The core-mantle boundary (CMB) is the most extreme discontinuity in Earth's interior and plays an important role in regulating planetary scale processes, including convection in both the mantle and core. The CMB is usually treated as a sharp discontinuity with direct contact between the silicate mantle and iron core, a scenario that is rarely questioned. However this is likely an over-simplification. This thesis investigates, using seismology, the possibility that the CMB is a layered transition, with an intermediate kilometre-scale layer sandwiched between the core and mantle.

This thesis uses seismic data from opposite ends of the frequency spectrum - high-frequency body waves and long-period normal modes. Before applying body waves to the CMB, this thesis first addresses the problem of ellipticity corrections for seismic phases such that they can be applied correctly to any seismic phase in any planetary model. Following this advancement, the effect of a thin layer at the CMB on P waves is examined, finding that a previously underutilised diffracted phase, PKKP_diff, is very sensitive to the inclusion of even very thin layers. A global dataset of over 12,500 PKKP_diff observations and 353 normal mode centre frequencies are then applied in turn to attempt to resolve whether a thin layer exists at the CMB.

Despite being vastly different data types with differing frequency contents and sensitivities, the two studies come to the same conclusion - not only is a slow kilometre-scale layer at the CMB possible within the bounds of seismic data, but both data types are better-fitted if such a structure on the order of a kilometre thick exists at the CMB. The favoured seismic parameters are a density increase of tens of percent coupled with similar magnitude reductions in both P- and S- velocity, with S-velocity more reduced than P. Both studies have their own limitations that inhibit robust conclusions that the CMB is truly layered. Nevertheless the results of this thesis suggest that this is a plausible scenario that is permissible within the bounds of current seismic data, and should therefore be considered appropriately. This conclusion has wide reaching implications which are also briefly explored.