26-10-2026, 05:33 PM
I tested whether individual words have habits of their own beyond their shape, which is what I’d expect if they carried meaning. They don’t, or at least not more than in my skribe, which has no meaning at all.
But the test showed something else: the first line of a paragraph is written differently. Besides p and f, it uses sh instead of ch much more (35% against 27%, in all five hands), t instead of k (43% against 35%), fewer ee and rarer words. I added this to the skribe, calibrated on the odd folios only, and it is the first rule that clearly improves it where it wasn’t fitted: on the even folios from 65 to 69 of 82 measures (median error 0.089 to 0.068). On the whole text it gets 78 of 82, and no other rule broke. Whether this first-line register is a heading (plant names?) or just a habit is open.
It has been updated in Github.
But the test showed something else: the first line of a paragraph is written differently. Besides p and f, it uses sh instead of ch much more (35% against 27%, in all five hands), t instead of k (43% against 35%), fewer ee and rarer words. I added this to the skribe, calibrated on the odd folios only, and it is the first rule that clearly improves it where it wasn’t fitted: on the even folios from 65 to 69 of 82 measures (median error 0.089 to 0.068). On the whole text it gets 78 of 82, and no other rule broke. Whether this first-line register is a heading (plant names?) or just a habit is open.
It has been updated in Github.