In previous phylogenetic analyses the hypothesis of homology of the furcula of birds and some other maniraptorans with the clavicles of ancestral archosaurs has been protected from the test of congruence through ad hoc hypotheses of non-preservation. Cladistic evaluation of the distribution of the evidence for clavicles and furculae within the Dinosauria casts some doubt on the homology of the furculae and clavicle-like structures that characterize various clades within the Maniraptora with the plesiomorphic archosaurian clavicle. Depending on the degree to which the absence of clavicles in particular clades is accepted as real, the test of congruence using the currently accepted phylogenetic relationships within the Ornithodira suggests that the loss of clavicles may be plesiomorphic within Dinosauria. This raises the possibility that the furcula of birds may be a neomorph, or may represent the reappearance of a “lost” structure. The test of congruence depends on acceptance of negative evidence for clavicles in particular clades. Preference for the alternative explanation of non-preservation seems to be based, at least in part, on a priori acceptance of the very homology statement that one wishes to test. Provisional acceptance of the absence of structures in fossils forms the basis for hypotheses that are subject to falsification and provides the only means whereby certain homologies can be tested via congruence.