An accidental usually only lasts until the end of a measure, but what if there are no measures? I'm writing some music without bar lines (blame Satie) and am wondering what to do about notating accidentals, and how people will read accidentals in the absence of bars. As far as I can see it there are three options:

  1. Accidentals last until the end of the measure. There is no end of the measure, so accidentals last forever.
  2. Accidentals only apply to an individual note.
  3. Accidentals last for "about a bar".

1 seems logical but could get confusing in a longer piece and would lead to lots of naturals to cancel accidentals. 2 could be more workable but might lead to lots of repetition of accidentals near each other. 3 would be asking for trouble.

Is there a convention for this, or do I need to explicitly state what rules I'm following?

I mainly write with pencil and paper, but I'm aware that when I come to put the score into the computer I may have to cheat and have bars with invisible barlines or something, so a rule that can be implemented easily in MuseScore and/or Lilypond/Frescobaldi would be preferable.

  • 3
    Of course, you could always put in bar lines (with as many meter changes as you like), and add instructions that downbeat emphasis is to be avoided. Commented Oct 9, 2019 at 14:35
  • After a bit more thought, and some unexplained dvs, what would happen when a note was sharpened, then later naturalled. Would there be a need to continue with natural signs or not..?
    – Tim
    Commented Oct 9, 2019 at 14:48
  • As a choral singer I've seen lots of very old music, originally written without bar lines, with modern bar lines added for clarity. We know in these cases that the emphasis and phrasing follows the musical phrase, rather than the artificial bar line. But having the bar lines (and bar numbers) is critical because how else are you going to get a bunch of singers to start at the same place? Not as big of an issue if this is music for a single performer, but it solves the accidental ambiguity.
    – Selvek
    Commented Oct 9, 2019 at 16:25
  • I've also seen quite a bit of music that changes time signatures on nearly every bar. That lets you write freely, while still using the bar line to add emphasis where you want it. It can get a bit cluttered though.
    – Selvek
    Commented Oct 9, 2019 at 16:28

6 Answers 6


Your #2 thought is the convention. Yes, this means you'll need to write lots of accidentals. That's okay. Contemporary players, especially those who regularly perform new music are used to reading lots of accidentals. A fair number of people (including myself) actually prefer reading accidentals over key signatures.

In addition to writing in every accidental, you'll need to include a note in the score / parts above that section saying "accidentals only apply to single pitch" as well as a note in the front matter detailing the same thing. I would also include a lot of courtesy accidentals as well, just to make sure everything is super clear.

  • 2
    A common convention is that if a note with an accidental is immediately repeated, you don't need to add an accidental to it. This is fairly normal even if there are bar lines. But it's worth while putting this into a note for clarity. And I agree about the courtesy accidentals.
    – Peter
    Commented Oct 8, 2019 at 17:15
  • 1
    @Peter Yes, so common I forgot to mention that point. Comment notwithstanding, it would be good to add that to directions in the score for sake of clarity. Commented Oct 9, 2019 at 2:51

There are two basic conventions.

The "second Viennese school" (Berg, Schoenberg, Webern, and their followers) chose to write every note with an accidental, including every note that has a natural.

That is not so bad for monophonic instruments, but it usually looks a mess for polyphonic notation.

The "modernist" or "neo-modern" convention is that an accidental applies to all notes that are immediately repeated at the same pitch. If any other note intervenes, the accidental is shown again.

The second Viennese school convention has the advantage that it is easy to proof read hand written music. Any note with a missing accidental is a mistake, and every note has an unambiguous pitch.

The big problem with the modernist convention is that most computer notation software doesn't support it automatically, and therefore you can "break the rules" but still get the correct computer playback. In practice it is advisable to add plenty of cautionary accidentals, to avoid wasting lots of rehearsal time getting everybody to play the right notes.

Lilypond supports a wide range of different convention for accidentals, including both of the above. See http://lilypond.org/doc/v2.18/Documentation/notation/displaying-pitches#automatic-accidentals.

The option that accidentals last "about a bar" is fairly useless, unless the music is such that everybody reading the score will have the same idea about the length of a bar - and in that case, writing the barlines isn't going to give anyone a misleading idea about the music, so just write them! You can have bars of unequal lengths with no time signatures, if you like.


I yield to the other contributors as regards conventions. However, I would like to share some intuitive thoughts in case they help you.

Meter is one thing, and accidentals are another. If you have a section that is in a certain tonality, for the most part, then you could provide a key signature for that section, and include accidentals in the usual way, within that section. To switch to a different key signature in another section, you could put a double bar line to separate the sections. Depending on the piece, this might help with readability.


When you mention Satie I think of his interest in Medieval music and then I think perhaps you could follow the conventions of modern notation for Medieval music.

For the unmetered part: https://en.wikipedia.org/wiki/Mensurstrich

For the accidentals: https://en.wikipedia.org/wiki/Musica_ficta#Modern_editions

My understanding of musica ficta is that it was about either raising the 7th scale degree or lowering the 4th scale degree (glossing over the complicated hexachord solfege history.) If the accidentals in your music go beyond that, or if there isn't any similarity to Medieval music, this probably isn't a good idea.


I‘ve invented another staff system:

enter image description here

You could use this alternative system: the lines are the black keys of the piano: the 2 lines are C# and D#, the 3 lines are F#, G# and A#. The natural notes D G A are between the lines, e.g. E is above the second of the 2 lines, F below the lower of the 3 lines, B and C analog ...

This system is quite easy to read as it represents a keyboard and you don‘t need any sharps nor flats.

  • I invented a different variation on piano-roll notation ages ago: all notes go in spaces, but lines between black and white keys are positioned slightly toward the black keys. On the other hand, very few performers are going to be able to read piano-roll notation, so such notations are mainly useful only for entering music that will be performed by machine, and don't really seem applicable to this question.
    – supercat
    Commented Oct 10, 2019 at 16:48
  • I had the idea of this notation 30 years before I heard of piano roll ... well, I think, someone who is able to read OP‘s lots of sharps and flats all in one bar would be able to read this kind of piano roll too. Commented Oct 10, 2019 at 17:19

An accidental lasts for as long as it does. At the point you need to cancel, write a natural sign. That then lasts until it's no longer needed. Then another accidental comes out to play.

There's no need to mark every note - which presumably will be a natural in its normal state. The problem for you may be that you notice loads of F♯s, for example, and at that point start thinking ' this could be marked in the key sig...'

  • 1
    A large amount of music written in this way actually follows the convention of accidentals only modifying notes that they touch. Imagine reading a B and having to remember whether it was natural or flat 15 measures ago...
    – user45266
    Commented Oct 8, 2019 at 17:08
  • 1
    @user45266 - so the antidote is putting an accidental against each and every changed note? Sounds pretty pointless. But that, presumably, is your solution. And - 15 measures ago? There were no measures!
    – Tim
    Commented Oct 8, 2019 at 17:35
  • @Tim No, it is not pointless; it is unambiguous.
    – ibonyun
    Commented Oct 8, 2019 at 22:30
  • 1
    @Tim - The proposition was to mark every accidental. Accidentals are only for things which differ from the key signature. If you're in C Major you would never mark a natural (except possibly as a courtesy if you had an accidental immediately preceding, like F#>F), but you would mark every F#, no matter how sparse or common.
    – Delioth
    Commented Oct 9, 2019 at 17:17
  • 1
    @Tim The fact of the matter is that there is an established prevailing convention which covers the OP's question. That is so say, there is a correct answer to this question, which other answerers have provided. Conventions aside, your answer is simply unhelpful because it does not solve the problem the OP faces. The problem is the ambiguity that arises when trying to notate non-traditional music using traditional notation. Perhaps you've never encountered such music so don't appreciate first hand the problems that come up.
    – ibonyun
    Commented Oct 9, 2019 at 18:57

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.