> Assuming -60dB antenna efficiency (0.00001%, that is the lowest value 
> accepted) at both TX and RX side, average terrain and 1kW TX power 
> the RX voltage according to GRNDWAV3 is 50uV at 33km (1 wavelength), 
> 18uV at 100km, and 1.4uV at 1000km.
> But I am afraid that -60dB it quite optimistic and that -80dB would 
> be a more acceptable value, what would reduce the RX voltages to 
> resp. 0.5uV (33km), 0.18uV (100km) and 0.014uV (1000km)

This calculations is for far field. But in this case near field is more
significal if distance << lambda. For near field one can derive simple
formula (extremally approximative!) U_{RX}=U_{TX}(h/D)^3 where D --
distance, h - antenna heigh. Similar RX and TX antennas assumed and very
high RX input impedance assumed also. So if voltage on TX ant is, say, 1
kV and RX can recive 1 uV and h=10m then maximal distance is of order 10
km. 

It isn't very difficult to derive more acurate formula. But details of
antenna design is needed for this.

I did some experiments with such frequency many years ago. I was
supprised then woods, walls e.t.c. near antenna yelds very strong signal
attenuations (much more then for 136 kc, many of tens of dB). If one think
a litle then it is not very strange. Inducted charges in, say, trees yelds
field of opposite sign then antenna field. Trees are good coductor on such
low freq an they 'short' a field. Keep it in your mind!
  

Regards,
Alexander/RA9MB