The scans *will* have the tower centered in different, non-indexed
beams.  But as long as we "collect", or add up all the energy that all
the individual tower pulses return, we should have consistent numbers.
In that regard, we have pulses at 500/second, and with a scan rate of
about 8.0 deg/sec, our pulse separation is 8/500 or .016 deg.  No pulse
can be more than half this difference from that pulse in a previous
scan, so we have azimuth matching in pulses that is accurate to .008
degrees.

Since each beam is really a linear sum of 50 pulses, linearly summing
six beams gives you a linear sum of 300 pulses which (if properly
centered) should really contain all the energy returned from the
tower.  There could still be differences due to un-matched azimuths
for the pulse, or due to phase cancelling within a pulse (and the
phase cancelling could change with changes in refractive index), but
these changes seem to be small.  Note that these small differences
*might* be able to explain why the variance in one polarization is
greater than the variance in the other.