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

[npcompl33t]

It would be interesting to see a graph of  f p counts plotted against their paragraph word size. A flat graph would show the 'top room' effect to be true.

More interesting may be a plot of parags as dots with (count of puffs on the head line) vs (total glyphs on head line) as coordinates, color-coded by section.   The "nothing interesting" hypothesis would have all the dots on the same straight line, except for sampling error.  Anything else would be interesting...

Here you go, its hard to really tell due to the number of P/F being fairly constrained. Total rise modestly with paragraph size.