but how would one determine its maximum frequency? Are we to look at the propagation delay of the 'not' gate and hence, calculate how many odd stages are required to produce the desired frequency?
in this case I would use a simple ring oscillator as it has already been suggested. Put down odd number of inverters and with the variation of their vdd voltage the output frequency will change as well. This vdd is obviously not vdd!, which is the common positive rail.
So: odd number of inverters + change of their power supply -> frequency changes.