It can be both, if it is the same control it has to go from +2 to -2sd, if it is across 2 controls, one goes up while the other goes down. From his website: “R4s indicates a rejection when one control measurement in a run exceeds a +2s control limit and another exceeds a -2s control limit. This "range" rule is sensitive to changes in the width of the distribution, therefore it is a good indicate of increases in random error or changes in the precision of the method.”