oh, it's a bug.
That is speculation 
good sir, i would take deduction based on facts and reasoning over blind faith any day.

what possible purpose would there be to ignoring the
most significant byte of a stat? while you see "speculation", i see borderline infatuation with Square. if they'd spilled sake on their keyboard during development, would you worship the stain 15 years later?
Here we go again.
i don't recall this aspect being argued here before. in fact, it seems to have been agreed upon as a bug from the 2004 GameFAQs threads forward.

=======================
Xardas:download this patch and document if you haven't already:
http://slickproductions.org/forum/index.php?action=dlattach;topic=2179.0;attach=1787so with that patch applied, do this: at C2/862D, put "JMP $80E9" (4C E9 80). done.
OR paste the new Knife code into the relevant part of the Chicken Knife function, which starts at C2/8626. like you said:
So is there any possibility to copypaste the M from knife formula to Chicken Knife formula? Since the speacial part of Chicken Knife lies in Attack Power, not the multiplier.
that'd involve copying the 18 bytes at C2/810B (in "Delta 3 to Fractional M.txt") to C2/864F.
that said, an issue:
1) that alternate patch apparently uses my idea of flooring of M at 128, even though we'd agreed that Jorgur's method of doing away with Minimum M was better. probably a copy+paste mistake from older patch versions, or being lenient on how "alternative" the patch is.
to remedy it, i recommend opening "Fractional M 1.2.txt" (found inside
http://slickproductions.org/forum/index.php?action=dlattach;topic=1975.0;attach=1312), and pasting:
a. C2/8A0D , 9 bytes
b. C2/940C , 8 bytes
(use the bytes that include NOPs)