Apatite's ubiquity in crystalline rocks, variable trace element contents (particularly with regard to the REE, actinides and Sr), and amenability to various datingtechniques based on the decay of the radioisotopes U and Th, permit specific provenance determinations. In this study, we first present a comprehensive descriptionof the trace element behaviour of apatite in various kinds of bedrocks (igneous rocks from felsic through to ultramafic compositions, metamorphic rocks from low tohigh grades and of diverse protolith composition, and authigenic apatite) in which we explain why apatite is so highly diverse in terms of its trace elementcomposition. Next, we present a synthesis of bedrock apatite trace-element compositional data from previous work, assembling a library of apatite compositions thatincludes the most abundant apatite-bearing lithologies in the Earth's crust, and many other less abundant rock types. Compositional statistics, classification, and amachine learning classifier are then applied to this dataset to generate biplots that can be used to determine the broad source lithology of detrital apatite, withmisclassification averaging 15%. This methodology is tested in three case studies to demonstrate its utility. In these examples, detrital apatite can be convincinglylinked to different lithology types, and combined apatite trace-element and UePb data can determine the terranes from which individual apatites were likely derived.The addition of apatite trace-element information therefore enables the determination of the source lithology, making the extraction of novel information and morespecific provenance determinations possible, and opening up new avenues in source-to-sink modelling.