Well, I do have a protocol that i'm working from.
What varies between them is the amount of solution used, particularly the amount of ddH2O used.
So, with this protocol, i don't see where the variables are changing, why, and how.
7.5% gel (20.015 mL)
10% AP| 200 ul
Monomer solution (30% acrylide/bis (29:1))| 5 ml
4x running buffer| 5 ml
10% SDS| 200 ul
ddH2O| 9.6 mL
TEMED| 15 ul
12.5% gel (10.01 mL)
10% AP| 100 ul
Monomer solution (30% acrylide/bis (29:1))| 4.15 mL
4x running buffer| 2.5 ml
10% SDS| 100 ul
ddH2O| 3.15 mL
TEMED| 10 ul
Note that the total volumes for these two solutions are different - multiply the 12.5% gel by 2 to get the same final volume as the 7.5%...
I took a look and noticed BIO-RAD has a nice document about their product. I suspect the formulas provided are what I would be using to determine percentage, right?
I wonder what I would do if I did not have the BIORAD solution, however. I have to make some solutions from powder as of late.
If you didn't have the biorad solutions, you can easily make solutions from powder to be the same as the biorad ones... these are all standard protocols for acrylamide gels. I suggest you find a copy of "Molecular cloning: a laboratory manual" by Sambrook et al., and have a look through that.
Ok, I made a spreadsheet now that I know the math. This spreadsheet is for the running gel solution. What strikes me is the amount of TEMED and Ammonium persulfate used. In my protocol, it seems like having at least 10 uL for a 10 mL 12.5% separating gel solution should be used. But my calculations provide otherwise.
Not that your maths is particularly wrong, but there are easier ways of calculating the volumes of dilutions for the percentage gels. The volume doesn't need to be too exact; 0.05 ml is well within the margin of error for measuring 100 ml, so your proportion aspect should be 0.1 not 0.0999...
If you really want to know how the volumes/concentrations are derived, I will refer you to Ornstein, 1964, Ann N.Y. acad sci, 121:321-349, Davis, 1964, Ann N.Y. acad sci, 121: 404-427 and Laemmli 1970 (one of the most cited papers ever, you should have no problem finding it).