Steve Olney wrote:

OK thanks.  I have a clearer picture now.   Doing a napkin analysis and
taking into account the logarithmic nature of the S-meter readings, it

seems

that, for example, if you have, say, S5 noise alone and then an S5 signal,
then the S-meter reads somewhere btween S5 and S6 - closer to S5.
Therefore an S5 report would be warranted as being pretty close.    The
problem seems to be when you have, say, S5 noise and, say, an S2 signal.
The S-meter still reads S5.   Giving an S5 report would seem to be suspect
under those circumstances.

What Steve uses as an example describes the situation at my station. The
noise level during daytime is usually S5. If I can just read a station by
ear I give it S5. Readibility could be between, say, R2 and R5, depending on
whether the signal is at the low or the high end of the signal span around
S5 (more than S4/5 and less than S5/6).
I  never give reports less than S5 because such signals would be too weak to
be copied by ear in the noise.

I calibrated the receiver according to the IARU recommendation for
frequencies below
30 MHz: S9 means a signal power of -73 dBm fed to the receiver. When the
receiver input impedance is 50 Ohms a power of -73 dBm is reached at 50
microvolts at the antenna terminal.

A step of 1 S-point is 6 dB.

73, Dick, PA0SE