As a technique for assessing the accumulation of photocopying errors, I propose that a given test image, containing a suitable range of greylevels and spatial frequencies, be photocopied (using a standard tv ‘testcard’ would be an obvious approach). That copy should be itself be copied. Let the process continue for a fixed number of times (I’ve typically used 100 such cycles).
The final image will show up regions in which the device has been introducing noise of various sorts, and thus provide a measure of imperfection in the system. In particular, try this with eg a name badge bearing your photograph. After about 30 recursive copies, you will find yourself transformed into Ben Grimm of the Fantastic Four.
Even if it is hard to make absolute measurements of image quality this way, it’s possible to use it to understand something about the relative strengths and weaknesses of different copiers -a kind of crude modulation transfer function test.