We present three experiments that systematically examine pointing tasks in fish tank VR using the ISO 9241-9 standard. All experiments used a tracked stylus for a both direct touch and ray-based technique. Mouse-based techniques were also studied. Our goal was to investigate means of comparing 2D and 3D pointing techniques. The first experiment used a 2D task constrained to the display surface, allowing direct validation against other 2D studies. The second experiment used targets stereoscopically presented above and parallel to the display, i.e., the same task, but without tactile feedback afforded by the screen. The third experiment used targets varying in all three dimensions. Results of these studies suggest that the conventional 2D formulation of Fitts' law works well for planar pointing tasks even without tactile feedback, and with stereo display. Fully 3D motions using the ray and mouse based techniques are less well modeled.