It looks like the originator has added a 1M\$\Omega\$ rheostat, which would have the effect of limiting the range of adjustment of the PWM pot. When turned down to minimum resistance you'd have full range of adjustment, at maximum resistance you'd have very little adjustment range. You can leave it out (ie. replace it with a short) if you want.
Call the PWM duty cycle electrical setting \$\alpha\$ where \$ 0 \le \alpha \le 1\$. Call the second pot electrical setting \$\beta\$ where \$ 0 \le \beta \le 1\$.
You can think of it (making some simplifications) as charging the capacitor through a resistance of
Rc = 4.7K\$\Omega\$ + 1M\$\Omega \cdot \alpha\$ + 1M \$\Omega \cdot \beta\$
and discharging the capacitor through a resistance of
Rd = 4.7K\$\Omega\$ + 1M\$\Omega \cdot (1-\alpha)\$ + 1M \$\Omega \cdot \beta\$.
The on and off times will be proportional to the charge and resistances respectively, whereas the total time will not vary much with changes in \$\alpha\$.
The brightness will be related to Ton/(Ton+Toff) if the frequency is high enough to eliminate visual flicker.
You should at least have a series resistor in the collector of the MPSA42. The way it is "designed" now it has an hFE-dependent current flowing at near Ub voltage across the transistor. That will tend to fry the MPSA42. At (say) 12V low-voltage supply, you'd get a base current of around 200-250uA, so collector current would be around 20mA (see typical DC current gain curves in the datasheet), leading to power dissipation of 4W with a 200V Ub. For a very brief period of time, followed by a 'pop', a nasty smell, and destruction of both transistors. So add something like 100K in series.