Gene has included the dE/dx calibrations and a few other changes into the acceptance.  Compared to my October 11th results, this creates a noticeable change.

Though the fits look nice, I did note that the raw distribution has some issues introduced.  Note the following are not from the same run.

(Left is the distribution I saw on Oct. 11th, Right is the new distribution after changes etc.)
                                  

Despite the oddity in the distribution, but the 1/r^N and the generalized howfit have matched the radial distribution nicely, as well as the azimuthal distribution - the radial fit.

The Order of N for all 17 runs now seems more luminosity dependent once again, but it is closer to the original expectation of N = 1.71:
(Left is October 11th results, Right is current result)

I have tried the more generalized model: [0]/(Radius*Radius) + [1]/Radius + [2]

The Howfit comparison is based on our current model:  [0]*((3191./(Radius*Radius)) + (122.5/Radius) - 0.395 ) / 15823  (Oct. 11th I showed with + [1] which was incorrect)

The [0]/luminosity coefficient is ~0.48 for the original howfit, which is much higher than the 0.15 from the generalized (though in the error bar range).
The [2]/luminosity coefficient is ~-6e-5, which is higher than the constant term from the generalized howfit.