Cladograms predict the order in which fossil tara appeared and, thus,
make predictions about general patterns in the stratigraphic record. I
nconsistencies between cladistic predictions and the observed stratigr
aphic record reflect either inadequate sampling of a clade's species,
incomplete estimates of stratigraphic ranges, or homoplasy producing a
n incorrect phylogenetic hypothesis. A method presented in this paper
attempts to separate the effects of homoplasy from the effects of inad
equate sampling. Sampling densities of individual species are used to
calculate confidence intervals on their stratigraphic ranges. The meth
od uses these confidence intervals to test the order of branching pred
icted by a cladogram. The Lophospiridae (''Archaeogastropoda'') of the
Ordovician provide a useful test group because the clade has a good f
ossil record and it produced species over a long time. Confidence inte
rvals reject several cladistic hypotheses that postulate improbable ''
ghost lineages.'' Other hypotheses are acceptable only with explicit a
ncestor-descendant relationships. The accepted cladogram is the shorte
st one that stratigraphic data cannot reject. The results caution agai
nst evaluating phylogenetic hypotheses of fossil taxa without consider
ing both stratigraphic data and the possible presence of ancestral spe
cies, as both factors can affect interpretations of a clade's evolutio
nary dynamics and its patterns of morphologic evolution.