[time-nuts] Comparing 3 oscillators using a 2 channel frequency counter?

[Erik Kaashoek]

As a good engineer I want to check the short and long term stability of 
all my oscillators but to do that a need a better oscillator...
But I've read that it should be possible to measure 3 independent 
oscillators together and use some kind of statistical tools or voting to 
get some better insight in the performance of each of the 3 oscillators.
Now the problem is: where to find a 3 input frequency counter? I don't 
have one, but I do have a 2 input frequency counter that can use an 
external reference. So I connected 2 oscillators to the two inputs and 
the 3rd to the counter reference and for each measurement of the two 
frequencies I also calculated also the ratio of the two measured 
frequencies scaled back to the oscillator frequencies (all 10MHz) and 
imported in Timelab. [1]
Looking at frequency difference chart two oscillators (DOCXO-RB) seem to 
be long term more stable with respect to each other compared to the two 
other combinations (OCXO-Rb and DOCXO-OCXO), these two act as if they 
are each others opposite. Of course this is just looking at the plot so 
my questions are:
- Is this the correct way to use a 2 channel frequency counter to get 
info on 3 oscillators?
- What (mathematical) tools can be used to get insight in the 
performance of the 3 oscillators individually?

I've read about 3-cornered hat and using the three frequency 
measurements I know how to calculate the taus of each pair and import 
into stable 32 to do the 3-cornered hat calculation but what would that 
tell?

[1] http://athome.kaashoek.com/time-nuts/3-freq.png
Erik