If you follow the schematic from V5a, the oscillator output goes through several rounds of attenuation via large resistors and divider networks, so that the signal at V4b is only a couple of volts. A couple of volts of bias swing on V4 is right where we need to be, as a small triode only runs in the 0-4V range. Remember, V5b is actually looking to the grid of V4b to get its signal. That's roughly where it taps in and starts its own voltage division process. Its ultimate goal, by design, is to make V4a see the opposite of what V4b sees. Same amplitude, same speed, opposite phase. And it looks like it's actually doing a decent job of it, it was just drowned out when trying to meter it vs scope it.