[split] Measure uncertain spaces?

[Koen G]

But they also report finding that "certain" spaces tend to fall between their study's learned units, while "uncertain" spaces tend to fall within them. That seems potentially more meaningful in terms of supporting the idea of a hierarchy of "strong" and "weak" spaces.

Thank you for your contributions in clear language, Patrick, I find this very informative. 

About these uncertain spaces: the phenomenon you describe here could still be attributed to subconscious split-second decisions by the transcriber, right? The manuscript trains your brain to recognize certain patterns, and as soon as something deviates from those patterns, you have to mark it as uncertain. In my opinion, there is a good chance this effect influenced the results. When transcribing regular medieval handwriting, we need to rely on expected sequences to determine spacing all the time, especially in kind of chopped-up scripts like these. Only, in regular MSS this is non-controversial because we understand the meaning of the words.

So I wonder: let's say we take this crucial sequence of [r] followed by [a], and plot them all according to the distance (pixels?) between the [r] and [a]. Will the no-space variants be all on the left, and the space variants on the right, and the uncertain ones neatly in between? How much overlap is there? Is it a continuum? If this could be measured objectively somehow, we would have a better idea of what space, no space and uncertain space actually stand for.

(Edit: this would actually be informative for other reasons as well, like when we split up by scribe).