I would use a 0.5" ball, and try to use the sigle point circle. On a cylindrical part Ø5, the diference of height between the highest point and the lowest point of a hole Ø0.25 is 0,04 mm (0,0015"). I believe that the uncertainty has approximately the same value on a Romer, I think you can do it by this way.
Grin...a half-inch ball? Really? Gonna have some trouble fitting that into the 1/4" hole.... You must mean a .5mm ball...
EDIT: Hmmm, maybe you mean just pressing the half-inch ball against the surface where the 1/4" hole resides...OK, I can see that, gets you location in 2 axes.
