[evlatests] Important Correction!

[Rick Perley]

    I earlier reported that  the 'cross-auto' correlations, when the 
same signal is fed into both R and L samplers, was low by 62% compared 
to the 'auto-auto' correlations. 

    This was erroneous.  I had utilized the bandpass plotting program 
POSSM for these values.  There is evidently another problem with this 
program when plotting cross-hand spectra ...

    When the general plotting program VPLOT is utilized, a quite 
different, and I think more believable result is obtained.  For the 
three better behaved antennas (15, 22 and 28) are viewed, we find for 
the autocorrrelation amplitudes, in Jy: (LR is the same as RL):

Ant         RR          LL        RL  
----------------------------------------
15           276         264       250
22          286          295       256
28          292          282       259

    The ratio RL/(avg RR,LL) is 0.90.  For the antenna with the most 
stable autocorrelations (15 -- see below), the ratio is 0.92.   Much 
better. 

    The program VPLOT plots amplitude of various 
baselines/antennas/polarizations/channels against time.  The 
autocorrelation plots show quite a remarkable difference in stability 
between antennas and polarizations ...
    On the RCP side, antennas 15, 22, and 28 give beautifully crisp 
square waves of about the corrrect amplitude as we go on and off 3C286. 
    On the LCP side, only antenna 15 gives similarly crisp patterns.  
The others -- particularly antenna 22 and 12, are severely degraded -- 
we can barely make out the square wave pattern.