Due to characteristics of the EA038 observations (short integrations,
changing array composition, single-pol), the normal standard plots
weren't optimal for summarizing the experiment.  We therefore made a
different set of plots manually, as described below.  The names of the
plot files and a brief synopsis of their contents are:

  ea038.{pass}-weight.ps.gz               station weight vs. time 
  ea038.{pass}-auto-[1,2,3].ps.gz         station autocorrelation real vs. freq
  ea038.{pass}-cross-[1,2,3][a,p].ps.gz   baseline amp/phs vs. freq
  ea038.{pass}-ampphase-[1,2][a,p].ps.gz  Ef-* baseline amp/phs vs. time

where {pass} = "cont" or "line" (the cont & line plots show the same
extracts of the data from the respective runs, unless specified).
The weight plots show the whole experiment.  The auto & cross plots
show 2 minutes of data taken near from [1] scan 7 (3C345), [2] scan 45
(J1928+0848), and [3] scan 143 (3C454.3).  The ampphase plots show 
the whole experiment without any target sources, and the amp(t) plots
also omit the fringe finders to avoid throwing off the y-axis scale.
In baseline plots, the 'a' or 'p' denotes an amp or a phase plot.

(a) WEIGHT plots - plot each 0.25s integration plotted as a separate ".".
For the continuum plot, the subband color-coding is:  SB0=red,  SB1=blue,  
SB2=cyan, SB3=green.  Colors overplot, so only green is visible on the plot 
(i.e., no individually bad subbands).  The line correlation has only
one subband.

(b) AUTOCORRELATION SPECTRA plots - show real vs. freq for each station,
using 2min of data from the middle of the scans listed above.  The y-axis
range doesn't necessarily go down to zero.  

(c) BASELINE SPECTRA plots - show (separately) amplitude and phase vs.
frequency for all baselines for the same 2-minute intervals (vector averaged) 
form the standard auto-plots.  The y-axes range in each amplitude plot
self-scales independently.  Phase plots all show +/- 200 degrees.

(d) AMP/PHASE(t) plots - show (separately) amplitude & phase for Ef-* 
baselines as a function of time for the whole experiment for non-target
sources.  The amp(t) plots further omit the fringe-finder sources, which
would otherwise have dominated the y-axis scale.  There are two versions
of the  continuum plots, in which only one subband is shown, as denoted
by the numeral in the filename:
  [1] - SB0; no Cm
  [2] - SB2; no Da,Df,Kn
Each 0.25s integration is plotted as a separate ".", with the middle 80%
of the band used for vector-averaging in frequency.