The triode is acting as a grounded-grid stage. The whole thing is a conventional Schmitt long-tailed-pair phase splitter, the difference from convention being that the LHS is a pentode, not a triode: this was done to extend HF response, as stated by Bailey when he described it.The triode is acting like a folded cascode with no voltage developed at the grid and cathode, only currents. So it has a high output impedance. the pentode has the grid as input signal and needs the extra pentode grid to increase the output impedance to make it more symmetrical to the triode.
The equations published by Schmitt show that it can never be balanced with such a low value of Rk and equal Rp values without a balancing pot. Leak in the Stereo 20 used Rk=23k, Rp1=91K, Rp2=100K. Mullard in the 5/20 used Rk=82K, Rp1=Rp2=100K which was still out of balance by 3%, which was less than the best resistor tolerance available in 1954: they only got as close as they did because Rk was already so high. The original 'real' Radford circuit as designed by Bailey uses Rk=8k2, Rp1=33K, Rp2=39K, taking into account that the pentode and triode sides may have different values of mu.
The missing pot must definitely be restored (or the original values above used, but this will change the grid voltage which alters the requirements of the previous stage).