In many cases the difference like that is it may have a metric screw and the error is in the conversion to the handwheel imperial dial.
Sorry to revive an old thread, but I just discovered EXACTLY the same error on my Grizzly G0762 square-column benchtop mill.
The dial has one hundred divisions that are supposed to represent 0.001", but one revolution of the fine feed handle moves the quill axially 2.5mm (0.0984"), not 0.100". That's a ~1.6% error which makes it hard to hit dimensions to the nearest thou (generally my goal when milling).
This is a pretty big deal on benchtop mills because we can't move the table in Z. On a knee mill, I think one usually raises the table for accurate Z adjustments and tends to leave the quill locked with minimal stick-out for rigidity when milling — the quill dial is only for boring holes. On a benchtop mill, head movements in Z are harder to do precisely (lowering vs. raising, usually with no dial or DRO for the Z-location of the head itself), so we pretty much always use the quill to adjust cutter depth (re-locking for each pass).
This explains more than a few of my blown dimensions, and explains why I always had to creep up on things with constant mic'ing instead of depending on my dials. If I hadn't just installed a DRO, I'm not sure I'd have discovered the error. I've been blaming lack of rigidity and (when nobody is around) lack of skill!
This is a big enough deal that I plan to start another thread in the Asian mill section to warn people. I'll show how to use an indicator and gauge pins (or end mill shanks) to check for errors in the dial markings. In hindsight, I think adding a cheap LCD battery powered DRO to these Asian mills should be a priority.