As a further to the thread attached to this one https://netrider.net.au/forums/viewtopic.php?t=20532 I thought i'd put together some pictures to help understanding. The first picture here shows your average engagement mechanism for a SYNCHROMESH gearbox such as you might find in a car - note that the synchromesh system is not included because it hides what i'm trying to show you. The gear (dark blue) is free to rotate on bearings on the shaft (light blue). The gear does NOT move at all along this shaft in a constant mesh gearbox (practically everything these days except most reverse and some first gears). Neither does the gear on a parallel shaft that mates with the gear I have shown you. The engagement ring (red) is fixed to rotate with the shaft and has a large number of small sharp teeth (yellow). To engage a gear, the shifter mechanism (light blue fork) pulls this ring that slides along the shaft until the teeth enter the slots in the gear (light brown). This is how the shaft from the clutch drives each gear in turn depending on which dog ring is engaged with which gear. Take note of how little relative motion you have to make sure that the teeth engage in their slots. The next picture shows the arrangement of a NON-Synchromesh gear The same system is in place, but the engagement dogs are much bigger, but there are only perhaps 3 or 4 of them. Their corresponding slots in the gear (light brown) are huge and allow a great deal more relative motion between selector ring and gear compared to a synchro box. It is this behaviour that makes a non synchro box so easy to shift without the clutch compared to a synchro box. Note that the relative speeds of the input shaft that drives the dog, and the output shaft (wheels) the turn the gears must be as close together as possible in both cases for a nice smooth shift. Crunching sounds that you get from a poor shift are from the little teeth (the dogs - yellow in both cases) crashing against the edge of the slots in the gear but not actually going in. Over time, you will wear away the bearing surface of these dogs (the part that presses against the side of the slot and turns the gear). Instead of a nice parallel slot, you get a slope on it - if you apply enough force or the slope is big enough, the dog ring will be forced out (ie back into neutral or no gear engaged) without you doing anything - obviously this is bad. I should probably also add that I have depicted, say, the input shaft of the gearbox. The output shaft runs parallel to the input and looks the same EXCEPT all gears are fixed to the shaft and turn with it, and there are no selector rings. EDIT: Have a look at the difference in the designs of the little pins (the dogs - in yellow) and their respective slots above. The chance you have of seating those dogs fully home in their slots is directly proportional to how easy a box is to shift without the clutch. The problem for a driver/rider lies in how close you can get the relative speed of the dog ring and the gear slots - obviously, the bigger the margin for error, the easier the process is for the user. In the given example, you have about 10 degrees of relative motion for the synchro (car) box, but you have about 80 degrees of relative motion for the dog (bike) box to seat the dogs BEFORE THE LOAD COMES ON. Granted, if the differences in speed are great, then you get a big jerk as road speed and engine speed are suddenly forced to be the same - but this is NOT the same as ease of shifting as far as getting the next ratio engaged. (multiple gears exist on the same shafts, and all are meshed together all the time - this arrangement I have drawn is a constant mesh type - only the selector rings lock a gear to the input shaft so that it drives it mate on the other shaft. The rest of the gears in the train that are not being used just spin on the input shaft at a speed proportional to the numbers of teeth they have compared to their mate.