f and p appear predominantly in the first lines of paragraphs

[Jorge_Stolfi]

[Is this the right thread for this post?]

Here are some statistics about gallows and parag head/body lines. I suppose that 65'534 people already computed and posted them, but maybe they will be new to some.

Let "puff" mean a one-leg gallows (p or f), and "tike" mean a two-leg one (t or k)

I took the parags text in each section ("hea", "heb", "cos", "zod", etc.), and split the lines into "head" (the first line) and "body" (all the other lines).

For each of these sets of lines I computed:

• nl = count lines in the set.
  
• nlp = count of those lines that contain puffs
  
• %lp = percentage of nlp relative to nl
  
• nlt =  count of lines that contain tikes
  
• %lt = percentage of nlt relative to nl.
  

• nc = total count of EVA characters [a-z] in that set
  
• ncp = count of puffs among those chars
  
• %cp = percentage of ncp relative to nc
  
• nct = count of tikes among those chars
  
• %ct = percentage of nct relative to nc
  

• %cg = percentage of ncp+nct (total gallows) relative to nc.
  

Here are the results:

                                        body                             
       -------------------------------------------------------------------
  sec      nl   nlp  %lp    nlt  %lt      nc    ncp  %cp    nct  %ct   %cg
  ---  -----  ----- ----  ----- ----   -----  ----- ----  ----- ----  ----
  hea    1042    78  7.5    991 95.0   31256     99  0.3   2972  9.5   9.8
  heb     322    58 18.0    320 99.3   14201     92  0.6   1548 10.9  11.5
  bio     700    90 12.8    679 96.9   29464    144  0.5   2797  9.5  10.0
  str     751    44  5.9    727 96.7   37085     59  0.2   3458  9.3   9.5
  unk     272    61 22.4    267 98.1   13353    114  0.9   1297  9.7  10.6
  pha     190    15  7.9    188 98.8    9520     35  0.4    792  8.3   8.7
  cos     129    17 13.2    101 78.2    3758     19  0.5    320  8.5   9.0

                                        head                             
       -------------------------------------------------------------------
  sec     nl    nlp  %lp    nlt  %lt      nc    ncp  %cp    nct  %ct   %cg
  ---  -----  ----- ----  ----- ----   -----  ----- ----  ----- ----  ----
  hea    167    129 77.2   153 91.5     6307    289  4.6    391  6.2  10.8
  heb     51     47 92.0    49 96.0     2577    143  5.5    158  6.1  11.7
  bio     40     32 79.9    40 99.9     1757     73  4.2    160  9.1  13.2
  str    331    258 77.9   331 99.9    19671    621  3.2   1816  9.2  12.4
  cos     45     24 53.3    38 84.3     1357     37  2.7     98  7.2   9.9
  pha     33     27 81.7    32 96.8     1850     62  3.3    127  6.9  10.2
  unk     33     32 96.8    30 90.8     1726    112  6.5     91  5.3  11.7

Some things worth noting, considering the major sections ("hea", "heb", "bio", "str"):

• Less than 20% of body lines have puffs, whereas more than 77% of head lines have them.  Therefore puffs are not exclusive to head lines, but are much more common in them.
  
• The percentage of puffs relative to all EVA chars (%cp) is ~3.5% higher on head lines than on body lines, while the percentage of tikes (%ct) is ~3.0% lower.
  

If we assume that the tikes have phonetic value, and that the sounds they represent have the same frequencies in both sets (an assumption that is weakly supported by the similar %ct across sections), the last observation could be explained by assuming that most of the puffs on head lines are embellished versions of the tikes (or of tikes combined with other glyphs, like te or ke), while a few puffs are substituting for other characters. 

However, word frequencies are probably different on head lines, and thus the frequency of sounds may well be different too. Thus the above conclusion should be taken with a ladle of salt...

All the best, --stolfi