It might really work.

So after spending most of yesterday installing ImageJ, reading through its documentation, playing around, recording macros and trying to figure out how to get a clean, reliable output of data with actually useful information... I think I have figured out a way to do it.

Well, this still entails having to make a visual survey card with very good contrast between card and yarn and then scanning it in at a suitably high resolution before being able to evaluate the threads, but that is still much less work and equipment than would be needed to do it the usual way (which is climatising the yarn, then cutting it into small, exactly measured snippets, then weighing each of the snippets and calculating the tex value, and finally getting an overview of how regular or irregular it is), plus it's a non-invasive method.

So now I only (hah!) need to figure out a way to implement my ideas and make things nice and reliable so that they work with all the visual survey cards, plus devise some macros that deliver the data to files in a sensible way to make it all less work and more click-and-be-happy.

Yay. And this, by the way, again proves my point that whatever skill you pick up, you will eventually be able to use it for work when you are an archaeologist. Including basic programming knowledge.