Its a bit difficult to explain but I will try. SRM are correct with their explanation. Imagine you have the crank sitting on the bench in front of you. First you would put the shims on, then the bearing, then the steel spacer [the one the seal runs on] then the splined sleeve and finally the nut. Once the nut is done up all these parts are trapped between the crank and the nut. This is why the nut must be done up tight to stop any of these parts being able to move. With a roller main bearing the end float is critical as even with the nut done up tight the whole crank assembly can move sideways. The early [plunger] engines had a ball bearing main and provided the outer of the bearing is tight in the crankcase the crank is held in position by the bearing. Some people use ball a bearing main in the later motors that originally had roller main and there are arguments both ways on this subject. Anyway, back to the cush drive. The sprocket and lobed piece and spring are independant of all the parts on the crank. Even when the spring is fully compressed the lobed part can not ride up and over the other lobed part. The idea of this is that as the two lobed parts work against each other, when you accelerate the lobed parts climbs up its opposing lobe and when you are running on a constant throttle it settles back to its normal position. I should also mention that sometimes when you dismantle these things you will find shims between the steel spacer [the one the seal runs on] and the splined sleeve. These are to align the sprocket to the clutch. If you have a six spring [plunger type] or a four spring type clutch it does pay to check the alignment of the sprockets and use shims if necessary. If you have the other six spring clutch [external springs with nuts on them] it doesnt really matter as the chainwheel moves anyway. I have that type of clutch in some of my bikes and have got them to work well but I dont like them, poor design I reckon. Well I am not sure if this will help or just make it all more confusing. Brian.