QTR-Quadtone RIP

I have struggled somewhat further and found a way to reduce my error 
rate with the color picker. So that works, but I still appreciate a 
pointer to a way to semi-automate this process like with the Measure 
Tool.

Now I have run into another issue in the linearization process. I 
scan the L-values with the Eye-one and fill the data manually in the 
linearization table of QTR. When done, I press "create curve", but 
then I get the error message "Invalid linearize curve - not 
constantly increasing".

Of course that's true, since the L-values decrease from approx. 95 
(white) to 15 (black), but according to the user guide that should be 
fine. I quadruple-checked the values: it is an increasing series, no 
strange wiggles. As a test I tried to paste value in the reverse 
(wrong order), so with increasing values. Same error message. 

I did some further dummy tests: using (L'/100) (to get the values in 
the order of what could be densities, with L' the inversed order of 
the measured L-values). Same error. Strangely enough using (L'/100)^2 
did NOT give an error. Of course the resulting curves are of no use 
whatsoever. These (admittingly strange) experiment gives me the 
suspicion that there is a computational bug in QTR. Or could I be 
doing something wrong?
 
If it's a bug, I can possibly circumvent it by transforming my L-
values into densities. But I could not find a formula. Anyone has it 
available?

Joost