The second building with the same supply is always a point of contention, and you need to consider if bonds to incoming extraneous parts are actually MAIN bonding conductors (to a distribution circuit), or if these are supplementary bonding. That is the point though Andy, they share extraneous parts and these must either be fairly high resistance apart or directly connected, in which case one does not need the bonding conductor. To get seriously large bonding currents the necessary circumstances are both very unlikely, or more likely impossible. Plastic service pipes have reduced the possibilities still further, and real Earth resistance hasn't changed. Try drawing out the possible fault circuits, and compare these with any buildings you know, are they similar, I rather doubt it?

Even with PME the bonding conductors are quite small really, and the risk is negligible, dangerous events are extremely unusual, again probably due to service pipes being plastic. In a TT installation the RCD protection protects against foreseeable circumstances, or are you simply thinking that the local CPC potential may be raised?

One scenario I had in mind was something like this:

so that the extraneous gas and water pipes in effect emerged in different buildings - and so the "bond" between them would be just the c.p.c. between the house and garage.

Say our house was all TT, but a little way down the road (further away from the substation) an all-electric house has been converted to PME and the other houses nearer the substation of course had bonds between gas and water,. Where would the diverted N currents flow?

   - Andy.

OK , you have made several assumptions there Andy, and whilst there is some possibility that they may be true, it is unlikely. The first is that the CPC is going to carry a large current, which it cannot because of its resistance and the fact that the diverted voltage driving the current will be small, unless there is a serious fault elsewhere. To cause a failure of the 1.5mm CPC a continuous current of somewhere around 50A would be required and a reasonable estimate for the cable resistance may be 2 Ohms, so the driving voltage would have to be 100V! This is obviously ridiculous and a voltage of perhaps 1V is possible, so a current of 500mA.

If worried about this situation an insulating link in the water pipe is the obvious answer, and you must admit that it would be an unusual scenario.

 that it would be an unusual scenario.

Of course, but it stands as just one example of range of possibilities - we've had decades of repairs and alterations to underground metallic pipework done using plastic so there must be numerous other examples of relatively similar situations out there.

I guess my hope would be that the regulations describe what in principle what would produce a satisfactory installation, rather than a specific policy that only mostly works in practice or merely satisfies the 80/20 rule or somesuch.

We could perhaps try some other numbers too - the outbuilding might not be 165m away (enough to get 2 Ohms out of a 1.5mm conductor at 12.1mΩ/m) - around here we're probably more modest and often have the garage just a footpath width away from the house - so say a 5m submain if the main CU happened to be close - so 0.06Ω might be more realistic. Also 50A feels a bit high to me - if we're looking for something that might prove satisfactory in the long term for a conductor in a cable shared with other possibly loaded conductors. The rating for a 1.5mm² in a cable with three loaded cores, method C, is 17.5A - so I could argue that twice that - say 35A might well be unsatisfactory in the long term. So V=IR then gives us just 2.1V - which doesn't feel quite so implausible.

    - Andy.

It is an interesting question - clearly if the water to the outbuilding comes from the house - perhaps verified if it can be turned off from a stop tap already on a pipe main bonded then there is no issue - but as drawn, where the water pipe and the mains are supplied from opposite ends of the installation then something akin to the problem of multiple PME supplies in the same block arises.
For a long time the fact that the regs specify main bonding size conductors even where there is no credible high current path has rankled, as a waste of effort and copper, but equally a broad brush abandonment is just as unwise. What is really needed is far more nuanced guidance where the installed can only reduce conductor size after verifying the route or impedance of the non cabled metallic paths.

(and while you could perhaps protect the inadequate CPC with a fuse or impedance to limit the current, but that too would be very much against the spirit of the regs  though I have often though t for FE it could be permitted so long as the safety earthing was separated.)
Mike.