[quote="Jorge_Stolfi" pid='94367' dateline='1793558449']
And? Latin alphabet 'p' 'q' 'b' and 'd' also look graphically to be a single class of glyphs (one circle plus one vertical bar), even more than EVA k/t/p/f do. Would you deduce for instance that Latin 'p' 'q' can be replacements for 'b' and 'd'? But I can surely agree with you that the distinctiveness of k/t/p/f suggests they belong to the same class, but "k/t/p/f belong to the same class" does not imply, nor is evidence for, "p/f are a replacement of k/t"
[quote="Jorge_Stolfi" pid='94367' dateline='1793558449']
Yes, and this indeed also points to p/f/k/t probably having something in common, but it does not follow that because "p/f/k/t have something in common" then "p/f are a replacement for k/t".
[quote="Jorge_Stolfi" pid='94367' dateline='1793558449']
This only estabilishes that p/f are common only in a limited section of the manuscript, noone doubts that, but it does not follow that they are replacements for k/t. You should at least show that where p/f are common k/t are rare, this would be a more promising evidence. But, given "In every word structure model, p/f occur in the same place as k/t", it's expected more p/f words means less k/t words. So it's not an evidence easy to find I guess.
[quote="Jorge_Stolfi" pid='94367' dateline='1793558449']
[quote]There are multiple lines of evidence now that they behave differently[/quote]
What are those lines of evidence? Here is a table I computed years ago with the frequencies of the glyphs that can precede k/t/p/f and their platform variants...
[/quote]
Looking at tables is suggestive, but relying on our own visual interpretation of a single table can be misleading. I have already posted the root-mean-square distances of p/f, k/t, and k/p and t/p in post [url=[quote="Jorge_Stolfi" pid='94367' dateline='1793558449']]#30[/url], I would really rather not have to post it again, but here's the quote:
[quote]Anyway: p/f and k/t do indeed seem to be paired, at least in Herbal A: the left context distance between p and f is 0.08 and the right context distance is 0.15 (for k/t: 0.07 and 0.12; for reference ch/sh are 0.23 and 0.08), rather small. But between k and p (the most frequent glyphs of the two groups) the left-context distance is 0.22, the right-context one is 0.34 (between t and p: 0.28 and 0.25).[/quote]
As you see the left-context distance of k/p, 0.22, is slightly lower than ch/sh at 0.23, which are often recognized to be 'similar' characters. But in effect, it's actually a moderate distance: ie. k/t at 0.07 are much more similar than 'k' to 'p' or 'ch' to 'sh'. To give another reference 'o' and 'y' are at 0.23. But from the point of view of the left-context distributions, yes, they might be compatible with p/f being replacements of k/t. Not exactly evidence for, but neither definetely against.
However the right-context distance of k/p is 0.34: it's not that big too, but now it's much higher not only than ch/sh (0.08 (*)) and k/t (0.12), but is now comparable to the distance between 'r' and 's' (at 0.33) or of 'k' and 'cph' (at 0.36).
(*) indeed notice how the similarity between 'ch' and 'sh' is mostly driven by the right context, while they are moderately diverse in the left context.
So I think left- and right-context distributions do not support the idea that p/f replace some random k/t when they happen to be in the top-left corner of the page, so this is the first evidence against (and notice it's based on the same kind of data you used)
But there are more: @npcompl33t presented some in this thread, based on different statistics than left- and right-context. So indeed there are multiple, independent evidences against the p/f = k/t hypothesis.
- Graphically, it does look like k/t/p/f are a single class of glyphs.
And? Latin alphabet 'p' 'q' 'b' and 'd' also look graphically to be a single class of glyphs (one circle plus one vertical bar), even more than EVA k/t/p/f do. Would you deduce for instance that Latin 'p' 'q' can be replacements for 'b' and 'd'? But I can surely agree with you that the distinctiveness of k/t/p/f suggests they belong to the same class, but "k/t/p/f belong to the same class" does not imply, nor is evidence for, "p/f are a replacement of k/t"
[quote="Jorge_Stolfi" pid='94367' dateline='1793558449']
- In every word structure model, p/f occur in the same place as k/t.
- They even occur in platform versions, like k/t
Yes, and this indeed also points to p/f/k/t probably having something in common, but it does not follow that because "p/f/k/t have something in common" then "p/f are a replacement for k/t".
[quote="Jorge_Stolfi" pid='94367' dateline='1793558449']
- The first glyph of a parag is very often a puff.
- In some sections, 4-6% of the glyphs on every headline, besides the initial, are puffs.
- Apart from the head lines, puffs are rare -- maybe 2% of the words have them.
- They are rare even on the second lines of parags that have very short head lines.
This only estabilishes that p/f are common only in a limited section of the manuscript, noone doubts that, but it does not follow that they are replacements for k/t. You should at least show that where p/f are common k/t are rare, this would be a more promising evidence. But, given "In every word structure model, p/f occur in the same place as k/t", it's expected more p/f words means less k/t words. So it's not an evidence easy to find I guess.
[quote="Jorge_Stolfi" pid='94367' dateline='1793558449']
[quote]There are multiple lines of evidence now that they behave differently[/quote]
What are those lines of evidence? Here is a table I computed years ago with the frequencies of the glyphs that can precede k/t/p/f and their platform variants...
[/quote]
Looking at tables is suggestive, but relying on our own visual interpretation of a single table can be misleading. I have already posted the root-mean-square distances of p/f, k/t, and k/p and t/p in post [url=[quote="Jorge_Stolfi" pid='94367' dateline='1793558449']]#30[/url], I would really rather not have to post it again, but here's the quote:
[quote]Anyway: p/f and k/t do indeed seem to be paired, at least in Herbal A: the left context distance between p and f is 0.08 and the right context distance is 0.15 (for k/t: 0.07 and 0.12; for reference ch/sh are 0.23 and 0.08), rather small. But between k and p (the most frequent glyphs of the two groups) the left-context distance is 0.22, the right-context one is 0.34 (between t and p: 0.28 and 0.25).[/quote]
As you see the left-context distance of k/p, 0.22, is slightly lower than ch/sh at 0.23, which are often recognized to be 'similar' characters. But in effect, it's actually a moderate distance: ie. k/t at 0.07 are much more similar than 'k' to 'p' or 'ch' to 'sh'. To give another reference 'o' and 'y' are at 0.23. But from the point of view of the left-context distributions, yes, they might be compatible with p/f being replacements of k/t. Not exactly evidence for, but neither definetely against.
However the right-context distance of k/p is 0.34: it's not that big too, but now it's much higher not only than ch/sh (0.08 (*)) and k/t (0.12), but is now comparable to the distance between 'r' and 's' (at 0.33) or of 'k' and 'cph' (at 0.36).
(*) indeed notice how the similarity between 'ch' and 'sh' is mostly driven by the right context, while they are moderately diverse in the left context.
So I think left- and right-context distributions do not support the idea that p/f replace some random k/t when they happen to be in the top-left corner of the page, so this is the first evidence against (and notice it's based on the same kind of data you used)
But there are more: @npcompl33t presented some in this thread, based on different statistics than left- and right-context. So indeed there are multiple, independent evidences against the p/f = k/t hypothesis.
![[+]](https://www.voynich.ninja/images/collapse_collapsed.png)