I found this very nice air-vairable tuning capacitor unexpectedly in the junk box. It has a 100:1 worm-drive reduction mechanism which after the application of a bit of oil, was easy and smooth to turn.

From http://www.qsl.net/ve3bdb/bc221photos.htm it's possible to see conclusively that the capacitor I have comes from the legendary BC221 frequency counter, in this case the BC221-AK. This is different to the BC221-T that I documented elsewhere on this site. The photos below are show the capacitor in the BC221-AK, and how the front panel of a BC221 looks.

The tuning capacitor has a 100:1 worm-drive reduction mechanism. I carefully measured the capacitance curve at 1 turn intervals for the 100 turns, and plot them below. Note the intended use is from turns 50-100, turns 0 - 49 are just the "back" of the vanes, as the capacitor is continuously rotating.

The BC221-AK tuning capacitor was designed with very carefully shaped plates, such that with a 50pF parallel fixed capacitor, the frequency-turns chart will be precisely linear and provide a 2:1 frequency range. This relationship holds whatever value of inductor is used in the LC tuned circuit. Different switched inductors can produce switched frequency ranges each providing 2:1 ratio. This simulated chart shows how the frequency vs turns chart would look.

In the case of my 6.0-6.5MHz oscillator, the addition of a 1200pF parallel capacitor is required in order to obtain the required 500kHz coverage for the full 50 turns. However the frequency vs turns chart is not linear. The simulated chart below shows how the frequency vs turns chart would look.

With proper choice of capacitor in series with the tuning capacitor, it IS possible to obtain linear frequency vs turns, or invert the curvature too if desired. One could argue that it might be nice to have lower kHz/turn at the CW end of the band, compared to the SSB end.