2002ApJ...569..362S

We discuss the results of long-term hard X-ray monitoring of Galactic black hole candidates 1E 1740.7-2942, GRS 1758-258, Cyg X-1, GX 339-4, and Cyg X-3 with the Rossi X-Ray Timing Explorer. The objects divide into two classes. In the first class, exemplified by Cyg X-1, luminosity and spectral hardness evolve simultaneously. In the second class, the relation is more complicated: the softest spectra occur while the count rate is dropping. Most models of accretion, tailored to Cyg X-1, do not predict the second sort of behavior. One interpretation is a simple model with two simultaneous, independent accretion flows: a thin disk and a hot halo. A drop in the accretion rate affecting both flows would propagate through the halo immediately but might take up to several weeks to propagate through the disk. While the inner halo is thus temporarily depleted compared to the disk, a temporary soft state is expected. This picture is supported by the observation that those sources that show delays (1E 1740.7-2942, GRS 1758-258, and GX 339-4) are expected to have low-mass companions, and those that do not (Cyg X-1, Cyg X-3) are known or thought to have high-mass companions. Low-mass companions imply accretion by Roche lobe overflow, with a high specific angular momentum in the accreting material, and therefore a large disk with a long viscous timescale. Wind accretion from massive companions is expected to result in a much smaller disk and thus little viscous delay.