f76r , text-only page in Q13

[RenegadeHealer]

Now if this parallel is valid in any way, that somehow gallows are used because this is the top line, then the implications are significant: it would mean that these gallows replace something else that takes up less space.

Koen, I've been much taken with this idea ever since Geoffrey Caveney brought it up when he presented his Judeo-Greek theory. One day I played around on Voynichese.com looking for vords containing [p] or [f] which have a large number of tokens. Then I tried replacing the [p] or [f] with a different glyph or string of glyphs – I think [d], [dch], and [chd] were the main ones I tried. I remember feeling amused and encouraged that many of these substitutions yielded common vords as well.

I'd like to test the hypothesis that [p] could be equivalent to another glyph or string of glyphs more seriously and systematically, though. As someone with a lot of experience doing statistical analyses on the VMS, would you mind offering me some feedback on my proposed methodology:

1. Tabulate all the types containing [p] with >4 tokens, in order of most tokens to fewest.
   
2. For each type, run a search that replaces [p] with a wildcard of any length. Tabulate the total number of tokens found in the next column, next to the count for the original type containing [p]
   
3. In the next few columns, break down the tokens by the character or string which replaces [p], with a count for each
   
4. Sort each row so that each column represents the same substituted character or string.
   

If this hypothesis is false, then I should expect to see:

• Orders of magnitude fewer occurrences of [n*n] than [npn], or none at all, for most types containing [p]
  
• No pattern to the glyph or string which most often substitues for [p]
  
• Ratios of [npn] counts to [n*n] counts that are nowhere near the ratio of the number of first lines of paragraphs to the number of non-first lines of paragraphs
  

Where I'm not sure how to proceed is in determining statistical significance. Obviously I wouldn't expect the ratios to be uniform. But how different is too different to be expected by chance, if [p] really is interchangeable with something else? I wish I'd paid more attention in statistics class. In my medical school, students who weren't going for an additional Masters of Public Health (MPH) degree or doing a doctoral research thesis only needed to take a brief, cursory 2-credit course in statistics, and I've barely used it since.