Two experiments measured line-dot alignment errors to either a long (2 deg 6 min) or short (12.7 min) oblique test line which was presented either in isolation or abutting a vertical inducing line. Alignment PSE's were obtained for a variety of line-dot separations at both the free end and at the intersect end of the test line, using a multiple randomly interleaved staircase technique. A number of aspects of the data were inconsistent with the hypothesis that alignment errors reflect whole-of-line changes in perceived orientation or intercept when the inducing line is present. Rather, the results were consistent with the notion that long lines are processed by integrating the outputs of analysers which respond to short line segments and that alignment errors reflect the differential weighting given to such analysers, depending upon the distance between the dot and the relevant segment of the line. An alternate hypothesis involving "cognitive mistracking" (Tong and Weintraub, 1974) could not be excluded by the results.