I've done some more testing. I scanned a perfect square and calculated the number of pixels it should be in height based on the DPI. It was accurate vertically, but about 99% of the correct distance horizontally. More alarmingly, I have confirmed the issue with non-linearity using a set square. In the first 12cm from the top right edge across towards the left, the error on my unit is about double Malcom's, and is not constant. That is to say, the gaps between the notches vary. Beyond that 12cm point, the error increases quite rapidly, and by 17/18cm, it is up to 10% out (i.e.10mm is scanned as 9mm). It would seem the issue may not necessarily lie with the motor, but how the scanner is distorting the image across its width. Are there any calibration tools which can adjust for this? Can Silverfast?