What a neat problem. Not something I've seen before. it took me a while to sort out what you meant by backfeed. To better understand what is going on I simplified the schematic:
Neptune1 (resized).jpg
The schematic above shows in red how the Q/Series and one of the Rollover relays should fire together when a rollover switch closes. There shouldn't be a path to the W/Wow relay because the stepper has selected a different rollover and is completing its path. What's really happening is more interesting:
Neptune2 (resized).jpg
This schematic shows that not only is the path to the Q/Series relay complete, but a longer path to the W/Wow relay is also complete even though the stepper isn't pointing to the active rollover switch. In the worst case any rollover will award the Wow as pinhead52 observed (once either the red or black series has been completed which closes a switch to the W/Wow relay not shown here).
Simplifying the schematic even more you get:
Neptune3 (resized).jpg
The top path through the Red 10 Rollover relay coil and the Q/Series relay coil uses coils that are intended to work together in series so that they can both operate properly given that they have to share the 24 volts between the power rails. They have very low resistance of roughly 2 ohms and can operate on about 12 volts. This works just fine until the red or black series completes and the 2nd path through the W/Wow relay is added to the circuit.
Once a series is completed what you have is the Q/Series relay coil in parallel with two other relay coils which changes the balance or the voltage that each coil sees. Even when it's working properly there is current passing through the lower path, just not enough current to fire the W/Wow relay or the Red 9 Rollover relay.
When pinhead52 shorts the lugs of the Q/Series relay the Wow problem stops (but I bet the rollovers don't score either since the Q/Series relay doesn't fire). The reason the W/Wow relay doesn't fire in that case is because the jumper essentially connects the left power rail to the junction where the two paths split. So there is no voltage difference between the left and right side of the W/Wow relay coil, no current flows and the relay doesn't fire.
The way to fix this problem and make it bullet proof would be to add (lots of) extra switches to prevent the back door path to the W/Wow relay. Gottlieb instead found a combination of relay coils that would work in the two scenarios with different voltages. After 50 years it's hard to say what changed. Maybe some of the coils were replace that slightly changed the balance?
Assuming that the W/Wow relay is using the right coil, I would experiment with adding a small resistance in series with the W/Wow relay coil. Just enough to limit the current through it so that it won't fire when the backdoor path with the two extra coils is complete, but not enough to prevent it from firing when connected directly to a rollover switch.