This table is silly, and doesn't match the design goal of making stats count more.

(0,3)
(1,5)
(2,7)
(3,8)
(4,9)
(5,10)
(6,11)
(7,12)
(8,13)
(9,15)
(10,17)
(11,19)
(12,22)
(13,25)
(14,28)
(15,31)
(16,34)
(17,37)
(18,40)
(19,44)
(20,49)
(21,55)
(22,60)
(23,70)
(24,80)
(25,85)

Beyond being in the ROM2.3 original source code, I'm not sure why it exists. Plotting the points, several are out of step with any sort of realistic curve. Most notably the section up from 17 that goes +3, +4, +5, +6, +5, +10, +10, +5.

The best curve fit I could come up with, which isn't perfect is this:

5.6e^{1.12 * 10^{-1}x}

(No idea if it will display here correctly)

25 = 92.1, 24 = 82.3, 23 = 73.6, 22 = 65.8, 21 = 58.8, 20 = 52.6, 19 = 47, 18 = 42, 17 = 37.6, 16 = 33.6, 15 = 30, 14 = 26.9 are the numbers by that formula.

Why did I do this math you ask? Because I'm really wondering what would be game breaking about allowing people to practice skills/spells/songs etc. to 90% or 100% in guild. For the 15 int char, it will now take 3 or 4 practices (31, 61, 90), 17 int are mostly the same-ish (39, 77), but a really smart arial, gnome, elf, etc. can legitimately still put 1 prac per skill, or just two and get that skill advantage via higher int without dumping extra time into unfun activities, or other advantages through having more practices/trains for gains etc. than their stupid counterparts.

This system (I'd probably put it at 90%, but I can see arguments for 100%) you make a cost benefit analysis with your characters resources (practices) vs. your resources (time).

Regardless of whether any Immortal agrees that we should be able to practice skills to higher %'s the current table is stupid.