Stellar models indicate that the core compactness of a star, which is a common proxy for its explodability in a supernova, does not increase monotonically with the star's mass. Rather, the core compactness dips sharply over a range of carbon-oxygen core masses; this range may be somewhat sensitive to the star's metallicity and evolutionary history. Stars in this compactness dip are expected to experience supernovae leaving behind neutron stars, whereas stars on either side of this range are expected to form black holes. This results in a hypothetical mass range in which black holes should seldom form. Quantitatively, when applied to binary stripped stars, these models predict a dearth of binary black holes with component masses . The population of gravitational-wave signals indicates weak evidence for a dip in the distribution of chirp masses of merging binary black holes near . This feature could be linked to the hypothetical component mass gap described above, but this interpretation depends on what assumptions are made of the binaries' mass ratios. Here, we directly probe the distribution of binary black hole component masses to look for evidence of a gap. We find no evidence for this feature using data from the third gravitational-wave transient catalogue (GWTC-3). If this gap does exist in nature, we find that it is unlikely to be resolvable by the end of the current (fourth) LIGO-Virgo-KAGRA (LVK) observing run.