The grain-size distribution of pyroclasts: Primary fragmentation, conduit sorting or abrasion?

Explosive volcanic eruptions expel a mixture of pyroclasts and lithics. Pyroclasts, fragments of the juvenile magma, record the state of the magma at fragmentation in terms of porosity and crystallinity. The grain size distribution of pyroclasts is generally considered to be a direct consequence of the conditions at magma fragmentation that is mainly driven by gas overpressure in bubbles, high shear rates, contact with external water or a combination of these factors. Stress exerted by any of these processes will lead to brittle fragmentation by overcoming the magma's relaxation timescale. As a consequence, most pyroclasts exhibit angular shapes. Upon magma fragmentation, the gas pyroclast mixture is accelerated upwards and eventually ejected from the vent. The total grain size distribution deposited is a function of fragmentation conditions and transport related sorting. Porous pyroclasts are very susceptible to abrasion by particle-particle or particle-conduit wall interaction. Accordingly, pyroclastic fall deposits with angular clasts should proof a low particle abrasion upon contact to other surfaces. In an attempt to constrain the degree of particle interaction during conduit flow, monomodal batches of washed pyroclasts have been accelerated upwards by rapid decompression and subsequently investigated for their grain size distribution. In our set-up, we used a vertical cylindrical tube without surface roughness as conduit. We varied grain size (0.125-0.25; 0.5-1; 1-2 mm), porosity (0; 10; 30 %), gas-particle ratio (10 and 40%), conduit length (10 and 28 cm) and conduit diameter (2.5 and 6 cm). All ejected particles were collected after settling at the base of a 3.3 m high tank and sieved at one sieve size below starting size (half-Φ). Grain size reduction showed a positive correlation with starting grain size, porosity and overpressure at the vent. Although milling in a volcanic conduit may take place, porous pyroclasts are very likely to be a primary product of magma fragmentation at or close to the fragmentation level. Given the high abrasiveness of pumice, hemispherical clasts should be observed if clast break-up followed efficient clast abrasion. As a consequence, finer grained pyroclastic fall deposits do not necessarily proof efficient secondary fragmentation in the conduit but may rather reveal the influence of conduit length on 'What size of pyroclasts can be erupted'?