Thank you very much for the response Rene,
I think I'm going to follow your lead and document bench variations rather than try lump them together. Luckily I have restricted myself to Q1 herbal pages, and (currently) my system only cares about framework. If we considered a bench to be ( ) the system doesn't care what you write in it, "ckhh" etc. As a later step it might be interesting to look at each non-conforming bench and see if there are any patterns such as "Sh" variations.
I will note some examples on "ci" when I tackle Currier B, I have a feeling it is a "B thing" but I don't know/have any work on it currently. Here's a somewhat suspicious "ch" from 77r
![[Image: ci.jpg]](https://i.postimg.cc/d01dYz1M/ci.jpg)
I have noticed in the past some examples where the \ stroke looks to be corrected to \ from ( (or maybe made more obvious), but it doesn't look like I noted the pages.
It is interesting that your hypothesis would solve a problem I have with my system, and also better explain something I already had in place...
I call a,o,y, "switch" (>) which generally means "the next thing is line based", but in Pfeasters work (You are not allowed to view links.
Register or
Login to view.) he notes "In building further on this model, Stolfi treats the glyphs a, o, and y as modifiers of the glyphs to their right, or as separate glyphs when they appear at the ends of words" which is better way of putting it in general.
I have been stating that my system doesn't care what is in the middle to simplify things, but to fully note what is happening in a benched gallows glyph I would have to imply "a,o,y" (switch) as we have a curve and a line in "ck" touching. So I would note "ckh" as "c>kc" (curve.switch.line-curve.curve) rather than "ckc" (otherwise every benched gallows would not conform).
Sequences of "hc" are common (644 matches), but so are "hoc" (154 matches) comparatively. "chch" only makes up 18 of 644, and "choch" 14 of 154, so this is mostly interactions regarding benched gallows.
Generally I have found things that use "o,a,y" are not also found without it in any great number. sr - 4, sor 94, sl - 5, sol 148.. and so on. Bench to benched gallows is strange in this way.
A specific example.
chcth - 143
chocth - 38
My system would gain it's missing "switch" for "cth" and so change the sequence in the middle of chocth - c
hocth
from "a curve switching to a curve (non-conforming) followed by an implied switch to a line based glyph" Messy.
to "a curve switching to a line based glyph" Which is how I see line based glyphs working.
Obviously my system could end up being nonsense, but it is just interesting that it also very elegantly solves every benched gallows problem I have, in that the start of the framework would now be "switch" rather than "curve".