Liaisons in the Voynich text

[nablator]

The interesting thing for me is the fact that the width of spaces is not random: no language has predictable variable spacing between words linking the patterns inside words to the patterns across word breaks. As crazy as it seems, there is a method to the madness of irregular spaces after all. Repairs don't explain them (unless I misunderstood how they are supposed to be applied): something is missing.

I had a "zigzag path" substitution cipher/code idea 10 years ago that uses two types of spaces as a switch between 2 different plaintext chunks for the same Voynichese word, but spacing is far too irregular to be reliably categorized, so I supposed, very speculatively, that the spacing irregularities might be a remnant of the replacement of many spaces with some other, less ambiguous, binary information, like the k/t gallows: a re-synchronization mechanism. So I started a full manual transliteration project, hoping to find evidence of a causal link between gallows and variable spacing, but I did not expect more than that, a very good correlation between patterns across word breaks and space widths. Amazing!