Thickness of lines

Music engraving isn’t only about placing objects in the right spots – it’s also about giving them correct appearance.  One of the easily overlooked properties that can make a significant impact on the score’s “feel” and readability is the thickness of various lines.

Now, ordinary books, newspapers etc. usually consist of lots of letters and an occasional table, drawing or picture – so, except for the tables and drawings, there are not many lines here and their importance is very small.  Musical scores are the complete opposite – they contain numerous important line-shaped objects: 

  • stafflines,
  • stems,
  • barlines,
  • ledger lines,
  • beams,
  • hairpins,
  • ties,
  • slurs,
  • lyric extenders,
  • volta brackets,
  • and so on.

It would be foolish to assume that all these lines should have the same thickness. For instance, beams should obviously be a lot thicker than other objects mentioned above.  Unfortunately, this is the place where many engravers stop – except for the beams, they make all lines very similar, which is a bad idea.

Let’s consider just these four most basic elements: stafflines, stems, barlines and ledger lines – they appear in almost every score and I estimate that they can take 50% or more of the ink used to print it.  What shall we do to make the results as readable as possible?

Firstly, in my opinion the stems of the notes should be slightly thicker than the stafflines.  That’s because we want to create an impression that the staff is in the background, with the notes being placed on its top.  I find it hard to understand why Elaine Gould in her excellent book Behind Bars recommends that the stems should be thinner than stafflines – it doesn’t make sense to me… The only explanation I can imagine is that she based this statement on the older engraving books (such as Ted Ross’ Teach Yourself The Art of Music Engraving & Processing), which in turn recommended such practice because the stafflines they used were already really thick.  While I agree that it may be acceptable to make stems the same thickness as stafflines, I would never make them thinner.

What the exact ratio should be?  Well, it depends on several factors, and I’ll probably write another post about it, but right now I can tell you that by default LilyPond produces stems that are 1.3 times thicker than stafflines and I think it’s quite fine.

The second issue is barline thickness, and here the situation is clear: barlines have to be thicker than both stafflines and stems; they must be instantly recognizable.  Again, the exact thickness depends on the context, but it is important that barlines stand out and make the measure boundaries obvious – unfortunately, I’ve seen many scores in which barlines had the same thickness as both stafflines and stems.  This is simply a bad idea.

Finally, we have ledger lines.  Engraving rules say that they should be thicker than the stafflines, and they say it for a reason: thick ledgers will not be easily confused with stafflines, and – somewhat suprisingly – they are easier to count.  I play the guitar a bit, and I’ve noticed that when the bass notes’ ledgers are thin (and there can be up to four of them), my brain starts looking at the staff itself (in addition to the ledgers), and easily looses its count.  When ledgers are thick, all that is necessary for decoding the pitch is counting the ledgers alone, without looking at the staff at all.

I’ve also noticed that ledger line thickness actually affects the performance and interpretation of the piece, at least in my case.  When I see a note on strong ledgers, they make me play it with more confidence – actually, this is even more noticeable when singing, especially high notes.  I sing bass in Epifania choir, and high notes (above one-lined c) are quite difficult for me.  There is one place in Antonín Dvořák’s Stabat Mater (in 14th and 73rd measure of 3rd movement, Eja Mater) where basses have to sing one-lined f, which is somewhat beyond our capabilities.  It usually sounds badly, like a cry of despair, and every time I look at the score (which was made in Finale and has thin ledgers), the anemic appearance of this note on paper reminds me of our anemic performance.  On the other hand,  when I look at the version with thick ledgers, I imagine myself singing this sound powerfully!

Time for examples.  In the first one there is an excerpt from a short guitar piece: on the top there is a scan from an edition published by Absonic (the jagged edges, for example on sloped beams, weren’t introduced by me in the scanning process – they were looking like this on the original print as well.  It seems the book was printed on a low resolution printer).  The middle one (B) is the same fragment typeset by me using LilyPond, with different objects having different thicknesses.  Since these two engravings use different fonts, spacing etc. it may be hard to compare just the impact of different thicknesses – therefore I’ve prepared version C, which is identical to B except for the thickness of stafflines, stems, barlines and ledgers.

guitar miniature

click to download pdf

The second example is the choir part of the 3rd movement from Antonín Dvořák’s Stabat Mater (the one with high f for basses).  You can see the version with varying thickness here, and the same-thickness one here (this is not the edition that my choir used).

Important note: due to low resolution of modern computer screens (even MacBooks with impressive Retina display still have lower resolution than paper prints), I highly recommend printing these scores using a decent printer (1200 dpi or more) instead of comparing them on-screen.

8 thoughts on “Thickness of lines

  1. Urs Liska

    Is the Version B of the guitar examples done with LilyPond’s default settings or have you adjusted them to your liking?
    (Of course Version C does _not_ use default settings but you have adjusted them to your disliking 😉 )

  2. Janek Warchoł

    Stems and barlines in version B are slightly thicker than default Lilypond values. I’ve noticed that these values should depend on the type of music, density of notation, and other factors. I hope to write a few separate posts discussing the details.

  3. Jan Nieuwenhuizen

    Yes, LilyPond comes with a sort of built-in “default stylesheet”. It would be nice to have some guidelines and experiments with what to tweak and how to tweak them. Great article.

    1. Janek Warchoł

      We definitely need to do some experiments! The only problem is sharing the results: printouts can vary greatly in their appearancee… We’d have to discuss the results in person, looking at the same piece of paper 🙂

      1. Urs Liska

        I think there are two different ideas here: It would definitely be interesting to see some discussion, examples and comparisons of this kind of engraving details.
        On the other hand I also think that an “Introduction to adjusting the visual appearance” would be a nice thing, as this is something that is usually quite daunting to anybody learning to use LilyPond. Would make up for a nice openLilyLib Tutorial, but I won’t have the time to write it myself …

      2. Gijs van Oort

        Just as a professional recording should sound ‘all right’ on a great variety of audio sets (from a car radio to a high-end HiFi installation), I think that the default appearance should be ‘nice’ on a great variety of printers and screens (don’t forget that I-Pads and E-readers probably will be used more and more on stage). So, probably, it is not so bad if two people look at different printouts; they should all look nice.

        (I’m not using Lilypond a lot, so I don’t know exactly what’s in there and what not…)

        1. Janek Warchoł Post author

          I agree that we should aim to have LilyPond scores look good on a wide variety of media. However, for discussing we need a common reference.

          After all – continuing with the comparison to audio recordings – a professional studio uses a specific audio setup which is designed to be “transparent”, i.e. have a linear characteristic and reproduce sound very accurately. In publishing industry this would mean using the highest print resolution possible.


Leave a Reply

Your email address will not be published. Required fields are marked *