Drugs are often metabolized and excreted from the body at different
rates, depending on how the drug was administered. This is measured by
the concentration of the drug in the bloodstream over
time. Researchers are interested in comparing a liquid, tablet, and
capsule form of a drug; each treatment contains the same quantity of
active drug. Large subject-to-subject differences are expected, so the
researchers need a design that can account for that.

A three period double-blind crossover design with washout periods is
set up with four sets of three subjects (12 participants total)
randomized according to a Latin square design.  Each drug will be
given to each participant once in one period in the order determined
by the Latin square. The response of interest is the area under the
time-concentration curve, which is proportional to the average amount
of drug in the bloodstream.

Question 1: Does the treatment effect differ by Latin square?

Question 2: Can the three forms of the drug be considered equivalent in
terms of area under the time-concentration curve?

The data are found in the file {\tt bioequivalence.dat} and the coding
used in the file is found in \newline {\tt bioequivalence.readme}.

These data are taken from Oehlert (2000) Table 13.3.