Based on feedback from the last round of timing scans with the endcap shown here, I made a small change to the slope determination.  I now fit from 5 sigma above pedestal to 25 counts above that, where there are greater statistics.  In the fits shown below that means I use the green line instead of the red line.

The slopes averaged over each crate (eemc_slope_v2.pdf) and for each channel in each crate individually (eemc_channels_crate*_v2.pdf) are attached.  I'm also attaching again the result of a gaussian fit to the timing curve for each channel (eemc_tower_fits_v2.txt).  If you want to repeat this yourself there is also a text file with each and every slope at each timing point (eemc_all_slopes_v2.txt)