Appendix B: The hole-cutting data reduction process

The basic tool which we use to measure the success, or otherwise, of our developing auto-folder is the distribution of $\ln A$ over the whole PS sample. However, an accurate estimate of $\ln A$ for each rotation curve requires the application of the "hole-cutting'' technique, which aims to remove that interior part of the rotation curve which is dominated primarily by the bulge. The details of this technique are given in Roscoe (1999A), and its effect on the data of the PS solution are summarized in Table B1 which lists the average and standard deviation of the 900 mean square residuals which arise from linearly regressing $\ln V$ on $\ln R$ for each of the folded rotation curves of the PS data base, before and after the application of hole-cutting.

   

Table B1: Effects of hole-cutting on power-law fits

Hole-cutting	Mean rms	Std Dev
Before	8.4 10-2	0.16
After	2.7 10-2	0.10
N = 900

Table B1 makes it clear that the deviation of rotation curve data from the simple power law model is very much concentrated in the innermost regions of the rotation curves, and thereby provides the empirical justification for the hole-cutting technique.