[evlatests] Correction

[Rick Perley]

    Previous circulars by me stating that the phase spin was not 
accompanied
by a loss of correlation amplitude are incorrect.  I am relieved to 
conclude
that upon careful review, the expected loss of coherence from the constantly
rotating phase is indeed seen in the data.   I am thankful to Craig for 
reminding
me that the amplitude loss is smaller than you may expect -- it goes as
sinc(f), where f = the number of turns in phase in the integrating period. 
With a 60 degree phase wind, f = 1/6, and sinc(1/6) = .958 -- not a large
loss. 

    The reporting error was caused by the loss being masked by other
effects.  Fortunately, the loss is fairly clearly seen in the spectral line
data, which was taken with 3.3 second averaging.  

    However, with the 416.6 msec averaging, the effect is not easily
discernible -- even though the phase wind is as high as 70 degrees for
each 416 msec. point.  There is a slow secular change in gain which
is larger than the 6% loss expected in the averaging. 

    A remaining mystery is why the phase spin is faster in the 416 ms
data than the 3.3 second data by the ratio of integration times:  8.0