Stereo hysteresis revisited

I have replicated the historic Fender and Julesz stereo hysteresis study [J. opt. Soc. Am. 57, 819-830 (1967)] but with a different method of image stabilization. My results generally support their findings that once fused, the two stabilized monocular images of a random-dot stereogram can be separated on the retinas by as much as 2 deg and still remain fused. However, there is one major exception: whereas they found it necessary to realign the images to within 6 min of arc for refusion to occur, I find that refusion of stabilized random-dot stereograms may occur well outside the classical Panum's fusional area.