Each bottle contributes several rows tagged (avg) / (sdev) / (min) /
(max). Only (avg) is read, for two reasons: the others describe the scatter
of the scans the bottle was fired over rather than separate observations, and
structurally they are not the same table — the (sdev) row omits the bottle
number and carries a time where the (avg) row carries a date, so it does not
share the column layout.