Data-driven discovery in mineralogy: Insights from natural kind clustering

Mineralogists have long asked how many different kinds of minerals occur on Earth, if that number has changed through time, and how Earth's mineral diversity compares to that of other planets and moons. Those questions have been explored recently in the contexts of mineral evolution, mineral ecology, and mineral network analysis, which consider how physical, chemical, and biological processes affect the temporal and spatial diversity and distribution of minerals. These questions are important for geology, because minerals have the potential to reveal the nature of the co-evolving geosphere and biosphere through billions of years of Earth history.

Mineral classification systems have the potential to illuminate this rich evolutionary story. However, the present mineral taxonomy of the International Mineralogical Association (IMA) is based on idealized major element chemistry and crystal structure. It does not consider the information-rich attributes of minerals—the trace elements, isotopes, inclusions, and other non-idealities that emerge during a planet's evolution. A complementary evolutionary system of mineralogy based on the quantitative recognition of "natural kind clustering" for a wide range of condensed planetary materials with different paragenetic origins has the potential to amplify the present classification system.

Natural solids, including non-crystalline condensed materials not represented by the current IMA formalism, can be categorized according to their distinctive combinations of non-ideal attributes. Cluster analysis based on the observed range of properties in natural specimens thus complements and amplifies the present mineral classification IMA scheme. By recognizing natural kinds of planetary materials, this approach to mineral classification incorporates an evolutionary component in addition to chemistry and structure. Based on this genetic, evolutionary definition of natural mineral kinds, at least three categories of planetary materials are imperfectly catalogued by the present mineral classification system: (1) distinct natural kinds that have been lumped in the IMA classification; (2) individual natural kinds that have been split by the IMA classification; and (3) non-crystalline materials.