Another interesting question, using #=IVTFF Eva- 1.7 from voynich.nu I get the following statistics for the paragraph initial character pairs:
122 + 122 = 244 or ~34% of 711
34% of all paragraphs start with either po or pc (i.e., pch) in exactly 50/50 proportion. If these two combinations pick a starting letter in the table, which two letters would start new paragraphs in exactly equal proportions, and would start 34% of all the paragraphs?
Edit:
For comparison the first letters of the paragraphs for Roger Bacon's Opus Majus:
Nothing similar to a 50/50 distribution covering 34% of the whole set. Note that I filtered for capital latin letters, because a lot of paragraphs in Opus Majus start with a bracketed number. But if I include those as well, the distribution is even farther from that in VMS.
Code:
# file version
> cat zl.txt | head -n 1
#=IVTFF Eva- 1.7
# number of paragraphs
> cat zl.txt | egrep -o "<%>.." | wc -l
711
# top ten combinations of the first two characters for paragraphs
> cat zl.txt | egrep -o "<%>.." | sort | uniq -c | sort -r -n | head -n 10
122 <%>po
122 <%>pc
56 <%>tc
43 <%>to
42 <%>ts
33 <%>ps
26 <%>ko
24 <%>qo
22 <%>kc
20 <%>pa122 + 122 = 244 or ~34% of 711
34% of all paragraphs start with either po or pc (i.e., pch) in exactly 50/50 proportion. If these two combinations pick a starting letter in the table, which two letters would start new paragraphs in exactly equal proportions, and would start 34% of all the paragraphs?
Edit:
For comparison the first letters of the paragraphs for Roger Bacon's Opus Majus:
Code:
> cat opmaj.txt | egrep -o "^[A-Z]" | sort | uniq -c | sort -r -n | head -n 10
148 C
86 D
70 I
34 P
22 S
21 V
18 H
16 E
15 Q
6 MNothing similar to a 50/50 distribution covering 34% of the whole set. Note that I filtered for capital latin letters, because a lot of paragraphs in Opus Majus start with a bracketed number. But if I include those as well, the distribution is even farther from that in VMS.