Deckel FP2 Vertical Column Scraping

After numerous iterations of blueing, scraping measuring and head scratching, I placed the saddle back and did some measurements with reference to the original machined surface on the vertical columns, here the results. granite parallels where placed between the ways on the machined surface, and a straight-edge spanning over it, and finally the measurement was taken from the top of the saddle to the straight-edge which are now in the same plane.

Y Represents the distance in cm from the top of the main column body.

x(0) represents the right outer measurement (over right column) , x(1) the center between columns, and x(2) the left outer measurement( over left column). The error is not nearly as big as originally measured, but it seems like the saddle nose dives into the worn area on the right column caused by the rusting cover plate. Then climbs back out and rises on the left top column section by 0,008 mm. I have repeated the measurements a few times to be certain. More scraping work on the left upper column will bring the saddle movement back within spec. no need to cast moglice or turcite. I will probably just scrape the back gib a bit to compensate. The measurement is in line with what I measured when testing the right back gib for parallelisms to the way. 

Visualization with https://plot.ly/create/

	Left Front	Center	Right Front
0	0.004	0	0
10	0.009	0.001	0
15	0.006	-0.002	0
20	0.003	-0.002	-0.003
25	0.002	-0.002	-0.002
30	0.002	-0.001	-0.001
35	0.003	-0.001	-0.001
40	0.004	-0.004	-0.006
45	0	-0.004	-0.008

Sleeping over it, made me think, there might be other reasons why I get the measurements.

• it might related to leveling of the main column body. 
• The right back gib was not mounted, which could indicate a slight tilt in the left column way
•  The worn area on the center of right column might cause the saddle to lift the right front part when the left rear dips into the worn area.

I have redone the measurement of the column height with reference to the machined surface, and get very similar results, thus at this point I will slightly scrape the top part of the left front column and call it job done. Next step would be to start with the saddle.