It was previously reported (Chernomordik and Sowers, 1991) that erythrocyte ghosts which were exposed to a 42 degrees C, 10-min heat treatment would, upon electrofusion, produce over 15-20 s a fusion product with an "open lumen" (i.e., the fusion product became converted to one large sphere), while electrofusion of ghost membranes not so exposed would lead to chains of polyghosts. In phase optics the chains of polyghosts showed a "flat diaphragm" at virtually every ghost-ghost junction (i.e., the ghosts do not appear to be fused even though fluorescent-labeled lipid analogs can laterally diffuse from a labeled ghost to an adjacent unlabeled ghost). In the present study we found that the diameter increase in open lumen- and flat diaphragm-producing fusion processes both had a rapid but short early phase (0-5 s after fusion) which was exponential or nearly so and a slow but long late phase (5-120 s after fusion) which was essentially linear. Heat treatments at 39 or 42 degrees C caused a minor acceleration in only the late phase, while temperatures of 45 or 50 degrees C caused an immediate and dramatic acceleration in the rate of diameter increase (spheres in 1-2 s). Ghost membranes in the presence of glycerol at 20% (v/v) did not form open lumens when exposed to the 42 degrees C (but not the > or = 45 degrees C) heat treatment. This suggested that the heat treatment was denaturing a critical protein. Both of these observations are consistent with the involvement of the spectrin network since it is the only protein in the erythrocyte membrane which is known (Brandts et al., 1977) to have a calorimetric transition over the same temperature range used in our heat treatments. The diameter versus time curves were sensitive to: (i) the residual effects of the fusogenic electric pulse only up to about 1 s after the pulse, (ii) the strength of the dielectrophoretic field after the pulse, but not before the pulse,(iii) the ambient temperature during the measurement.