Jefficus, this question, and the whole discussion, are why this is a rule of thumb, and not an exact rule. It's an interplay between length, leverage, limb to string angles, etc...and has to do with Badger's mass formula. It always needs tiny adjustments, often to width.
Have you read the TBB's? baker has some really good examples of how this works with his logical extreme Andaman/Holmegaard bow (which we are now more correctly calling a Mollegabet rather than Holmegaard).
So, imagine a pyramid bow, with limbs as long as draw length. You can make that bow shorter IF you make it wider. BUT, if you keep doing this, ever shorter and wider, at some point the shorter bow gives bad leverage, bad string angle, and thus stacks more and stores less energy. Likewise, you can go longer and narrower until you start to get efficiency losses due to limb length, weight, and string weight. Eventually you are under-straining the bow, unless you adjust tiller.
Now consider a Mollegabet, with wide, flat, bending inner limbs, and the stiff, narrow outer limbs, same thing; total limb= draw length. This style has very favorable leverage due to the long stiff tips, and will hold that string angle better than the pyramid, BUT, since only the inner limb is bending the risk of extra set needs to be offset by moving more mass (width) down closer to the handle. If total limb length stays the same, but you make the outer limb longer, you have to keep moving more of that mass to the inner limb. The string angle stays more favorable because of the lever, but again as you change the ratios you hit a law of diminishing returns where you can't eliminate enough tip weight in the outer limb, or add enough width to the inner limb to help.
Next, imagine that pyramid bow from before, that was too short. If you start adding a longer and longer handle, eventually, you regain your string angle and leverage, but it takes MORE added handle length to do this than it would take added LIMB length, because of where the bend is happening out at the tips. It's whip-tillered. You can still have problems with the string angle.
Baker's pecan bow was basically a stiff long middle, WIDE bending midlimbs, with long stiff tips. Almost a hybrid of the Sudbury bow with Mollegabet outer limbs, sort of. In his case, more total limb length let him get away with less bending limb length, and PUT THAT BEND WHERE IT DID THE MOST GOOD.
Now, this gets really too general, but here are some "rules" Bending enough limb LENGTH reduces strain, which reduces set. Bending in the right spots minimizes the EFFECTS of set. Bending enough limb WIDTH can compensate for bending less length, as far as set, but not as far as string angle. Bending bows too much close to the handle can induce high set. Bending bows out by the tips can reduce leverage. Bending anything anywhere too much is bad, and adding length can counter most of these problems within reason, but creates other problems.
So, the original limb length rule gives us a starting point: a decent balance between leverage and string angles, enough wood to bend, predictable set, and options about where to place limb mass and various bend radii, in several styles of essentially straight-limbed bows.
