So to answer your question about staff sizes:

You're asking the wrong question.

Since the 1850's music has bee blown up and shot down to any size you want.
This is, for reasons I'll get into later, often a really bad mistake. This
is also the reason why looking at scores and trying to measure their size,
and then trying to make sense out to the result can be so frustrating.

In real engraving, everything, and I do mean everything, is set up on a
horizontal AND vertical grid. The real question is not how large is the
staff, but how many spaces across.
 
If you take the height of the staff and divide it into the length, and then
multiply by 4 you will have the number of units on the staff. Different
publishing houses have different engraving areas. The old Breitkopf
classical piano format was 107 accross x 154 high. The modern piano format
is about 119 accross. The vertical varies with the kind of music and the
publisher. Because C.F. Peters has a horizontal engraving size of 7 3/8
inches, there staff is 118 accross. G.Schirmer is 7 5/8 so they wind up with
121. Score's default is 7.5 inches so you wind up with 119. This is what is
usualy called a Rastral 3. Rastral 2 is about 107 or 108, Rastral 4 is about
127 - 132. 

Rastral 3 translates in SCORE to a Code 8 P5 value of .72. This is very
convienent since the staff is already locked on the grid (which means you
can move the staff by intering only two digits, rather than four, or using
the cursor arrows.

In SCORE, the P5 value multiplied by .35 will give you the staff height in
inches (FORTRAN'S default resolution is 4000 x 4000 dpi). Again divide the
height into the satff width (7.5 inches for SCORE) and multiply by four for
the staff width in units that are the same as the vertical space between two
adjacent staff lines.

Since the default spacing for P5=1 is five spaces between staves, and this
(for reasons that I will never, ever understand) remains constant when the
staff sizes is changed, if you want to lock onto the vertical grid you have
to divide 18 by the P5 value and multiply by -1. For P5 = .72 this will give
you a value of -25. If you set Code 8 P4 staff Nr. 2 to -25, staff one and
two will print right on top of each other. If you set the P4 of staff Nr. 3
to -50 all three staves will print on top of each other. And so on. This is
very handy for engraving more than one voice on a line since the edit
function (EDI) will always work. Otherwise it doesn't. 

Now this is getting too long. Think about it, and I'll answer your
questions. Don't look for any of this in the manuals, it isn't there.

To close up, the trouble with reducing and enlarging is that, as
typographers figured out in the 16th century, when you change the size of a
font, the shapes of the symbols have to change too. A nice fat serif in 72
points will dissapear if the symbol is reduced to 7 points. SCORE's font
isn't too bad around Rastral 5. Otherwise it needs help. If you look at good
engraving in SCORE you will notice that different engravers have their own
symbol libraries. A real music engraving program would have to have a least
8 different sets of symbols. Which is a bit of work.

george mcguire

****


There isn'y really anything usefull written by high quality engraving. The
reason is simple - the whole system was based on apprenticeship, and if you
want to sell it, you can't give it away.

Also engravers don't tend to be very verbal. The one great teacher I had,
Walter Boelke who apprenticed at Roeder and became the chief engraver at
G.Schirmer in New York, never told me anything. But he would sit next to me
and grunt when I did something right.


*******


>
>My best reference (Wanske) says that Rastral are fixed sizes of
>staffs, so you are saying that the staff lengths come in fixed sets as
>well.
>

The sizes were fixed for the publisher she was working for (Schott), which
are very close to Breitkopf.
But the Roeder sizes were different. There is a long history behind this -
starting with the fact that the first German engraving workshop (methods,
machinery, tools and engravers) was imported from England (?).


******

>If I understand you correctly, you are saying that the scaleable part
>of msuic isn't so much the height, but how many symbols you can cramp
>onto one line, and how many lines (systems) on one page.  Or do you
>mean that I should not be thinking in "dimensions" but "ratios".
>

Yes, basically the rations are what is important. The horizontal size was
dependent on the piece of metal.
On the other hand metal was expensive and the sizes and layout had
everything to do with how much you could cram on a page.

****

That's okay as far as it goes. But if you look at different size noteheads
you will notice that they are ovals, and that the angles from the horizontal
of the main axises change with the size. Of course this is something Tex
deals with easily and well.

****

Table from Wanske:


16.5 15.5 14.5 13.5 12 11.5 9
143 12 11 10 9 8.5 7
11 10 9 8 7 6.5 5