The abundance of water on Earth has a profound impact on how our planet operates: aqueous fluids facilitate melting and subduction, and hence plate tectonics, control the redistribution of elements to make ore deposits, and determine the availability of (trace) element nutrients to life. Knowledge of the compositions of these fluids is therefore essential. Unfortunately, direct fluid samples are rare, especially for deep in the Earth and for its earliest history. However, minerals with preserved compositions are readily available, sampling environments to more than 200 km depth and back to at least 4.2 billion years ago. Minerals capture a fingerprint of the associated fluid by element exchange, and in my research I investigate how to read this mineral record. This involves understanding how elements partition in mineral-fluid and mineral-melt partitioning experiments from surface conditions to high temperature and pressure; computational modelling of mineral lattices and aqueous complexes, and in-situ trace element work on natural samples. The key mineral used in these studies is tourmaline, although we also work on fluorite, albite, chlorite and perovskite.