Optimal Horizontal Dynamics Placement

Basic horizontal dynamic placement tends to lack nuance in current practice. The various software packages simply define a default position for a given text style and leave it to the user to do any manual changes. In practice, this often means that basic dynamic positions are simply left as is with little thought to historical convention or good aesthetics. They tend to get adjusted in awkward situations, but rarely otherwise.

Standard dynamic placement isn't treated with much depth in the major texts, but two main objectives are evident:

Gould: "Centre the dynamic on the notehead" (p. 102)
Ross: "A multiple mark straddles the stem of the notehead which it is beneath" (p. 186)

The following chart proposes default positions for the major dynamics; offsets are given in spaces from Sibelius defaults. Finale treats dynamic placement differently, but the idea is the same.

Optimal Dynamic Placement
(Click for larger version)

In a score system, I will take the leftmost dynamic on a given beat and align all the others to it. It's more important in that situation to have a unified line of dynamics so the conductor can quickly grasp the change. These positions are for isolated dynamics and to establish baseline left side positions.

I will be interested to hear your thoughts! It would be very nice if the software packages were able to make these adjustments automatically, but that's why I stay in business.

Update:

Jeremy Hughes suggests that my positions are a little bit further to the right than he would like. I agree with him. Instead of centering the cleft of the p and the crossbar of the f, I am now centering the bowl of the p and the upper curve of the f. Here's a revised chart with a couple of extra dynamics:

Revised dynamic placements
(Click for larger version)

Let me know what you all think in the comments!

Hammered glissando for percussion

I was recently required to make a glissando for mallet percussion instruments that asked the performer to hammer all the way through rather than just slide the mallet across the bars. The standard way to indicate articulation in a glissando is to have headless stems with the desired rhythm. This was my first proposal, but it fails to fully capture the requirement:

Since it explicitly requires a rhythm, the performer would be likely to attempt to be precise in its execution. The desired outcome is to have an unmeasured flurry of hits all the way through the figure. My eventual solution was as follows:

The 32nd notes with extended beams suggest that the figure is unmeasured and very fast. The whole note defines the starting pitch and duration of the gliss. The text clearly states the requirement. Result: no confusion.

Update: John pointed out that having the mixed roman and italic text looks odd, especially following the line. I think that a better solution is to treat the text as a technique and have it all in horizontal roman type as follows:

Behind Bars readthrough: Chapter 1: Stem thickness and Center line direction

Welcome back to the readthrough for Behind Bars. To get started, you can see the previous posts in the series:  Introduction and Staves, clefs and noteheads.

In this post, I cover stem thickness and single-note direction.

STEMS


Thickness: Gould suggests that the stems should be thinner than a stave line, but not too thin. I believe that she is quoting Ted Ross here. However, the examples in Ross' book have (to modern eyes) quite thick stave lines. Perhaps the lines had to be scored thicker or deeper in order to stand up to the punching and smoothing process.

Stephen Powell in Music Engraving Today (Brichtmark, 2007) suggests that both stems and stave lines should be the same thickness, but to increase that thickness slightly as the size of the staff gets smaller. Sibelius puts the default thickness of both stave lines and stems at 0.1 spaces; Finale puts the default thickness of both stave lines and stems at .075 spaces.

Edward Tufte, the information display guru, suggests that stave lines are a form of "chartjunk," clutter that obscures the data in an information graphic. In Envisioning Information (Graphics Press, 1990), he recommends printing stave lines thinner and in gray. This has the effect of popping the notes off the page and making them very clear while maintaining the pitch information.

Variations in stave line thickness alter readability

To my eye, Ross' lines are too thick; they imprison the music rather than encourage easy reading. I find that Tufte's approach is very interesting, but may lack feasibility in the real world. The more delicate lines certainly have the effect of putting the notes dramatically frontmost, but care must be taken to ensure that the lines are not so light that they blend into the background or disappear altogether. Inferior reproduction and printing methods will tend to wash out the lines, rendering the music unreadable. Neither Sibelius nor Finale offers a method to easily make stave lines gray.

My recommendation is to stick with the middle ground – stems and stave lines should be the same size, or perhaps the stave lines could be a little thinner to push them into the background more. Experimentation will reveal a balance point somewhere, much like choosing lenses at an optometrist's. It's curious to note that Gould's examples all feature stems and stave lines that are exactly the same thickness, contravening her own recommendation.

Direction: For single notes, the standard convention: notes above the middle line have down-stems; below have up-stems. The middle line is where things get interesting.

Gould surprisingly recommends an option for the middle line: the direction of the stem is determined by the surrounding stems. If there's no clear-cut case for either direction, use a down-stem.

A selection of quotes on the matter:

Read (1969): "When the note is centered on the staff, ... the stem may go in either direction, although it is the more common practice to draw it down."

Stone (1980): "The old rule that the stem direction for notes on the middle line of the staff is governed by the majority of the other stems in the measure ... is rarely if ever followed any longer, at least not by today's professional engravers and autographers."

Ross (1987): "Some engravers consider the middle line neutral, and take the option of using either up- or down-stems for notes that fall on it. However, more up-to-date engraving no longer permits an option; now a down-stem is always appropriate"

Chlapik (1991): "Alle Noten, die unter der Mittellinie stehen werden hinauf-, ab dieser und darüber heruntergestrichen."/ "All notes below the middle line will be up-stemmed, on and above the middle line will be down-stemmed."

Powell (2007): "When there is one note and it is on the middle line of the staff or higher, the stem goes down."

So: why does Gould go back to the old style? I expect she feels that it just looks better: stemming the note with its neighbors assists with the flow of reading and maintains smooth phrasing. What prompted the shift away from this practice? Why is it so bad to have an up-stemmed middle line? None of my sources gives an explanation – does anyone know?

Next time: Stem length and tails!

Behind Bars readthrough: Chapter 1: Staves, clefs and noteheads

Ground Rules: in which the basics of notation are established and expounded upon. For an introduction to this project, see the first post. Throughout Behind Bars, British naming conventions are in use: stave, crotchet, minim, etc. I shall use her terminology where applicable and revert to standard American where appropriate.

Sections:

The stave [staff]

Clefs

Noteheads

Stems

Tails

Beams

Ledger [leger] lines

Octave signs

Rest symbols

Barlines

Rhythmic spacing

Spacing symbols

The stave:

The stave is the basis for all musical measurement. A stave-space (or, in American, staff-space) is the distance between two stave-lines and serves as a fundamental scalable unit. The stave size may change, but the proportional relationship of symbols does not. We will discuss appropriate stave sizes much, much later.

Almost all music is written on a five-line stave as it is immediately recognizable. The other frequently-used stave type is that of a single line. Its primary use is for indefinite pitch of all sorts and as a baseline to show approximate pitch. Two- and three-line staves are useful in percussion music. Kurt Stone recommends against four-line staves because players can more readily grasp positions on a five- or three-line stave. (p. 216)

Clefs:

Recommended
Percussion Clef

Gould outlines the basic clefs: treble, bass, and the various C clefs. She recommends a double vertical-stroke clef for percussion and notes that it is optional for staves of fewer than five lines.

We will talk about the horizontal spacing of initial items (clef, key signature, time signature and initial note) a little later.

Her recommendations for changing clef are quite thorough and expand significantly on Ross, Read and Stone. Stone notes that clef changes should come one space before the barline and that "there are no specific rules for clef changes within a measure" (p. 46). Ted Ross and Gardner Read have some recommendations, but fall somewhat short of comprehensive.

Consensus rules for clef changes:

1. Use a clef change where there is an extended constant use of ledger lines
2. Avoid frequent clef changes (Gould: use up to three ledger lines rather than changing frequently. Preserve the contour of the line wherever possible.)
3. Use a clef two-thirds of (Ross: two or three sizes smaller than) normal for the change.
4. Generally place the new clef immediately before the note involved.
5. If the change affects an entire measure, place the clef before the barline
6. For mid-bar changes, use the whole beat immediately preceding the affected notes (Gould adds to use half-beats or other subdivisions if necessary.

Gould's additions:

7. Changes between beats come after rests.
8. It is better to keep the phrase contour intact rather than to keep to whole beats.
9. Avoid changing clefs between ties; if you must, use a system break
10. After periods of rests, change back to a conventional clef at the end of the system after the player has finished. If changing to a nonstandard clef, change according to foregoing rules.

Common clef change situations. Examples adapted from Gould: Behind Bars.

Comments:

Gould’s recommendations amplify the standard texts in very helpful ways. Her central aim is to preserve the contour of the line and disrupt spacing as little as possible. The suggestion to change back from a nonstandard clef to standard is very interesting. When you put the change back at the end of a line, it's almost invisible. The next line starts with the proper clef and the performer doesn’t get the double-take when the new clef appears immediately before playing.

Noteheads:

Black noteheads: oval shape that slopes away from stem.

White noteheads: 

• Minim (half note): oval shape with diagonal shading following slope of the oval.
• Semibreve (whole note): wider oval shape with shading slightly left of perpendicular.
• Breve (double whole note): may have two lines, one line or be rectangular in shape.  

Diamond shaped noteheads are used for the following functions:

• harmonics in instrumental writing
• silently depressed keys in piano music
• differentiate unconventional techniques
• singing through a wind instrument
• multiphonics
• unvoiced vocal sounds
• falsetto

Crossed noteheads are unshaded and fill a stave-space. Steven Powell in Music Engraving Today (Brichtmark, 2007) suggests to use a double-sharp sign in place of unshaded lines because it is more visible on the page. Gould demurs, recommending reserving the double-sharp sign for its role as an accidental.

Showing a minim or semibreve with a crossed notehead is done in one of two ways: a small x in the middle of an unshaded circle the size of a minim or semibreve or as diamond noteheads. I find that diamond noteheads are significantly easier to read.

Uses for crossed noteheads:

• notating sounds of indefinite pitch on a five-line stave
• to distinguish certain instruments in percussion writing
• spoken text
• differentiate between speech and singing
• showing approximate pitch

Triangular noteheads are used for the highest or lowest possible pitch, where it cannot be specified. It may be black or white depending on duration and does not take a leger line.

Comments:

Gould specifies that semibreve noteheads have a ratio of 2.5 to 3 black noteheads. This seemed off to me. I took several music fonts in current use and compared the sizes of black to semibreve noteheads and came up with a different result. Her ratio works out to .833; the actual number ranges from .71 to .76 with an average of .73. The designers of these fonts need to balance two things: ease of legibility and evenness of color. When the size of the black notehead and half-note is closer in size to the whole note, the page can flow more evenly, but it is subtly more difficult to quickly tell the difference between half and whole notes. If the ratio is too small (and Petrucci approaches this point) the line tends to have hitches in it which impede facile reading. 

An interesting side-effect of this exercise is the illumination of the various widths of the fonts. Opus is by far the widest music font and Petrucci the narrowest. Opus' generous space invites easy reading, but Maestro and Helsinki may be better suited to tightly-packed spaces. Sonata and Petrucci are quite outdated at this point and engravers will be better served with other choices. I would welcome comments from those more informed than I on this topic.

Next week: Stems, tails and beams (hooray beams!)

Behind Bars readthrough: Introduction

Elaine Gould's Behind Bars is a fantastic new reference work on all aspects of music notation. Over the next 6 months, I will be covering a chapter per week, summarizing the content and commenting on various issues that arise in the text. I will welcome thoughts and comments from the community as I progress through the material.

To begin with, the introduction covers three areas: the need for scores and parts prepared with an understanding of basic notational standards, the historical context of this book and a guide for using the book. Her argument for the need for expertise in the principles of music notation is as follows:

Summary: The need for notational standards
If computer software, given a competent user, can produce sharp-looking music, why learn about music notation? The computer can only go so far. It will apply notation principles, but without the knowledge to determine appropriateness to a situation, the result will just as likely hinder the performance as not. "Through a mutual understanding of the rules and conventions of notation, the composer can 'speak' effectively to the performer, who then has the best chance of achieving a faithful interpretation of the composer's intentions"

Commentary:
As an engraver, my goal is to facilitate the communication between composer and performer to the greatest degree possible. My lifeblood consists of these rules and conventions and their proper application. Frequently I find that composers are often extremely good at writing music but don't have a thorough grounding in the layout and disposition of a score and parts. The effect is usually to make performers work too hard to play the music.

Gould notes that in her experience, "professionals are frequently too tolerant of poor presentation." As a performer myself, I know that I almost never have any say in the quality of the part. Indeed, a particularly bad part will be the subject of jokes and bellyaching in the back row, but it will rarely get to the conductor's ears, let alone the composer's. The performer's job is to read whatever chicken scratch or too-small music is put in front of her and play it beautifully. If there are poor page turns, make a copy. If it's too scrunched, squint. When the player is working this hard just to get the music off the page, excellent performance becomes more of a luxury than a given. How much rehearsal time is wasted for someone to ask if a passage should be slurred or tongued like his neighbor? How many questions never get asked and the assumptions and errors made impact the performance negatively? How much new music is never played again because the parts were terrible?

Summary: Historical Context
The next part in the introduction concerns the historical context of this book. Her central point is that established notational conventions provide the most help to the musician because they can be read and understood quickly. In the mid-twentieth century, a wide variety of new notation was created to solve particular compositional challenges. Gould establishes her lineage here: Karkoschka's 1972 Notation in New Music and Stone's 1980 Music Notation in the Twentieth Century. Karkoschka sought to outline the multiplicity of new notational approaches and Stone sought to establish conventions that addressed the needs of contemporary composers. The intent of Behind Bars is to examine the developments of the past thirty years and propose recommendations based on current practice.

Commentary:
I have loved Kurt Stone's book for years. It was a light to me in a world full of arbitrary conventions. Here it was: these are the best ways that we could agree on to notate music. Based on the findings of the 1974 Ghent conference, more than 400 disparate notational methods were evaluated and the clearest chosen. Too often, notation manuals were based on the direct experience of one individual who undertook to poll a handful of colleagues for their input. Ted Ross' 1970 Teach Yourself the Art of Music Engraving and Processing is a wonderful resource for spacing and basic notation, but it's very idiomatic and limited in scope. Gardner Read's seminal Music Notation: a Manual of Modern Practice from 1969 is a monumental effort, but suffers from being from the perspective of a composer and scholar. The ideas put forward are intensely personal and yet presented as being absolute fact. He's done the research, but has come to his own unilateral positions. It also tends to be overwritten and pompous.

One of the reasons that I'm excited about Behind Bars is that it comes from the point of view of a music editor. Her primary goal is to create editions that make money. Assuming that the choice of music is out of her hands, she must produce the finest presentation of that music possible. High-quality engraving says to the purchaser that the utmost care was given to the production of the edition and they will be assured of a smooth and enjoyable performance of the music. To that end, she is concerned with what is best for players and how to make the work as consistent and dependable as possible. Her thirty years of experience in the field and the production of uncounted thousands of diverse scores give her the authority to make solid recommendations. Of course, the viewpoint of one person lends itself to idiosyncratic pronouncements. She acknowledges a wide variety of experts in various fields that have helped her form opinions. The result is a carefully-considered, well-presented reference work that has clear, thoughtful recommendations for an extraordinary range of situations.

Summary: Using Behind Bars
The book has a clear hierarchy of headings with related subjects grouped and cross-referenced. Topics are structured "definition -- design -- placing -- use" and progress from elementary to complex. Technical terms  in other languages are presented if in common use. Items are illustrated with good and bad practices.

"Effective communication results from establishing a convention and adopting a consistent approach." Any new recommendations are simple, clear and based in traditional practice when possible.

Commentary:
I'm looking forward to going through this book with you and exploring the current state of music notation and how it differs from and complements my own current understanding. Adventures in Music Engraving indeed!

Optical note spacing

I recently obtained a book by Herbert Chlapik entitled Die Praxis des Notengraphikers, or (roughly – my German is tenuous at best!) The Practice of the Music Engraver. With a little help from Google Translate, I'd like to present his description of optical note spacing. Neither Sibelius' nor Finale's spacing algorithms take this into account, so every instance must be corrected by hand.

Optical Compensation

Every typesetter or graphic designer understands that the different letters, especially capitals, require their typographical design be taken into account. It is an optical phenomenon that

uniform spacing between letters in certain combinations creates precisely the opposite impression. Only kerning – introducing a a small expansion or contraction between letter pairs – creates an optical illusion that suggests complete uniformity.

Though unkerned pairs in may be perceived as ugly, the readability is not affected much. This may in fact be the case with music notation as well. The widely disparate forms of musical symbols and their position to each other tends to cause irregularities which are disproportionately larger than those found in a line of text. The spacing between the characters is defined by the rhythmic structure and must assist the musicians' spontaneous comprehension. Notes must therefore be spaced not just according to their calculated position, but also so as to be visually correct.

An optical compensation must be made in the following situations:

1. When stem-up and stem-down notes are next to one another.

2. If there is a (lower) stem-up note after a rest.

3. Cross-staff beaming [this is a subset of the first case]

As examples 1–3 show, one does not calculate the distance between the notes, but between the stems. This seemingly-erroneous calculation of the note spacing causes the optical illusion that looks correct to the eye.

4. Large intervals.

Reduce the space between large intervals by about a third.

5. For the space between stem-up short-value notes and the barline.

The space between the last note and the barline should be slightly larger than the exact calculation.

Only in multisystem situations is the optical compensation improper because it can cause distortions in the other voices.

[Here he discusses vertical spacing between staves. In short, the uppermost and lowermost characters on adjacent staves define the closest possible distance between them. Any extra space left over after all the staves are spaced may be equally distributed.]

Since it is impossible to show all variants of the optical compensation, examples must be limited to the essentials. The trained eye can find other situations where a slight compensation would give a better visual result.

The translation (such as it is) is mine, but the text and images are copyrighted 1987 by Ludwig Doblinger (Bernhard Herzmansky) KG, Wien – München. Published here under the fair use provision of section 107 of the US Copyright Law.

Toward Engraved Braces

Treat braces as first-class citizens

As I was working on duplicating braces from the 1960s, I realized an important limitation of both Finale and Sibelius. Both programs use a single stretched shape to handle all possible staff widths. The implications are clear: for staves that have narrower or wider spaces than the symbol was designed for, it becomes distorted and ugly. My solution? Replace the single shape with several shapes specifically designed for a range of distances. The program should dynamically swap them out as staff distances change.

Computer fonts have a similar problem. If a font is designed to look good at 12pt, it will look too thin at 6pt and too fat at 72pt. Type companies, realizing this, release versions of high-end typefaces that are appropriately-weighted. Six-point caption-weight fonts have thicker lines and are slightly wider. Seventy-two-point heading fonts have a greater difference between thick and thin lines and have more delicately-pointed serifs. Adobe has an illuminating article on this practice here. The computer won't do it for you, though. It's up to the layout designer to know about the issue and select an appropriate typeface for the job at hand. For music engravers, the choice is not quite so easy.

How did they do it before?

For much of the history of music engraving, braces were set using differently-sized punches. Each punch size would be individually crafted to have correct proportions. With the advent of photographic reproduction and dry-transfer sheets, manufacturers began to cheat. Instead of having individually-designed braces, different sizes would be taken from one master symbol and optically reduced or enlarged. Very small and very large braces continued to have their own symbols. By the time computer software started to be written, I surmise that memory limitations and programming expediency led to the one-symbol solution. Since it is generally innocuous, it has persisted to the present day.

Let's do it right.

"Generally innocuous" does not excellent engraving make. You can see in the image below what happens at large and small distances. Sibelius scales a little better than Finale, but the braces are still distorted. If one were to apply this kind of stretching to a familiar character, one can immediately see the violence done to the letterform. As the level of stretching increases, the line weights become increasingly out of proportion. With the Boosey braces, two strategies are employed. The brace at the left is the model; the middle brace is an optically-reduced version of it. The smallest brace is a unique symbol. This is the strategy I recommend to the MakeMusic and Avid.

There should be four or five unique braces, each tailored to look good for a particular range of staff distances. Each brace should be uniformly scaled to match the specific staff distance. As distances change, different symbols should be swapped in. The number of unique symbols should be enough that its extremes are not noticeable (a brace designed for a 1-space gap would look funny enlarged to fit 4 staves). This would produce the highest possible quality for these delicate and graceful symbols.

Sub-Sub-Bracketeering

What's a sub-bracket?

It's not often that I feel like I've explored every aspect of a problem and come up empty, but in this circumstance, the feeling is warranted. The subject is instrument sub-brackets. To bring us all up to speed on this tiny bit of arcana, please examine the examples (click for a larger version) and away we go!

In full scores, instruments are bracketed according to group, then subgroup, then (very rarely) sub-sub-groups. In the example before us, we see the flute section from the Rite of Spring. The big bracket is for woodwinds, the first sub-bracket for flutes, and the sub-sub-bracket for Flutes 1 and 2,3. In older scores, sub-brackets were frequently drawn as braces (the curly bits to the left of piano staves) rather than brackets. Neither Finale nor Sibelius handles these situations very well, but MakeMusic's software does a better job than Avid's.

The crux of the matter for me is the gap between brackets and staves. Please notice on the original that the sub-brackets do not go through to the staves. The hook on the sub-bracket stops when it intersects the larger group bracket. The sub-sub-bracket's hook stops when it intersects the sub-bracket. This particular style is possible only with great effort in current engraving software.

Is there a generally recommended style?

In my entire collection of standard engraving reference books, the issue is barely, if at all addressed. Ted Ross mentions only one sub-bracket, Gardner Read indicates a preference for braces instead of brackets, and the usually reliable Stone doesn't even mention brackets at all in connection with scores. In many published scores that I have examined, this seems to be a choice colored by region, date and publisher. So it is not surprising, perhaps, that neither Finale nor Sibelius offers the option to specify the length of the sub-bracket's hooks. (I am not familiar enough with Score to render an opinion, but my understanding indicates that such a specification is possible.)

That said, Finale is the only package capable of delivering usable sub-sub-bracketing. Finale's system of grouping offers much finer control than Sibelius does. I can define an arbitrary number of staves to be part of a staff group and assign any of six bracket options. I can individually adjust the line thickness, distance from the left hand of the staff and adjust the top and bottom of the bracket. With these tools, I can create sub-sub-brackets that behave correctly under any given circumstance, even though the hooks run through to the staves.

I'm sorry if this is getting too nitpicky, but one of the reasons that Finale is often touted above Sibelius is the near-infinite adjustability of items. I'll continue at risk of losing some of you.

Mind the gap?

Finale's sub-brackets have a hook length of about 1.5 spaces. This means that I can achieve the desired gap by simply moving the sub-bracket over so that the hooks are buried inside the big bracket. The problem, as should already be apparent, is that the sub-bracket now sticks out too far. To extend the technique to the sub-sub bracket results in an uncomfortably oversized mass at the left edge of the staff. The only option, therefore, is to have brackets that protrude through one another – a valid choice, but one that I would prefer to have left up to me.

Sibelius doesn't offer much control at all over bracketing. I can only feasibly create one level of sub-bracket. The only adjustments available to me are the distance from the left of the staff, the number of staves enclosed, and line thickness. Even so, it might still be possible to implement a solution similar to Finale's, except for one thing: the distance from the left of the staff is a global variable. All sub-brackets are equally affected, affording me no way to differentiate a sub-sub-bracket. Further, no matter how far left the sub-bracket is positioned, the hooks always connect to the staff. This makes it impossible to maintain the gap even on the single sub-bracket allotted me. The only way I was able to accomplish the above example is by creating a one-off line and manually positioning it.

In fact, the only way to achieve the desired effect in either program is through painstaking and fussy effort. In Finale, I would be required to define a specific bracket shape expression and attach it to the appropriate position on every page. Such shapes would not update to changes in either staff distance or measure reflow.

My conclusion? Live with the brackets going all the way through to the staves. It's not worth the substantial effort needed to effect such a minor detail. I do prefer the cleaner look afforded by the gap, but not enough to go through the hassle.