Discussion

In order to bring structure to the discussion that follows, we will summarize the pertinent results from this paper.

1. We clearly see reduced fractional variability in the iron line band (5-7keV) as compared with the continuum band (2-4keV). This is the origin of the additive offset, K, that was introduced in Section 3. The spectral fitting results of Lee et al. (1999b) suggests that this is due to a combination of a constant iron line flux and flux correlated changes in the photon index.
2. Our analysis finds no evidence for iron line reverberation effects. By running a number of simulations, we find that any reverberation time delays must be less than or greater than ks. Together with the above result, this suggests an approximately constant iron line flux over these timescales. Thus, we can extend the work of Lee et al. (1999b) and infer a constant iron line on timescales down to 0.5ksec.
3. Any overall time lag between the 2-4keV and 5-7keV band is less than s. However, we do find that the 8-15keV band is delayed with respect to the 2-4keV band by 50-100s. We can use this time delay to obtain a rough size scale for the Comptonizing cloud that is producing the hard X-rays. Assuming a coronal temperature of , it takes approximately 3 inverse Compton scatterings for a photon to be boosted between the 2-4keV and 8-15keV bands. Thus, the mean free path of a photon is approximately 15-30 light seconds. This is a lower limit on the size of the Comptonizing region.

As we will see, this combination of facts presents problems for current models.

• Simple reflection models
• More complex scenarios