Classifications made by the powerful human grouping instinct are dilat
ed (swollen out, enlarged) around any groups that are probably incompl
ete for some reason. For example, evidence may suggest that some taxa
of a group may have been poorly sampled by collectors. Some small grou
ps may be important remnants of major fossil taxa or may have undiscov
ered allies. Dilating classifications around such incomplete groups ma
y raise them to higher, more appropriate ranks. Dilation control is ra
rely provided in computer programs. The Uniter program lets the user g
ive more dilation around the smaller clusters. It does so more for the
later-linking clusters, based on an assumption that these are more li
kely to have missing allies. For clusters of various sizes, the user c
an control their individual prominence or concealment. Methods are com
pared and discussed with examples. It is concluded that controlled dil
ation is essential for the high resolving power one expects to find in
computer based classifications of evolutionary groups.