100MHz oscillator. Set it up as a voltage divider: osc output -- FB -- 50 ohm terminated cable (e.g. scope/speccy/RF voltmeter input).
Measure several values of resistor (0, 50, 100..) first, and undo the voltage divider equation to solve for source resistance.
Now you're ready. Insert FB, measure the voltage, and (based on the calibration above) calculate its impedance. You'll actually be measuring the magnitude, which is close enough. Most FBs will be somewhere between slightly inductive and resistive at this frequency, so it shouldn't be too far off.
FBs are sold by their impedance at some frequency, most often 100MHz, so this is as good a method as any.
I think you'll find the physically larger ones usually have more impedance, but it will also vary significantly with type, as there are three major classes of material used: LF, MF/wideband, and HF. If you sort them by both size and impedance, you should see a pattern like this, more or less.
Such a setup will also work well for RF inductors in the range of L ~= (50 ohms) / (2*pi*(100MHz)) = 80nH, give or take a factor of 10 let's say. Or 10x higher inductances at 10x lower frequency, etc.
Tim