This must have been one of the longest running threads ever... and who was it who virtually said "Does it really matter.." so long as the finished article rides well.
I have been building frames since the early to mid 50s and, frankly, and perhaps I am now about to display just how "thick" I am, I have never contemplated any other measurement than that which runs along the length of the chainstay, from the centre of the bracket to the "centre" of the drop-out, in the case of a long drop-out, in order to allow some adjustment in either direction, particularly in the case of touring frames and multi-use ones ( in the 50s we used to ride our bikes out to races with "clinchers" on, then change over to tubulars for the race.) in order to accommodate diferent widths of tyres.
Most local frame-builders used to build "free" ie without a jig, and the "C-to-C" method for chainstays was universally adopted. The drop-out would be brazed into the chainstay which in turn would be tapped into the "pipe" of the bottom bracket. A tape measure would be held alongside the chainstay, the drop-out would be gently prodded into the pipe with a lead mallet until the desired length was obtained on the tape-measure. At that stage the "waste" length of the chainstay ie that amount that protruded into the bracket shell, would be scribed round, ready for cutting off with a hacksaw. A refinement of this technique was to tap both chainstays into place and fix an axle, provided with stops to set the width of the rear end, between the drop-outs. A piece of string would be run around the head-tube, crossing over in front of the seat-tube before being tied in a regular fashion to the drop-outs. The chainstays would be tapped into length and then the drop-outs checked for alignment by measuring the distance between the seat tube and the string on each side. With a little care and a few mallet-taps, an accurate back-end could be set up.
The FRench tool-maker VAR in it's admirable pocket-book "le petit livre jaune" gives examples of the dimensions of the frames ridden by Merckx, Van Impe, and Zootemelk, these dimensions being given alongside simple line drawings of a frame. In each instance the length of the chainstay is given as that length measured alongside the stay ie sloping upwards and not parallel to the ground. Well... if it's good enough for the likes of them it'll do for me. To support my claim to be correct I would cite the measurements I have taken from a very wide variety of high-end frames such as TVT, LOOK etc in cases where I had the manufacturers'spec sheet in front of me. Whenever the LOOK, TVT, Vitus etc spec said that the chainstay length of a frame was say "X cms" that measurement always was the same as the measurement along the length of the stay upwards and not horizontally.
Talk of French frames reminds me that most French manufacturers measure the seat angle by using the distance by which a line running up the centre line of the seat-tube is distant from a line rising vertically through the centre of the bracket. Using trigonometry the angle between the centre lines of the seat and top tubes can be calculated.
Just a last thought...Bearing in mind that the long slots in drop-outs such as standard Campags, Shimanos etc are never actually parallel to the ground when they are brazed into place, instead they slope gently up from front to back, how do framebuilders calculate the height of the bottom bracket, because the height will vary depending on whereabouts the wheel is clamped along the drop-out. Or am I not thinking too clearly after all these years?
Norris Lockley .. with my mind now in a turmoil of angst.. Settle UK