This has surely been much discussed, but you wouldn’t know it from my everyday experience. Proper spindle length calculation almost seems like a mysterious secret held within the BMG (Bike Mechanics Guild). I’ve never gotten a straight forward answer to this question from a bike mechanic, but then again, maybe I was asking the wrong mechanics.

I know there are many complicating issues involved with the correct answer; but the essential issue seems to me to be achieving a straight chainline. So I ask, in order to determine the spindle length for a blank-slate frame, what is your method?

My starting place is what was in there before, and whether I want the same, or longer or shorter. For a new crank it's a bit harder, but most used to be sold with BBs, or at least suggested spindle lengths in their specs.

In some cases where I had no idea where to start, I used a "best guess" spindle with a bearing and right cup. I'd fir the spindle and use a straight edge across the inner chainring to gauge the offset with respect to the right cup. Once I have that, I can estimate chainstay clearance or chainline, and adjust accordingly.

The above applies to square taper, and/or spline BBs with a spindle and 2 mounted cranks. More modern stuff often has a spindle fixed to the right crank, and chainline is adjusted via spacers.

I can see know that I should have provided at least a little more detail. Yes, for my purposes we are talking about tapered spindles. It seems like someone would have come up with a formula for this by now. I have an old steel frame spread to 130mm rear that I want to build with 9 speed cassette and a triple crank.

My only thought is to slip a known length spindle through the BB shell and slip on the cranks, then run a straight line from the center of the cassette to the center ring, try and measure how off the centerline is, and subtract or add. This doesn’t seem as professional a method as I’d hoped for however.

You don't have to put it into the frame. The outside surface of a right cup is 1/2 the shell width + the cup flange thickness from the centerline. (34 + 3mm for a typical 68mm BB) So it's simply a matter of working with that.

It’s this part that is the mystery to me. If no one has a better solution, my idea was to just lay an un-mounted spindle of known length gently across the open BB shell approximately correctly oriented to where it would be if completely installed, then slide the cranks on and run a straight line to the center of the cassette. The problem is, if the straight line is wide or short, it will be just a guess-estimate by how much. Surely there is a better way.

Complicating the choice is that many square taper bottom brackets are not symetrical so the drive-side spindle length change isn't simply 1/2 the difference in overall length. These remain a matter of trial and error unless the crank manufacturer makes a specific recommendation as Shimano did for the 6400 Ultegra double at 115 mm or the 1057 105 triple at 118 mm.

Octalink and ISIS bottom brackets simplfied the problem by offering only a limited number of spindles. I.e. road Octalinks (V-1) double cranks took 109.5 mm bb and triples 118.5 mm and that was all that were available. External bearing bottom brackets are even simpler since the spindle is part of the crank and there is no length choice at all.

Also, while chainline can't be ignored I wonder about agonizing over it to the nearest mm. The chain is almost never in a straight line anyway.

yes, there are two better ways to measure chainline.

Directly - lay a straightedge (if using a flat ruler, use the edge not the surface because only the edge is dependably straight) against the outer surface of the outer chainring on a tangent, and extend it back to the cassette. Allow for the offset from the straight edge to the center of the gap between the rings, (or middle ring of a triple), and that should line up with the center of the cassette.

Indirectly - use a chart like this one to determine correct chainline range, then measure from the tips of the chainring teeth to the down tube, add 1/2 the DT diameter, and adjust for the offset to center of the chainset as before. If your bike is unique or you want a more precise chainline target, measure from the inside of the dropout to the center of the cassette and subtract from 1/2 the hub's width to know the centerlind distance to the middle of the cassette.

Reading Sheldon's info on "chainline" should clear it up sufficiently. There is no need for a formula, and I doubt any formula would be versatile enough to cover all situations without making the task more complex, not more simple.

Single speeds require a more precise chainline than geared bikes- well at least it looks better if the chainline is close. My main concern with BB width is that the c-rings clear the chainstay. Often times I have a crank I want to use on a specific frame. Each crank will have different requirements for spindle length. It's much easier to figure this out by trial and error, or some guestimated measurement. Depending on frame design, and the specific crank/spindle length combo, sometimes I can't use the crank I'd like. If I can make the rings clear the chainstay, and there's only a slight misalignment to the chainline, spacers or shims on the chainring can get the job done.

Easiest for me is to have a sampling of bb's in varying lengths, and select the one I suspect to be close, for a fit test. No formula could be devised to simply cover my needs.