Combining digital and physical worlds is today largely simplified thanks to many advances in the capabilities of optical see through AR head mounted displays. In many scenarios such as augmented maintenance, the implementation of AR is now based on the use of physical referents associated with digital information. Although the display space with optical see-through Head-Mounted Displays (HMDs) is almost unlimited, its field of view (FoV) is still small: overlapping the FoV with all these physical referents is therefore very constraining. This raises the challenge of pointing physical referents around the HMD FoV, where the feedback cannot be displayed. To this end, our work characterizes the pointing of physical referents inside and outside the HMD FoV to establish the differences. Our studies focus specifically the nature of the target and the form of feedback used during pointing.