Re: Beams not connecting

[Hans Aberg]

On 9 Oct 2007, at 15:00, Mats Bengtsson wrote:

> The true bug, in my opinion, is that beams are subdivided, even  
> though subdivideBeams is set to ##f (which it is by
> default). This is related to bug report #11, for example, but has  
> nothing to do with triplets. See for example, the following
> example.
>
> \version "2.11.33"
> \paper{ ragged-right=##t }
> \relative c' {
>  \time 6/8
>  c8 c c16 c d8 d16 d d8 |  e8. e16 e e f8 f16 f8 f16 \break
>  \set Timing.beatLength = #(ly:make-moment 3 8)
>  c8 c c16 c d8 d16 d d8 |  e8. e16 e e f8 f16 f8 f16 }
>
> I have not been able to find any setting of beatLength or  
> beatGrouping or anything else that gives a full beam
> (without any subdivisions) for both the rhythms d8 d16 d d8
> and e8. e16 e e .

In order ti illustrate my reasoning (see below), change the meter to  
3/8:
----
\version "2.11.33"
\paper{ ragged-right=##t }
\relative c' {
 \time 3/8
 c8 c c16 c | d8 d16 d d8 | e8. e16 e e | f8 f16 f8 f16 \break
 \set Timing.beatLength = #(ly:make-moment 3 8)
 c8 c c16 c | d8 d16 d d8 |  e8. e16 e e | f8 f16 f8 f16
}
----

From the musical point of view, in 3/8 time, there is a primary  
metric accent at beat 1, and subordinate metric accents at beats 2  
and 3. In 6/8 time, in CPP, it is a rewritten 2/4 time with triplets;  
so it has a primary metric accent at beat 1, a subordinate one at  
beat 4, and no metric accents at beats 2, 3, 5, 6.

Now look at the 3/8 example, and think of the metric accents (count  
the beats with the fingers and sing the rhythm, at least mentally);  
then LilyPond actually typesets the first row (first 4 measures)  
correctly. This is formally correct if the metric accents should be  
brought out. The second row would be correct if one subdivides the  
meter 3/8 = 3/16 + 3/16. This kind of subdivision is similar to that  
of Bernstein's song "America", but it never occurs in CPP.

In fact, in traditional CPP typesetting, measure 7 is preferred over  
measure 3. despite the fact that the metric accentuation comes out  
wrongly, if the beaming should be used as an indication of that. My  
guess this happens because the other musical forms do not occur in  
CPP, and therefore it does not matter how they are typeset; one can  
tie them together just to make it look nice.

But if I should analyze it musically: in a subdivision of time into k  
equal parts (as in a measure k/n or a k-tuplet), the "k" can have two  
meanings: either as a primary accent on beat one and subordinate  
accents on beats 2, ..., k, or only a primary accent starting at beat  
one, and no separate accents starting on the other beats. The latter  
form just indices a measure of time that the beat takes up; write  
this from as k', with an accents. Then the two meters are 3/8 and  
(3'+3')/8.

And in typesetting, this latter form k' can probably be typeset just  
by grouping the beaming as to look nice, as there is no need to use  
beaming to bring the accents out. (I think the k' form can also be  
used when one does not want to indicate any acents, even though they  
may be present.) This is what I think is done in traditional CPP  
typesetting in the examples above. LilyPond, however, always chooses  
a combination of the k-subdivisions (and not the k'-subdivisions).  
This breaks off from traditional typesetting in some circumstances,  
even though, from the musical point of view it may be correct.

So in summary, there are two different typesetting styles in play,  
with different musical meanings (the k- and k'-subdivisions). In  
traditional CPP music, the musical context is clear, so it has a  
special way to typeset this.

  Hans Åberg