I may be wrong here (we need an expert engineer's opinion) but I suspect that "Lustrox" worked like this - it would have maximum abrasive effect on tight, closely fitted items (such as a tight bushed bearing) and correspondingly less effect on looser fitted mating surfaces. Even Duke himself recommended caution in using "Lustrox", but in the case of Chinn's Fox 36RX, it may just have needed that pinch of powder to free-up the bearing, whilst having little effect on the other fits.