Thanks for the advice.
I think this type of AGC may be suitable for my own application because the sinewave output isn't actually going to be used for audio purposes so some small distortion of the wave wont make too much difference, as long as the general shape is there.
I've worked out that when the diodes conduct the effective feedback resistance becomes 8.3 ohms, which because of the 6K7 input resistor, gives a gain of 1.22 rather than the original 3.9
As I understand it, a Wien oscillator requires a gain of 3 to sustain oscillation, so when the AGC kicks in this could temporarily effect the circuit's operation - is this the distortion you refer to?
If we're talking about using this kind of arrangement, then the output voltage will surely be restricted to around 1.4-2V? After that the diodes would be conducting all of the time...