and G, but instead some other combination like E, B, and D? I imagine it would sound terribly dissonant. I have heard that the reason there is the conventional progression of sharps/flats (i.e. the circle of fifths) and that there cannot be other strange combinations of sharps/flats is because the keys are “built on each other”. However, I’m not sure what that means exactly.

I have a very limited understanding of music theory, and occasionally I have dabbled in (piano) composition and made a few amateur sketches. It is always the case that I find myself “gravitating” toward a certain key in any given attempt. I can never seem to, for example, make a sensible or at least not-terrible-sounding sketch without being in some conventional key, and if I want to introduce anything different I would simply notate accidentals in individual measures (but that would usually only be temporary, e.g. a single one-note dissonance). So what I am wondering is, is it possible to actually compose a solid, structural, perhaps even melodious piece strictly with an unconventional combination of accidentals in the key signature, and where individual measures or segments are not in some conventional key signature (which seems to usually be the case, e.g. a piece may be said to be “in C major” but can be in other keys at different moments), which would usually achieved by introducing accidentals? (I mean, one could put B as the only sharp in the key signature, but then simply add a natural to each B throughout the piece… so effectively it’s just C major.)

Could a key with unconventional accidentals make sound, structural, perhaps even harmonic and melodic sense... a key that is perhaps separate and distinct from the conventional key signatures, and maybe even with a different "quality", "temperament", or "feel"?

  • 1
    Unconventional key signatures are possible, see for instance Ligeti. They might make sense in some cases, but might also confuse the musician.
    – Karlo
    Commented Aug 16, 2016 at 11:18
  • Here's just simple proof of existence: youtube.com/watch?v=zSugkgY-tKA
    – nonpop
    Commented Aug 16, 2016 at 18:46

6 Answers 6


As computer scientist I hold, that key signatures are intended as help to reduce the amount of information to be processed at each note, so a piece in G flat major looks quite similar to one in C major. They are a convenient convention, but there is no strict requirement to use key signatures and you could ornament each note with the appropriate accidental instead. (If all F notes receive a sharp, this would be an indication, that you have at least G major, however).

I remember my son received a piece (20th century) to play as prima vista exam, which had the quite unusual signature of one sharp and one flat, so yes, strange pieces exist.

  • Could that piece be, as I suggested, in D minor?
    – Tim
    Commented Aug 16, 2016 at 7:08
  • 1
    @Tim, probably Bartók, and probably (non-diatonic) modal as all Hell.
    – user16935
    Commented Aug 16, 2016 at 13:14

I think there are a couple of aspects of this question - there's the issue of what notes you can choose for your piece and still have it sound harmonious, and there's then the issue of how you notate that.

To tackle the second question first, western music notation (to an extent) assumes use of the diatonic scale - witness the fact that it has 7 notes per octave, as opposed to the 12 notes per octave that might make more sense if everyone were wanting to explore the full range of possibilities within the chromatic scale. All the conventional key signatures are those that comply with this assumption. You could use non-conventional key signatures, but you could argue that you are then going against some of the assumptions of the notation system in the first place. (Of course accidentals and other tactics can be used to make much more flexible use of standard notation, but that doesn't negate the fact that the system takes these assumptions as a starting point.)

As to whether a piece with an unconventional key signature could sound fine, there's no real reason to think not. A piece using the whole tone scale (starting nominally on C) could be said to use an F#, G#, A# key signature, for example. In fact any scale that goes outside the diatonic (e.g. I think some middle eastern and Indian scales, and arguably even blues) is not going to be able to be notated accurately using only the conventional key signatures - It's just that the conventional way round this tends to be the use of accidentals, rather than using an 'odd' key signature.


Your use of 'gravitate' is the clue. Most Western music has this propensity. It gives the listener a feeling of stability. In fact, the tension and release aspect of a lot of music uses this phenomenon to work. Notes and harmonies that put an edge on the music do the same for the listeners' feelings.

Over many centuries, the keys, their scales, and the distance between each of the notes thereof established themselves - rather than 'humans made it happen', if that makes sense. All we did was formulate what was already there, as accepted form.

Yes, one could write a piece with a strange key sig., but that wouldn't help musicians who are reading it. The key sig. is a sort of code, which gives the reader a clue as to what will be coming. Some pieces in, say, G, with one sharp, could actually sound in the key of Em, somewhat different, but a reader would spot the D# that would often appear. Sometimes, in modal compositions, the 'proper' key sig. of the note name appears, as in, say, one flat for Dm, but if the piece is actually in D Dorian, then all of the Bs get naturalised. This again is a clue for the reader.

I sometimes used to think o.k. this piece is in D minor, which usually has Bb and C#, so why not use those as the key sig...

If the key sig. was, say, B#, as you suggest, and then all the Bs were naturalised, then the process would be pointless.


It's very common in traditional musics to have 'non-standard' key signatures; Swedish traditional fiddle music, for example, has many examples, picking one of my books of Swedish tunes and flipping through quickly yields key signatures of F# Bb , and C# (only). Music based on modal scales also pushes notation out of the 'standard' key signatures, for musics such as Breton, Hungarian, etc.

Western classical music has imposed the idea of a logical progression of ever-increasing sharps or flats; out in the 'real world' there's all sorts of other key signatures (or, more correctly, scales depicted in standard notation as key signatures) than that.


It was very much the thinking in your last paragraph, @fioritura, which led me in 1993 to use non-standard key signatures in a book of Swedish fiddle tunes I was putting together. I also had a bit of an ideological point, I think — conventional notation evolved in response to the needs of certain musicians and musics, and doesn't necessarily meet the needs of others.

In retrospect, though, I think it was a mistake. For people who are not fluent and habitual readers, it probably makes very little difference, but for anyone who reads all the time it's a stumbling block — something like a spelling reform, where the change might make things slightly easier to people learning to read but much harder for those who already read well.


I believe that musical notation has to be seen as one kind of message. You, as the composer writes the message. The notation is the way to transfer the information. The musician interprets the notation, and hopefully comes close to what you the composer wanted it to sound like.

As such you have to consider what the musicians are used to read, to help them on the way. Classical musicians spend most of a lifetime reading notation following a set of rules. So if your message, sorry music, can be massaged to fit into this form it is a good thing to follow the rules (real rules as well as unwritten ones). But then again, if you cannot convey the music using conventional notation, then use other ways. If the music requires it, do write it differently. You can find examples of this if you look around a bit.

But only using a non-conventional set of sharps and flats in the key, to mind mind, really does not convey anything you could not convey following the conventions. But it might, only the result of letting the musicians playing the notation will really show.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.