Several important points to note.  The maps need to be aligned with each other.  I do that in the example by opening fit atomic models, moving the second atomic model to align with the first, and then moving the second map to the same position as the moved atomic model.  For the subtraction to give a useful scale factor you want to limit it to a region where the two maps mostly overlap.  I cropped the first map to such a region in the example.  Then I subtracted the first map from the second -- the RMS is minimized within the contour level of the map that is being subtracted (#1 in this example), so you also might want to choose the contour level.  The result in the example is that map #1 needs to be multiplied by 0.13696 to minimize the RMS difference.  I think this is prone to giving a bad scaling factor if there is significant non-overlapping density for the region within the map contour level.  You can judge how well the scaling worked by looking at how much density is left in the difference map.

Tom