I could not do any sensible measurements at 40MHz because my hybrid couplers are not good enough at that frequency.
So I used a different approach and made a measurement at 50MHz.
The picture shows the 50MHz (+8dBm) carrier passing through a 50.125MHz crystal filter, the filter attenuates the 50MHz signal by almost 50dB.
What is left is the true RMS level of noise passing through the filter, The cursor shows the analyser noise floor at -122.5 dBm.
The Leo Bodnar device noise measured at 50.100MHz is -107dBm.
As to how that will relate to 40MHz I will leave to others. This is my final opinion on the subject.
Regards everybody and stay safe.