Don,
You are right. The closer the phase and return conductor are (to being parallel with no separation), the less reactance there is, to the level of essentially zero. The current (at an instant in time) that is coming from the source on the phase conductor equals the current returning on the return conductor. Each current has a magnetic field of the same strength, but a field of opposite direction, so the magnetic fields cancel each other. No magnetic field, no inductive reactance. The further the two conductors are apart, with equal and opposite currents, the less canceling of magnetic field occurs, hence, more reactance. The Green book is starting off from a point (2" conductor separation) where there is a small amount of reactance, because, although close together, the magnetic fields don't quite cancel. Moving further apart (30"), the canceling effect of the other conductor's magnetic field falls of with the inverse of the square of the distance (1/d*d). The canceling effect decreases very rapidly, at first, then becomes very gradual, but never quite going away. Between 2" and 30", reactance in each conductor rises to about 90% of its theoretical maximum. So, the increase over 2.65 times that of the 2" amount won't be much.
Here's the other thing that balances the water pipe impedance: The neutral from house #3 to the transformer has a reactive impedance created in it by the unbalance current equal to the current in the water pipe, but the current is in the phase conductor. It's like this:
Hypothetically, house #3 has a 100 amp load on one phase and no load on the other phase conductor. As a result, the return current at the service neutral bus is 100 amps. Let's say, then, that 50 amps goes on the water pipe, and 50 amps goes on the service neutral conductor (for the purpose of this discussion). I think we're agreeing on what happens in the 50 amp water pipe current.
The 50 amp service neutral current is the interesting one. It creates a magnetic field that cancels half of the field created by the 100 amp current that is in the phase conductor right beside it (between the transformer and house #3). The remaining magnetic field from the phase conductor has a direction that generates current in the neutral in the direction of the current of the neutral -- a leading inductive reactance that is almost the same magnitude as the lagging inductive reactance in the water pipe.
Both of these reactances add to the resistance in the pipe or wire (a vector addition) and, IMO, the magnitude of the total impedance in each path almost the same, just with a difference in direction. The direction difference cancels at the transformer neutral terminal.
It's the extra 50 amps (or whatever the amount might actually be) in one of the phase conductors to house #3 that creates a matching reactance in the neutral conductor.
Al