Astronomy has entered a new era in which a large number of exoplanets has been discovered (more than 5500). In the search for an Earth twin, and possibly alien life, understanding the composition and fate of these worlds is of paramount importance. Nature provides an unlikely source of information in the form of dead stars called white dwarfs. Many white dwarfs have accreted remnants of their own planetary systems, causing their atmospheres to become polluted by heavy elements. The relative quantities of these elements contain a wealth of information about the composition of the accreted planetary bodies.

This thesis focuses on the interpretation of white dwarf pollution, with the aim of better understanding planetary composition and evolution. There are a large number of processes that can affect, and potentially explain, the composition of planetary bodies. For example, exposure to high temperature can remove volatile elements. The formation of an iron-rich core is another significant process: certain white dwarfs exhibit iron-rich (or iron-poor) pollution, which are often interpreted as the accretion of core (or mantle) material.

However, given a set of elemental abundances, identifying the physical processes which best explain the data is a highly non-trivial exercise. Bayesian modelling is a powerful method to disentangle the most likely explanation from the myriad of possibilities. This approach reveals evidence that core formation and volatile loss, which shape Solar System bodies such as Earth, also occur in other planetary systems.

In addition to their formation histories, polluted white dwarfs can be used to uncover the ultimate fate of planetary bodies as they are accreted. Different accretion scenarios alter the composition of detected pollution in a probabilistic way, which must be investigated at the population level. I use population synthesis to calculate the sample size required to distinguish between accretion scenarios.

The number of white dwarfs with detected pollution will increase in the coming years, largely due to follow-up of candidate systems identified by the Gaia mission. Applying the methodologies presented in this thesis to the resulting data will help us learn more about how planets form, about how they are ultimately destroyed, and about the Solar System's significance within the galaxy.