Thats a pity. I have a couple of old Heidenhain rod420 or 426 encoders lying around somewhere. You could have one of them.
Thanks I would be intrested in buying one to play with if you have a price in mind?

Linear encoders are generally very expensive. One cheap way to make one that should be plenty good enough to experiment with is to use a rotary encoder, put a timing pulley on it and have it run on a flat (open ended belt). Wrap the belt round the pulley with two idlers, so when the linear slide is moved it causes the pulley to rotate...I doubt it would be good enough to use in 'real life', due to small errors in the belt pitch and stretching, but worth a try I reckon.
I haven't been able to get timing belts working well for me and liked the idea of a linear encoder as there would be very little possiblity for any sort of backlash. I also thought it would be a nice way of measuring the backlash in my marchant dice router. If I had a way of documenting the backlash or flex in the router I could see how different mods effected the backlash/flex and also determine how accurate the machine is when it is air cutting in comparision to cutting say hardwood or soft foam. This would show the limits of the machine and possibly help idendify where I would need to beef it up, although this would involve a lot of trial and error.

I think the Kflop has the possiblity to output positional error, if you could plot the error over time as you cut a certain program into varying materials you could hopefully compare. Could you take this one step further with an encoder on the motor to measure missed steps and linear encoder on the gantry to measure actual postion?