Concurring with views regarding the OPT tap, but I do find the input circuit rather strange.

As Wavebourn said, it resembles the usual loudness control - except for the fact that as shown it operates rather the opposite way!

One can view the amplifier as an inverted topology, with the summing point at the junction of the 3,9K and 220E resistors. This is a low impedance point, 3,9K being the feedback resistor and the input potentiometer the input series resistance. With the input grid near the bottom of the pot the topology holds. As the slider is advanced upwards the input series resistance not only decreases but the NFB also decreases, thus increasing signal volume. (One is presuming a low input feeding impedance.)

But then at low volume - say with the pot slider at the 100K tap position, where one would expect greatest volume increase for l.f. as per loudness, the input *attenuation* is in fact increased as frequency decreases. (The impedance of the 3,3nF and 10nF series caps increases with lower frequency, increasing the 'series' input resistance and thus decreasing the volume instead of the other way round.) I presume that the schematic is drawn correctly. With the volume slider at max. position, NFB is minimum, then increasing l.f. where it is not normally desired by increased NFB at mid. frequencies. The input impedance is thus also not constant with frequency, etc.

??? There are several ways of implementing such a circuit; in this case I would imagine the more normal way of feeding NFB to the input tube cathode and leaving the grid 'free' to normal circuitry would have been better. (I would have liked to see the design response graphs for the various settings - easily simulated although I have not done that at this time.)