It worked! Replacing that resistor moved the circuit characteristics enough that I was able to tune and scale all of the 101's three exponential converter circuits. I installed a 2.55K 1% resistor in place of the 2.2K resistor in the reference voltage circuit, and I measured 4.46V vs. the specified 4.4V. That wasn't a huge change from the 4.6V that it was outputting with the other resistor, but it was enough to give me the leeway I needed on the tuning and linearity trim pots. It now plays in tune at both ends of the keyboard. (Although, curiously, it's ever so slightly flat in the middle octaves. Maybe it's always been like that and I never noticed.) And I can now play all four VCOs in unison without much beating.
The process is a PITA if you don't have a high-accuracy frequency counter, and I don't. I use the A440 reference signal built into my Synthesizers.com Q123 standards module, and route that into the external input of the 101 so I can mix it with the 101's own signal. I set up a patch with VCO 2 only. I then play high A and tune it to eliminate audible beating. Then, I unplug the A440 from the 101's external input, and connect a Q106 VCO which is driven by control voltage from the Q123. I turn the Q123's octave knob up to the fourth octave and tune the Q107 to match the EML's high A. Then, I turn the Q123 down three octaves and compare it to the Q107's low A. Adjust the tuning trimpot until it matches. Set the Q123 back to the high octave; check the high A and adjust the scaling trimpot. Then, check the low A again, etc. It took me about 10 tries. Once VCO 2 and the keyboard 1 expo converter are tuned and scaled, do the other VCOs, and then the keyboard 2 expo converter. Finally, scale the filter (involves moving a few wires around).