[time-nuts] HP53132a counter noise floor

[Club-Internet Clemgill]

Hi, 
I am testing my 53132a counter floor.
Connecting the 10Mhz rear output to Channel A, and reading ADEV with Timelab in frequency mode.
Attached picture shows first results.

We see log-log lines with a -0,5 slope, and a vertical spacing ratio consistent with (Tau0)^2
(Tau0 = actual gate time given by Timelab) 

I understand the 1/square(Tau) slope of the lines, as RMS resolution is enhanced with the length of samples
(And ADEV # RMS of fractional frequency in this configuration)
But I don’t understand the (Tau0)^2 spacing ? 

According to the counter manual, RMS resolution should be proportional to 1/(Tau0  x square(Tau)) so not to 1/(Tau0^2)
I looked at literature, including Rubiola and many others but could not find the answer.

Any clue ? 
Best, 
Gilles