Figure 4: The 
slice through the 
surface produced by
folding the 2-4keV light curve through the trial transfer function
and comparing with the 8-15keV light curve. Note the small (1
bin) delay between the two light curves (with the harder band lagging the
softer).
By considering the slice through the surface
produced with trial transfer function , we can examine
overall lags and leads between energy bands. Examining the 2-4keV and
5-7keV light curves for MCG-6-30-15 in this way, we find that the
slice possesses a minimum at zero lag -- i.e. we find no
evidence for overall time lags or leads between the continuum and line
bands down to 64s, the bin size of the data. Performing the same
procedure for the 2-4keV and 8-15keV light curves reveals a one bin
offset in the position of the minimum (Fig. 4), indicating that the
8-15keV light curve is delayed by 
50-100s as compared with the
2-4keV light curve.
Figure 5: Discrete Correlation Function (DCF; Edelson & Krolik 1988)
between our 2-4keV and 8-15keV light curves. Note the asymmetry
in the DCF which validates our detection of a 
time lag
between these two bands using the PRH92 method.
Lee et al. (1999b) have applied CCF methods to this RXTE
dataset. By carefully comparing with simulations, they find evidence
that the 7.5-10keV band lags the lower energy bands with a phase
delay of 
. They also find evidence that the hard band
(10-20keV) lags the softer bands with a time delay similar to
that found in this work. Figure 5 shows the DCF for our 2-4keV and
8-15keV lightcurves (this is very similar to Fig. 17 of Lee et
al. 1999b). A small time lag of 50-100s between these two bands is
evident. Thus, CCF methods and the optimal reconstruction method both
suggest a time lag of 50-100s between the 2-4keV and 8-15keV
bands.