### Abstract:

An experiment was conducted during the period November 1967
to November 1968 at Oregon State University (OSU) and the Umatilla
Branch Experiment Station (BES) at Hermiston, two trials at each
location, to answer several questions: (1) could comparable rations
for swine be formulated by a computer on a least-cost basis; (2)
would the rations produce equal results when fed; and (3) what effect
would the switching of these rations during the feeding period have
on the rate of gain, feed efficiency, and carcass characteristics of
the pigs.
At OSU there were 45 pigs in each trial divided into 15 groups.
Each of these groups received all rations in different sequences. At
BES there were 168 pigs in each trial divided into 14 groups. Eight
of the groups, in replicates of two groups per ration, were fed a
given ration for the entire trial; the other six groups received all
rations in different sequences. A total of seven different rations were fed in the two trials.
Six rations were formulated by the computer on a least-cost basis;
the other was a standard experiment station ration. Three different
rations were used in each trial and the standard ration was used in
both trials.
Analysis of the data from the experiment showed that: (1) rations
for swine can be formulated by computer on a least-cost basis
if care is taken to insure that the input data given the computer to
formulate the rations is correct; (2) these rations will produce nearly
equal results when fed if both the economic and physical aspects are
considered and not either alone; (3) switching the rations during the
feeding period will produce little effect on the pigs for rate of gain,
feed efficiency, or carcass characteristics. The significant differences
between sequences, direct ration effects, residual ration effects,
and constant ration vs. switched ration were not consistent
and could not be interpreted to indicate that any of the above conclusions
were invalid.