[1] Compared to "dry" atmospheric eruption of magma or "dry" magma/rock contact, intensity and time scale of heat discharge from magma to the surroundings is strongly modified by an effective coolant: water or water-sediment mixes. In the case of subaqueous or subglacial eruptions magma-water contact must take place and can result in phreatomagmatic explosions. Even if no explosions occur, rapid cooling results in the formation of pyroclasts by thermal granulation. To study this process in detail, a short-term calorimeter was built for the direct measurement of the heatflux from a magmatic melt to a coolant. Volcanic rocks from recent eruptions in Iceland were remelted and used to produce jets of melt poured into a coolant-filled container. Particles could be produced in a non-explosive process, that are practical identical to those from natural hyaloclastites. The process' fragmentation energy is about 10% of the total heat transferred from melt to coolant. Citation: Schmid, A.,

[1] Much of the volcanism on Earth takes place in subaqueous settings where magma has direct contact with a water reservoir of restricted or quasi unrestricted volume. In order to assess the intensity and timescale of non-explosive interaction of magmatic melts and water, experiments representing these settings were performed. Natural volcanic samples were remelted and poured as a continuous jet into a water-filled calorimeter where the melt interacts with its coolant. The rapid cooling results in granulation, i.e. brittle failure of the material. Granulation needs energy, which is taken from the thermal input of the hot melt. Energy used in granulation was found to require 5%-20% of the melt's initial heat content. This energy loss fraction is insensitive to variations in coolant-and melt temperatures but instead depends on the melt's thermomechanical properties. However analysis of the experimentally produced granulate indicates a strong correlation between the initial coolant temperature -i.e. the heat sink-and the grain-size distribution, but also shows variations due to material properties. The maximum of the grain-size distribution was determined to change from a diameter of 1 mm up to 4 mm due to coolant temperature increase. Properties of the heat source (melt ) dominate the efficiency of the process, whereas both heat sink and source characteristics determine the products.