Are the obtuse angles the key components of the Poggendorff illusion?

An experiment is reported which confirms and extends a previous finding that amputations of the Poggendorff figure do not necessarily result in large positive effects with obtuse-angle stimuli, and small or even negative effects with acute-angle stimuli. Indeed, the acute-angle effects found were significantly greater than the obtuse-angle effects, and the full Poggendorff error was not explicable in terms of the linear summation of the component-angle effects. An 'alignment displacement effect' reported earlier by Hotopf and Obonai was shown to occur, but could not be an important component of the Poggendorff illusion.