• Eruptive processes
• crystallization
• Averno 2
• magma storage

The Averno 2 eruption (AV2) occurred 3.7 ka b.p in the northwest sector of the Campi Flegrei caldera (CFc), at the intersection of NE-SW and NW-SE fault systems bordering the resurgent block. The eruption, one of the youngest of the caldera, was followed only by the 1538 AD eruption of Monte Nuovo. The eruption of AV2 was fed by alkaline magma tapped from a chemically zoned (weakly peralkaline alkali-trachyte to alkali-trachyte) chamber [Civetta et al., 1988, Rosi and Sbrana 1987]. Eruptive products of AV2 eruption were subdivided into three members (A to C) by Di Vito et al., [2001]. 21 representative samples of AV2 deposits were analyzed for grain-size and componentry whereas 16 depositional units (about 100 juvenile clasts each) were sampled to measure bulk density, bulk vesicularity and assess texture characteristics of individual clasts. According to bulk density, vesicularity and textural data three main pyroclasts types were identified: i) low-density, light coloured, microvesicular, microlite-free; ii) high-density, dark-brown, microlite-rich; iii) banded made up of mm-thick stripes of (i) and (ii) types. The three types record variations of magma properties resulting from various degassing histories and syn-eruptive crystallization of groundmass. Characteristics and relative abundance of juvenile clast types throughout the stratigraphic sequence were used to make inferences and place constrains on temporal variations in magma rise and eruptive mechanisms. Occurrence in the very early fallout bed of dense juvenile clasts suggests the initial disruption of a limited volume of degassed magma. Member A was dominated by type (i) clasts suggesting that plinian episodes were fed by the fast supply of volatile-rich highly evolved magma from below whereas subsequent eruptions were fed by volatile poor-magma less evolved (member C). Members B and C contain (i), (ii) and (iii) types; an increasing abundance through time of clasts (ii) and (iii) suggests increasing importance of intra-eruption degassing, mostly occurring during time breaks between two successive eruptive pulses. Additionally, these degassing pulses were likely responsible for the emplacement of the fine-ash fall out beds. The pulsating activity of AV2 appears to be transitional in style between subplinian and vulcanian. We propose that transitions in eruptive style from plinian to surge activity were controlled by magmatic processes and not by water-magma interaction. This study also allows us to characterize the magma chamber of AV2. Petrography and chemical analysis (SEM-EDS and electron microprobe) were performed on pumice samples, melt inclusions, and host minerals. The least evolved melt inclusions are found in zoned diopside clinopyroxene phenocrysts from the pumice fall of member C (Cmb layer), whereas the most evolved melt inclusions are found in hedenbergite clinopyroxene phenocrysts from the pumices fall of member A (A0 layer), the first-erupted pumices. The observed compositional trend indicates that the erupted deposits were drained by a single magmatic reservoir. The compositional evolution of the AV2 magmatic system seems to have resulted from different processes occurring at different stages, pre-eruptive magma mixing, due to an input of a small batche of mafic magma into the reservoir, and then crystal fractionation to produce the most evolved compositions. The pumice products are characterised by a heterogeneous mineral population represented in part by crystals formed at the margin of the reservoir and a syn-eruptive mixing is identified in intermediate samples during the magma extraction. The H2O concentration determined by FTIR measurements of the melt inclusions suggest that this magma body resided and crystallized at shallow level prior to eruption, possibly about 4 km depth.