For some years I have used an earth-electrode antenna to TX my earth-mode signal on VLF. The earth-electrode antenna consists of two earth rods, or one earth rod and a connection to a copper water pipe separated by 15-20m of cable running along the ground. Using this system on 8.97kHz I can detect my 5W VLF beacon to at least 6km.
Recently I have been experimenting with the very same antenna at 472kHz and getting some quite remarkable results with reports from over 1000km on WSPR using low mW EIRP levels. Another local MF station G4HJW has also carried out experiments with a very similar antenna, in Bernie's case he is using 30m spacing between the earth rods, and getting results from over 900km away when running about 6dB more than me.
It is early days yet but here at least, where the bedrock is chalk, I am getting results only 2-14dB below those I get with a 6m top loaded Marconi antenna on 472kHz. There is some evidence from the signal strengths in various directions that the earth-electrode antenna behaves like a virtual loop within the ground, radiating a signal much like that from an elevated wire loop i.e. with directivity. Signals in the line of the antenna are strongest and those orthogonal to the wire weakest.
Testing the same 15-20m baseline earth-electrode antenna on 137kHz give similarly encouraging results.
At the new QTH I have a shorter baseline earth-electrode antenna with one end (shack end) connected to a mains earth. The spacing from the far electrode is about 12m with a short wire about 1.5m off the ground running along the garden fence to the far electrode. Best DX on 472kHz with a measured ERP of under 10mW is just over 1300km.
This diagram, video and photo may make this clearer. So far, at this QTH it has given decent results on 630m,160m, 80m and 60m.