Moving about the web the spider Agelena labyrinthica continuously adjusts the prospective return angle. The amount of path integration was indicated by two compromise angles, return angle e and goal angle t# (Fig. 2). The spider was primed to one of two perpendicular light azimuths, L1 or L2. Subsequently, the discrete effects of a 90 ~ change in light azimuth on the return direction were recorded (Fig. 3). When primed to L1, and the spider was exposed to L2: (1) while homebound, the deviation from straight home was clockwise and largest, (2) both while outbound and returning, the deviation was smallest (Fig. 4), (3) while outbound, either during the first or second half of the outbound run, the deviations were between those obtained in (1) and (2). When the spider was primed to L1, but given L2 while outbound and L1 again while homebound, Agelena deviated counterclockwise, the most with L2 on during the entire outbound run, and less when L2 was available only during a portion of the outbound run. The degree of adjustment of the home path direction is not correlated with the actual path length but with the shortest distance ('bee line') between the two points during which the spider was exposed to one of the two light azimuths.