Re: Right-Left stoke matching

["John Grove" <John@xxxxxxxxxxxx>]

> Jorge Stofi wrote:
> [...]
> >Speaking of which: a while ago I posted a message calling attention to
> >the above, and to the apparent correlation between the right stroke of
> >each glyph and the left stroke of the following glyph.

I'm not quite sure I follow - the right stroke of most characters fall under
roughly five types specifically: d,j / s,r  / ch,ih / n,b / & possibly y,l.

All of which are almost 100% of the time followed by a 'c' stroke character
(including a).

The only time (98% or better) that the 'left stroke' will be an 'i' is when
it is preceded by an a - which is itself a 'c' stroke character. Therefore
the left stroke is almost always going to be a 'c' stroke. The finals being
d,s,h,b,y, a + j,r,h,n,l? (I have a problem with l).

This is confounded by multiple 'c' and 'i' extensions including words with
multiple 'c' preceding an 'a' followed by multiple 'i' extensions before
reaching the final.

Of course none of this goes into Gallows territory, except that Gallows are
almost always followed by a 'c' stroke character as well.

By my definition of a character 'code' a character begins with a 'c' stroke
and ends with one of the five finals - then a new character may begin.
Multiple 'c' strokes means different 'words or characters' are represented
by the stream.

Thus, deey consists of two character  c-stroke with d final & 'ccc' stroke
with y final.

Again the problem that I have with this definition is that most words seem
not to need the multiples and relies on a very small character set to make
up the words. 'i' multiples outweigh 'c' multiples by a long shot (I think).

John.

> >
> >--stolfi
>
> Yes indeed, I did a calculation to find a correlation between character
> pairs and the the following numbers:
> 105547 occurences of char pairs with r-l matching strokes
> 28459 pairs with non matching strokes.
>
> The matching test was done "visually", i.e first I wrote an awk-script to
> find all pairs of characters with an overall count (absolute number),
> converted the output into MS ACCESS and then inspected the table using the
> EVA-Font (only 228 differnet pairs whre found by my script, .*,-=% in
pairs
> where excluded.
>
> If interest exists, I can export may ACCESS table as text and post it.
> Cheers
> Claus Anders
>
>