The evolution of P/F vs T/K raises important questions

[Jorge_Stolfi]

I'm sorry Jorge but sincerely I cannot see how it's still possible to believe that p/f are variants of k/t (or ke/te iirc).

However:

• Graphically, it does look like k/t/p/f are a single class of glyphs.
  
• In every word structure model, p/f occur in the same place as k/t.
  
• They even occur in platform versions, like k/t
  
• 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.
  

For all these facts, it is hard not to draw the conclusion that p/f are variant versions of k/t, or glyph combinations that include k/t, which can be used (not obligatorily) for those three purposes: (1) parag initials, (2) highlighting head lines, and (3) highlighting some special words, or words in some special contexts.

In particular, the observations above seem to exclude the theory that they are letters of the alphabet with distinctive sounds or meanings.  That is, opchedy must be equivalent to some other "lower case" word, that is, a word without puffs-- probably one that begins with o and ends with chedy. Except that perhaps the version with p is used to indicate respect, importance, or some special grammatical role (like sentence starts in English spelling, or nouns in German spelling)

There are multiple lines of evidence now that they behave differently

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:
   
A "lexeme" is a "word type", that is, an abstract string of Voynichese glyphs; while a "token" is a specific word in the text (a specific occurrence of a lexeme) 

As you can see, the columns for p/f are very similar to those of k/t, and ditto for the platform variants.  That is true whether one counts the frequencies among lexemes (counting the digraphs in each lexeme only once) or among tokens (counting repeated uses of the lexeme as multiple occurrences of its digraphs).  The differences are not significant, considering all possible sources of noise in the data. The main differences is that puffs are somewhat more likely to be preceded by word space (□) than ordinary gallows; which then means they are smewhat less likely to be preceded by any of the other glyphs.  This difference may be explained by the use of puffs as parag-initials. 

On the other hand, the frequencies of glyphs that can follow p/f are significantly different than those of k/t:
   
Actually the main difference is that p/f are never followed by e; which naturally enhances the frequencies of other glyphs -- specifically, Ch. 

On the other hand, the next-glyph distributions for CTh, CKh, CPh, and CFh are essentially the same.  (I bet they are like the next-glyph distribution of Ch).

A pity that I did not think of adding the rows for ke and te; but surely you can find that data elsewhere.  In fact, please don't bother questioning those tables; there are many things I would change if I were to recompute them today.  For one thing, I would do the analysis separately for each of the big sections (Herbal-A, Herbal-B,Bio, and Stars), because each seems to use puffs in somewhat different ways. And I would consider only pure parag text, excluding anomalous pages, labels, etc.  And I would use a different parsing of EVA into elements, and different "error correction" (like ir -> iin, Ih -> Ch, CTHh -> CThe, etc.).  So, rather than questioning these tables, I urge you to create your own.

All the best, --stolfi