Thinking about it some more, I have a vague recollection of Black Sheep mentioning that the "line disconnect plugs" in the pole-top "coffins" were prone to come loose as the lid was being closed. So much so that the manufacturer eventually modified the design . . .
Almost on the money, B*Cat ...... you have an incredible memory btw.
If it's a BT18C pole-top block (affectionately called a coffin block), the issue we found and still do, is that the connection modules used to bond the incoming underground (UG) feed wires to the outgoing dropwires, are a push connection on the UG side of the module.
A truly awful design IMO, and if the UG feed is anything less than 0.5mm then there is a large potential for a HR fault. Alas, the fault is so miniscule it can quite easily be missed by the PQT, DSL and Eclipse tests ............ it's simply a case of knowing this can happen and climbing the pole to check.
If it does prove to be the fault, a very simple fix is to 'crimp' s piece of 0.5mm Cab jumper wire onto the existing UG wire, and the connection should then be good. I am old-school, and will always prefer a screw terminal connection method (despite it's obvious future weak point - corrosion), over a PVC push-connection.