Unavoidable variation in experimental units
Identical experimental units result in the most accurate estimates of the effect of a factor. In practice however, we usually have little choice and the available experimental units are often very variable.
If we have no prior knowledge of the experimental units, a completely randomised experiment is the best possible design.
Matched pairs
Knowing something about the characteristics of the experimental units allows us to do better.
A simple example arises when there are two treatments. A matched pairs design groups the experimental units into pairs that are similar (using either numerical measurements about the units or a subjective judgement). The two treatments are randomly allocated within each pair.