The history of climate research in the 20th century has been characterised by a crucial shift from a geography-oriented, two-dimensional approach towards a physics-based, three-dimensional concept of climate. In the 1930s, the introduction of new technology, such as radiosondes, enabled climatologists to investigate the high atmosphere, which had previously been out of reach. This “conquest of the third dimension” challenged the surface-oriented, geographical notion of climate patterns and opened up climatology to a three-dimensional approach, which deeply changed the character of (...) climate research. Two decades later, by drilling deep into polar glaciers and using the downward vertical dimension as an archive of the earth's history, ice core scientists began to reconstruct past climates layer by layer. The data retrieved in deep glacial layers contributed crucially to a temporal expansion of climate history far beyond human timescales. However, the inaccessibility of glaciers and the practical challenges of bringing fragile fragments of ice into transnational networks of scientific exchange meant that this vertical extension of climate knowledge production proceeded through a range of new scientific practices, and was shaped by new forms of international collaboration. Furthermore, this vertical approach to glaciers also asked for a new understanding of glacier volume. Drawing on archival and printed sources, I argue that ice core research represented a second discovery of the third dimension, this time downwards into the depth of the earth's surface, but again with decisive consequences for the research practice, for collaboration politics, and for understandings of climate, spatially as well as temporally. (shrink)