(1) The propagation is definitely "utilities assisted" conduction apart from the last few metres of induction into the loop. Even so, I'm amazed that the signal can be seen so clearly on Spectran when running QRSS3 with such low power (4W) to such a small earth electrode antenna with just 20m spacing. 5.6km is a very long way from my QTH.
(2) Even at 2.4kms from the home QTH, the signal induced in the ground via buried pipes was stronger than any detected by pure induction. I tried various loop orientations to see if there was any "pure induction" signal, but I was unable to see any.
(3) There is weak evidence that at the more distant points some assistance may be being given from water channels criss-crossing the wet fenland. This needs further testing, but I suspect there is something in this.
(4) There is no evidence that overhead electricity cables are helping: when orienting the loop for best coupling into these the signal actually decreased.

(1) Try using a larger loop at the RX end. I have in mind a 3 or 4 turn loop some 5 x 5 metres that I can lay flat on the ground. This will allow stronger coupling into the loop from any buried utilities. Also, with a small loop (80cm) it is a bit "hit or miss" where this is placed to get decent coupling when signals are weak. Also, a simple wire loop can be left in the road when cars go by, whereas my 80cm loop on a wooden frame always has to be moved...or be crushed.
(2) Try using an electrode pair antenna at the RX end. I'm thinking of straddling the roadways or drainage channels with this to see if the received signal is stronger and can be detected further from the utilities.
(3) Increasing the TX power. 100W should allow more chance of communication over moderate distances with less dependence on buried utilities or by pure induction (or even some radiation). I need to get hold of a 100W amplifier.
(4) Try an E-field probe antenna at the RX end using an earth electrode RX antenna.