Volcanism - the generation and eruption of molten rock from within the earth's interior - is one of the most visible manifestations of plate tectonics. Growth of the earth's crust occurs either when magma is stored and solidified within the crust, or is erupted at the earth's surface. Eruptive activity at subduction zones can be explosive and highly disruptive, and represents an important natural hazard, with implications for life, health and financial stability when it occurs. One of the major challenges facing volcanologists is the accurate forecasting of this eruptive behaviour. Abundant evidence of past volcanic activity shows that large volumes of magma can be erupted in a single event. However, geophysical techniques used to image below the earth's surface fail to distinguish large volumes of melt (magmatic liquid) stored within the crust. Instead, melt may be stored as "crystal mush", i.e. an accumulation of volcanic crystals separated by only small amounts of melt that is hard to image geophysically. However, a crystal mush with low melt content behaves like a solid and cannot be erupted. Researchers therefore suggest that the mush contains 'eruptible' lenses that have higher melt content, yet remain thin enough to be unresolved by geophysical techniques. If so, then wholesale spatial reorganisation of crystals and liquid in the whole mushy region could change its overall physical behaviour, such that it quickly becomes eruptible. In contrast, other scholars predict a prolonged existence of more liquid-rich (potentially eruptible) mush bodies within the crust. In this case, the lack of currently observed geophysical signals for large, melt-rich magma bodies may simply result from the ephemeral nature of magmatism. To make progress, more information about the longevity of eruptible mushy regions is essential. This proposal will develop a new method to determine the lifetime of melt-rich regions, enabling us to resolve this current conflict. Time 'chronology' information about volcanic systems is commonly recorded in the mineral zircon, which contains radioactive elements that are sensitive to time. Zircon chronology shows that crystal mushes can persist over long time periods (e.g. 100s kyr), but these measurements hold significant uncertainties. The lifetime of the more eruptible, melt-rich 'mobile magma' is much harder to investigate, because it occurs at higher temperatures where zircon may not be stable. However, this information is a critical link between geophysical observations, which record a snapshot of the state of the earth's crust, and volcanology, which records information about magmatic processes over very long times. This project will develop a new method to determine the lifetime of mobile magma crystallisation directly by analysing crystals that grow from melt at high temperatures. Specifically, we will relate the aspect ratio (length/ width) of the silicate mineral plagioclase, which grows from almost all subduction zone magmas, to the time available for crystallisation. Our preliminary work suggests a strong relationship between aspect ratio and time for water-rich, silica-rich magmas that erupt at subduction zones. Using high-temperature experiments, analysis of well-dated plagioclase crystals, and mathematical approaches, the team will derive a universal relationship that can be applied to all magmatic environments. We will apply the method to intermediate subduction zone volcanic systems that have recent geophysical information, in order to re-evaluate the architecture of the subterranean magma plumbing systems. Finally, we will integrate our crystal-scale observations with existing geophysical information and chronology datasets, to bring new insights into the distribution of melt and our ability to see it geophysically. This will lead to novel constraints on the identification, recognition and definition of mushy plumbing systems in future.