What voltage would you expect on an IT system with an undistributed neutral between live and earth?

There is rather more here on that here..
https://www.nationalgrid.co.uk/downloads/4102/5-arc-suppression-coils.pdf

 The decisions are essentially driven by the problems of a tricky soil type, but buried cables are more like giant TV co-ax and have a far higher capacitance per unit length, making the tuned circuit method less safe, as the impedances  get lower and the circulating resonance currents get correspondingly higher - segmenting the network into shorter lengths is only a partial solution.

The path finder fault detector basically sees the region of faulted cable as carrying an extra current, like a rather wide single wire with earth return, while beyond fault, balance is restored - the fields under the line are quite different, so fault finding from a safe distance is often possible.

Those interesting in some related  history may find the tests of the Alabama Power Company in 1923 ( page 31 of the PDF onwards ) an interesting read, as they tuned their reactor earthing coil up for the first time. The problems of capacitance in an IT system are similar, but much less severe ;-)
Mike

Hi Mike, Yes it was that '5-Arc-suppression-coils' file! It was back in January that I downloaded it - time flies.

I'll have a look at that AIEE journal paper. Always interesting how things were presented back then.

The comment about capacitance reminds me of an anecdote of a substation switch that was always burning out. The solution (probably temporary..) was simply to wire in as drum of cable (open ended!) - the problem was the resonance of the [original] low Capacitance connection  and local load transformer created an HF arc that never quenched. Adding the extra cable lowered the frequency allowing enough time at zero crossing to quench the arc. Funny stuff this electricity thing