Here at the NCMI we're running two Nikon 9000 scanners on two Windows machines. We have modified the holders (FH-869M) in-house so that the film is sandwiched between two sheets of anti-Newton glass. We also found that the film holders/scanning system had a slight tilt downwards from SW the NE (looking from above the machine).

The bottom piece of glass is glued to the film holder, on which the film is placed, and then the top sheet of glass is placed on top by the Nikon scanner operator. The top piece of glass and film are held roughly in place by make-shift guides. Important points are

• The bottom sheet of glass is leveled relative to the scanning system - we did this by elevating the corners of the glass plate using tiny cubes of sticky tape piled in stacks. Then, in the Nikon scanner GUI, we took focus readings on the corners and in the middle a test film, until it was clear that the point of focus was relatively consistent across the entire film. All testing was performed on several films for thoroughness. Also, [EMAN1/FAQ/e2scannereval results::e2scannereval.py] proved a very useful tool for the qualitative detection of tilt, and verification measurements made using the Nikon scanning software.
• The anti-Newton glass is only coated on one side, this is the side that faces the film in our design.
• We did this for two different Nikon 9000 scanners, and found that there was roughly a consistent tilt, though varying in extremity. The tilt is only slight, so its effects are not all that great.

The glass didn't come with the holder we had to order it in. Because the bottom piece of glass is glued it stays there, but the sandwiching piece just slides around unless you think of someway to hold it still - so we just put some bits of thick tape here and there so everything stays approximately still. We had to shave some plastic from the bottom too, just using a pocket knife. It's a bit of a hack but it works fine.

I will add some pictures shortly.