I've updated the plots to now list the cross section value from the fit.

For run 10097097:

For run 10103044:

The next step is to repeat this process for a cleaned version of the barrel -- removing the trigger patches that seem to be high.  

At the moment, the code seems to be fairly sensitive to some parameters (max_time, as well as the initial value of sigma) that it really shouldn't be.  If max_time is set too low, we won't find the right offset, and if it's set too high, it seems to pick up an artificial offset that ends up putting everything half a cycle out of phase, resulting in a high background and a negative cross section, which is clearly wrong.  Thankfully, I can correct it by hand (and have working settings for both the vernier scans).