In three experiments, it is demonstrated that the negative Poggendorff illusions that Restle (1969) and others have reported are a special case, that, in general, acute- and obtuse-angle effects are the same, that there are additional special cases which modulate the illusion, but only slightly, and that these special cases derive from particular configurations which impose other effects, including observer strategies, on the basic illusion. These additional effects do not bear necessarily upon the fundamental explanation of the illusion. The experiments also show that effects obtained with amputated illusory figures are not linearly additive. Recent experiments that have measured apparent oblique separation in the Poggendorff figure and have found large obtuse-angle but small, or negative, acute-angle effects are also discussed. It is argued that these experiments have not demonstrated that the two dependent variables-alignment and length judgments-are correlated, or that one error causes the other; and it is argued that uinaHolland's theory of parallel attraction to explain the illusion is not tenable.