Many thanks to Andy Kunz, Peter Bains, and Keith Kotay for the
information about servos.  My one missing piece of information was that
servos are only supposed to turn 90 degrees.  The Futaba S3003's have
the physical stops a little over 180 degrees apart, so I thought that
180 degrees was how they were supposed to work.

I did some measurements -- sure enough, the servos are pretty linear in
the 1-2mS range, and lose linearity outside of that range[*].  However,
since this is just a personal project, not anything for production (and
there doesn't seem to have been any damage to the servos from my fooling
around so far), I think I'll continue to violate the specification
rather than complicate the mechanism of my camera platform.

Thanks,
Brian

[*] I hesitate to post these figures at all.  I attached a stick to the
servo output shaft, and plotted its position (by hand) for a few pulse
widths.  I couldn't find a protractor in my junk drawer, so I figured
out the angles trigonometrically.  The pulse widths are accurate, but
there are probably only two significant figures on the angle
measurements.

Pulse width (uS)        Theta (degrees)
2062                    189
1742                    157
1614                    148
1422                    130
1166                    104
1102                     94
 974                     81
 782                     63
 718                     55
 462                     32
 398                     27

The physical stops on the servo get hit at about 180uS and 2200uS.

Brian