Many thanks to all, havols, larryderouin most recently, for the comments on this old thread! Yes, seems true, this power supply design was not Fisher's finest moment and it will run like this, mine did for 30 of the last 50 years.
But I've resolved my problem with it to my satisfaction... Q951,a 0.8W max power dissipation part was dissipating 1.8W according to my measurements and calculations so I too limited collector and emitter currents by changing R954 to 135ohm/4W and R953 to 235ohm/4W resistors, they had been running hot too. Also, to address the 19.1V reading on the 15V supply going off the board at pin 9E, I replaced zener diode CR952, pin 9E is now at 15.6V. I can't be sure it was bad, the overdriven Q951 may have pushed the diode reverse current along it's V-I curve to 19.1V - the diode is a 5W unit, might be able to take it (it did out-of-circuit bench test fine, 15.2V at 75mA).
To cool Q951, the puny heat sink is replaced by two finned back to back Thermalloy units staked to the PCB, each with twice the thermal conductivity of the original. The board was browned and the transistor solder joints compromised to the point of intermittent operation.
Bottom line - only wanted to reassemble once, did so with these changes - Q951, R954 and R953 run noticeably cooler now and the sections that had been running on 15V supply at 19.1V should be happier. My signal strength meter that had been pegged out when any station of reasonably good power was tuned in is usable again, back to a 4.5 reading max.
New parts:
Q951 = NTE128 TO-39 NPN
CR952 = NTE5130A
heat sinks (2) = MarVac Electronics
www.ebay/str/marvacelectronics HS-14 snap-on HS T0-39 Item ID:25587573538
I don't know if my Flir camera thermal images will embed here below, was trying to show these temps, recorded after only the new Q951 heat sinks were added:
Q951 30.2 deg C
Q951 HS 72.7 deg C
R953 84.6 deg C
R954 76.9 deg C