Dealing with backlash when using a manual mill can be problematic for some, especially for beginners. Here is a tip on how to deal with backlash that really isn't very difficult but does require some planning, discipline, and consistency from the machinist.
Let's cover planning first. I almost always set up my drawings and jobs with my datums ( or zero, zero edges ) on the upper edge and the left hand edge of my work piece, as in the drawing below. All of my movements will then be in the X+ and Y- directions. There are parts that need to be handled differently, a round part with a bolt circle for example, but once you establish your zero point, the technique works the same.
The discipline and consistency comes into play when you pick up your edges or center point and make your moves. Using an edge finder, ( I run mine at about 600 RPM ) pick up the datum edges of your work piece being sure that you move into position by tuning your cranks clockwise. Then carefully zero out your dials. With a round part, use an indicator to find center and again be sure that you move into position by turning your cranks clockwise. Now the backlash will always be removed when the numbers are climbing, you don't have to worry about remembering which way you set it up.
Always, and this is important, be sure that when you make your moves, you move into position turning the cranks clockwise so that the numbers on the dial are climbing. Even if you have to move an axis backwards to get to a position, as you would drilling a bolt circle in a round part, over shoot that position and come back to position turning the crank clockwise. Always move to a position turning your cranks clockwise, Consistency!
Even though I have a readout on my mill, I always set my dials this way and approach positions in this manner. An added benefit is that if your power goes out, you don't have to pick up your edges again. You simply crank back to your zero zero position and reset your readout!
When I started my apprenticeship, everything in the shop was manual, so you learned to be consistent when dealing with backlash. When we got our first readouts, (some of the very first in our area) no one trusted them and always used this technique as a double check of the readout. We were surprised to find that the readouts were showing more movement than the dials did. This was especially prevalent on our older machines. We determined that this was due to worn lead screws. When the nut moves from the thin worn thread section of the screw to a less worn area the screw thread, the thread acted like a ramp and pushed the table more than the dial showed! We started trusting the readouts a little more after that.
I hope that some will find this a useful and not confusing way to handle backlash.
Richard