I've searched and can only find one comment on how to measure the chain stays and that was from the middle of the BB to the middle of the rear axle?

I ask as I have an old Schwinn Varsity (not the standard guys model) that measured from middle of BB to middle of the rear axle is 18 inches/45.7cm. Is that possible or am I doing something wrong? I'm measuring the stays to figure out what set of panniers to buy to put on the back. Here's a pic of the old girl:

What would the axle have to do with anything, as many drop-outs are horizontal and the axle can be moved around according to adjustment. The diagram here should help you. The point that this builder uses is the apex of the end of the dropout. This should IMHO work the same for vertical d.o.'s as well as track ends. You may get other opinions.

And that's the problem---other opinions. You hear that 'x bike has chain stays of'....but measured from where to where? Looks like the middle of the BB is a good starting point, and the apex of the dropout makes sense. Of course, that's about the middle of the axle.

My belief is that chainstay length is center of bottom bracket to center of rear axle. If the frame allows different rear axle positions, like with horizontal dropouts, then the (effective) chainstay length is adjustable.

I think this is the correct answer.

With horizontal dropouts, the chainstay length is effectively variable, so you measure from the bb axle center to the rear axle center wherever the rear axle happens to be. For years, Schwinn has used the axle in the center position in horizontal dropouts as the nominal chainstay length.

IMO, this is a good call and it is at least a better specification in that it's defined as a fixed center to center. An arbitrary "effective" chain-stay length would not be much use to say reproduce a specific geometry — frame building. Depending on where the axle is, the angles will correspondingly change — not by much, but they will change. The OP is just trying to fit a rack, so I can't imagine that such fine points will matter. But for argument's sake, I think the Vitus drawing above is a better spec. It seems to show the apex of the the d.o., but even if it is defining the apex as being the axle center (as loneviking16 says), when the axle is cradled at the extremity of the d.o.'s, it is a workable, fixed specification.

Perhaps the members over on the frame-building forum would really have the final say. This stuff must be second nature to them.

I like the Schwinn drawing. I am noticing that the "frame size" is measured as C-T rather than C-C — a real source of confusion