So I plugged my measured figures into the equations to work out ERP on the 472kHz band with the very low height Marconi. Measured antenna current is 0.64A and effective height about 1.6m. The formula gives EIRP taking into account the lack of directivity.
EIPR (472kHz) = I^^2 * Heffective^^2 /100 (source LF Today 3rd edition)
= 0.64*0.64*1.6*1.6 / 100 = 10.48mW
ERP is 1.8/2.52 times the above figure, so around 7.5mW.
My estimate of 10mW ERP is a few dB too high, but not a bad estimate, considering experimental error.
As the ERP is proportional to current and effective height both squared, it is easy to see how ERP can be much improved by increased antenna vertical height, increasing the power, or doing both. The other thing is to reduce system losses. An antenna/earth that looks like 300 ohms is probably lossy.