C. Cahillane

Per Jeff's suggestion, I have detrended the calibration time-dependent kappas to get the true all of O1 uncertainty associated with the kappas.

This was done to eliminate the spikes in calibration uncertainty we would see from changes in the detector, such as ESD bias flips.

I detrended by finding the median kappa for a range of 100 kappas around a specified gpstime.  Then I would subtract the original kappa from the found median.  I then took the standard deviation over all of O1 from the detrended kappas.  These are the numbers reported in the histogram titles. 
From this method, we get sub-percent and sub-degree uncertainty, due to the massive numbers of kappas we have.  I think this is fine as long as we understand why the detector configuration changed at every kappa-jump.

To see the LLO kappa detrends, please see LLO aLOG 25677