[Gnu-music-discuss] Re: Tie issues...

[Scott Ballantyne]

David Raleigh Arnold <address@hidden>
> 
> It is important to be clear about what is necessary and what is not when
> making up rules. :-)
> 

I guess you and I have different rule books. The rule I learned tells
me that an accidentals life ends at the bar line, even if the note it
affects is tied over. This rule requires that accidentals be repeated
on any non-tied notes in the new measures. These are not 'reminders',
they are required to establish the correct pitch. This seems to me to
be common practice by the finest composers, and examples supporting
this contention are not difficult to find: Mendelssohn octet, around
measure 60 in the first movement. Second violin part:

       f1~|f4 f2 f4(|)e1~|e4 e2 e4|

Since this is in E flat major, the first e requires a natural. The e2
in the next measure also requires a natural sign, which one sees in
the printed score. You see the same treatment of the first cello in
the same measures:

      d1~|d4 d2 d4(|)df1~|df4 df2 df4|

Again, the df2 requires an accidental, and is given one in the score.

This is consistantly followed by Mendelssohn throughout. Here's an
example from the scherzo, second violin part:

        af2~|af2~|[af8 ef( af g| af ef af )g| fs2 ~ | fs8 fs( g )fs|
                           !                              !

Again, the places I've marked with an ! require an accidental, and are
given one by Mendelssohn, in spite of the fact that ties to the same
pitch exist.

Just to demonstrate this is not unique to Mendelssohn: Dover, complete
symphonies of Brahms. 3rd Symphony, letter B, first violin part has D
flat for three measures, all of which are tied over the bar line, all
of which Brahms repeated the accidental. There are multiple other
examples in this work, incidentally, again, quite consistently
followed by Brahms.

I certainly support duplicating the composers desires in notation over
any 'rule' or modern 'taste' that may be prevailing today. So if you
can come up with enough counter examples from the literature, I would
hope that the lilypond implementors should provide a property that
will set one or the other default treatment of accidentals. If not,
then I devoutly hope the present bug will be fixed soon.

sdb
-- 
address@hidden