QRPp How is it possible

When I started as a HAM, 20 years ago, I used my homebrew transceiver HM7, from Solid State Design, page 214...218, with an Inverted Vee, to work many different DXCC countries in Europe in CW. Because my power was only 500 mW, I had to look for stations, that I received with 599 or more, so the other station could hear my QRPp signal of 500 mW.
During daylight hours my RST would be 559, or even lower.
Please notice that a power of 500 mW is only four S-points weaker than 100 Watts.
See the PA1B QRPp page.

D and E layer
The radio wave is not attenuated when it is reflected by the E-layer. (Total reflection) But during daytime the signal is attenuated by the D-layer, when it passes through the D-layer.

During daylight hours, the signals are attenuated by the D-layer, which is absorbing

I discovered, that when darkness fell, the signals from the south of Europe would be very loud. So this made it easy, to make QSO's over a distance of 1000 km or more. When the D-layer disappears during the night, the signal can reach the the E-layer without attenuation in the D-layer.

The E-layer reflects the signal without attenuation

When the D-layer disappears, there is no attenuation in the radio path. The signal is reflected by the E-layer without attenuation. The attenuation of this sky wave is (only) 6 dB for a doubling of the distance. So the signal will be very strong.
At first, I did not realize, that my own QRPp signal, could also be an earsplitting 599, until I got genuine report from Italy, that my signal was S9, while using just 500 mW.

How low can you go

I was wondering, what the lowest possible power would be, if I would reduce my power in that situation with this excellent propagation.

If I reduced my power, so my RST would be 559, the difference is 4 S-point. This is a reduction in power of 4 x 6 dB = 24 dB which is 4 x 4 x 4 x 4 = 256.
So I could reduce the the power from 500 mW to about 2 milliwatts.

I was surprised by the low value, that I calculated, but this value is very realistic.
From 2001 on, I made many CW contest QSO's over more than 1000 Miles with 2.5 mW and 5 mW,
using PA1B homebrew attenuators.

I discovered that the frequency must be absolutely clear, when you use very low power.