During recent runs with Hants BSAOC I had been advised more than once that the rear wheel on my 1960 A10 (Golden Flash) was running well out of true. A quick check confirmed this runout to be over 0.1" (2.5 mm) side to side. Last winter I had both wheels rebuilt with new spokes, using new chromed Devon rims, which I had been assured were pretty much the best available. So it was back to the rebuilder for a true up pronto. What he found was really only a little run-out, most of which he skilfully tweaked out as I watched. So imagine how dissapointed I was when I replaced the wheel and measured with a dial guage 0.12" run-out. Now the wheel truing rig used its own half spindles and I assumed at that time that somehow this factor was introducing the error. But before a further trip with the loose wheel, I had the thought to mount the wheel in each of the four available sprocket positions and observe the run-out. This proved enlightening. The next position gave a run-out of 0.09", the third just under 0.04", and in the final slot it was back up to 0.08".
My conclusion is that the sprocket mounting hole circle must surely be eccentric, thereby offsetting the wheel axle a variable amount in each position. Just a 20 thou offset at the sprocket would translate into the 120 thou at the 19" rim. Now with the sprocket in the best position I rapidly managed to tweak the maximum run-out to less than 20 thou (to my surprise and pleasure). Then I permanently marked the wheel hub and sprocket for future matching. I suppose this is just a further example of innaccurately made pattern parts; although the sprocket fitted must be 30 years old it is certainly not a BSA original. What is most surprising is that both times the wheel builder must by chance have had the sprocket in the same position, and I had selected the worst of 4 positions at random, so a 1 in 16 chance. I had not seen this issue referred to anywhere so it must be worth passing this on to other riders with this type of wheel, as also fitted to A7 and B31/33 of course.