Considering the rather terse introduction I gave to these house rules, I thought maybe a little more insight was in order.
My main goal was (and has been for a long time) to come up with a tweak of the magic system in Chainmail that was unpredictable, wide open, and carried an element of danger.
I started with the casting table straight from Chainmail. I took the numbers for a Seer casting a complexity I spell and added +1 to all the numbers. That was my baseline. The extra +1 to the numbers offsets the fact that most any character worthy of being called Wizard should have at least a +1 mod to the casting roll. Next, I looked at the spells-per-day table and plugged that baseline in whenever a new spell level was attained. So, for example, the baseline is used for Rank III spells for level 5 Wizards. Then, I just extrapolated the numbers up or down as needed for spells of higher or lower rank.
I could have just used some formula, but that requires math on the front-end, then more math at the table as mods are considered. Besides, tables are much more old-school. There was another reason, too. I wanted the raw number rolled to mean something when compared to the roll that was needed. Specifically, the chance to lose the spell and the chance to fail catastrophically. That becomes harder to track at the table when the casting roll is just based on pure math. Not terrible, but enough to slow things down.
Speaking of those raw rolls, with the 2d6 bell curve, keeping track of the original number needed was important. If your base chance to cast-with-delay was a 7-8, then you'll lose the spell on a roll of 7, which is roughly 16% of the time. If that was based on the modified roll, then bonuses would quickly make losing spells highly unlikely.
That's it for now. As many of you know, a Chainmail based game has been an elusive dream of mine for some time. That combat modification that I linked to at Howling Tower a few posts back really got me started back looking at it. Now that I have a magic system I like, the project is gaining critical mass. I have character generation ideas almost done (at least conceptually).
Taking those three components and mashing them up with whatever other subsystems I like will likely be the way to go with it, for now. I may or may not try to work up my own advancement, equipment, endgame, and other subs, but for now, I'm happy using my combat, magic, and character subs houseruled into other rulesets.