熔体包裹体对长白山天池火山千年大喷发的指示意义

长白山天池火山在全新世曾有过几次喷发,其中距今约1000年发生过大规模布里尼式爆炸喷发（即“千年大喷发”）,其喷发产物——灰白色碱流质浮岩和喷发柱垮塌形成的火山碎屑流分布范围极广,除长白山区外,在朝鲜半岛和日本北部也有大量浮岩降落和堆积。根据野外较大范围的系统采样、镜下观察和测试分析,在天池火山千年大喷发产物的碱性长石晶屑中发现了两组颜色、形态、化学成分迥异的“火口组”和“圆池组”熔体包裹体,对揭示天池火山千年大喷发的成因具有重要意义。根据电子探针分析结果,“火口组”熔体包裹体成分为英安岩和粗面英安岩,寄主晶多为透长石;“圆池组”熔体包裹体成分为粗面英安岩和流纹岩,寄主晶为歪长石。相对“火口组”熔体包裹体,“圆池组”包裹体具有高SiO2、高H2O和高Cl含量的特点,化学成分也更为演化,可能是天池火山千年大喷发时岩浆结晶分异后期的产物。两组包裹体的存在为千年大喷发前的层状地壳岩浆房和成分并非单一提供了证据,它们可能是在同次大喷发的不同序列中喷出的。由于地幔岩浆注入地壳岩浆房,导致不同层位岩浆的扰动和混合作用,因挥发分出溶在岩浆房最顶部形成挥发分梯度和过饱和,最终触发了天池火山的千年大喷发,对当时的气候环境造成过较大影响。     

长白山地区新生代火山岩及其幔源包体的矿物化学及矿物包裹体研究

长白山地区新生代火山岩主要为玄武岩类、粗面岩类和碱性流纹岩类。其中奶头山期碱性玄武岩中含有幔源的尖晶石二辉橄榄岩包体和辉石岩包体。幔源包体及不同期次火山岩的主要矿物是橄榄石、单斜辉石、斜方辉石、尖晶石、斜长石、碱性长石。不同寄主岩石中的矿物组成及其化学成分具有一定的变化，反映了岩浆分异演化的特征；矿物及火山熔岩中的包裹体成分及玻璃熔体结构的激光拉曼光谱分析结果表明，地下深处的岩浆含有较多的挥发分，岩浆上升过程中发生了强烈的出溶作用；岩浆由起源经分异演化上升到地壳浅部直至喷发，幔源的挥发分减少，而浅成或壳源的挥发分增多，特别是在岩浆喷发过程中，地下水或大气成分起了重要作用     

长白山地区新生代火山岩及其幔源包体的矿物化学及矿物包裹体研究

长白山地区新生代火山岩主要为玄武岩类、粗面岩类和碱性流纹岩类。其中奶头山期碱性玄武岩中含有幔源的尖晶石二辉橄榄岩包体和辉石岩包体。幔源包体及不同期次火山岩的主要矿物是橄榄石、单斜辉石、斜方辉石、尖晶石、斜长石、碱性长石。不同寄主岩石中的矿物组成及其化学成分具有一定的变化，反映了岩浆分异演化的特征；矿物及火山熔岩中的包裹体成分及玻璃熔体结构的激光拉曼光谱分析结果表明，地下深处的岩浆含有较多的挥发分，岩浆上升过程中发生了强烈的出溶作用；岩浆由起源经分异演化上升到地壳浅部直至喷发，幔源的挥发分减少，而浅成或壳源的挥发分增多，特别是在岩浆喷发过程中，地下水或大气成分起了重要作用。     

云南腾冲新期火山岩矿物及其熔体包裹体研究

腾冲火山群位于我国云南省西部和缅甸的交界处,由黑空山、大空山、小空山、打鹰山、马鞍山等一系列火山组成,是我国著名的第四纪火山群,从喷发活动时间上可分为老期火山和新期火山.前人研究证实,打鹰山、马鞍山和黑空山为新期火山,它们在全新世都有过喷发活动.这些新期火山岩的主要斑晶矿物为辉石、橄榄石和长石,主要的斑晶矿物中都含有熔体包裹体,它们形态多样,分布不规则,且部分显示后期有变化.探针分析表明,新期火山岩矿物中熔体包裹体成分的变化范围大于其基质玻璃成分的变化范围,它们的化学成分分布范围涵盖了玄武粗安岩、粗面安山岩、粗面岩和流纹岩等岩石类型,与腾冲火山区晚更新世以来火山岩的成分分布范围基本一致.根据对新期火山岩斑晶和微晶矿物中熔体包裹体及基质玻璃成分的测试研究,其中挥发分Cl的含量在包裹体和基质中变化不大,但F、SO3在熔体包裹体中的含量高于基质,总体上腾冲新期火山喷发时岩浆的脱气率较低,喷发时并未向空中喷出大量气体,推测对当时的气候环境影响不大,但未来喷发的灾害效应不容忽视.     

长白山天池火山岩浆演化——来自主矿物成分的证据

长白山天池火山岩的主要矿物成分为橄榄石、辉石、长石和铁钛氧化物，经电子探针测试，其成分变化特征明显，自造盾阶段一造锥阶段一全新世喷发，橄榄石中铁含量增加，一直达到铁橄榄石的端元组分；单斜辉石中铁质和钠质含量增加；长石则从基性的拉长石向含钾、钠渐高的歪长石和透长石演化；铁钛氧化物含量则明显减少。三阶段斑晶矿物成分的变化反映了天池火山近200万年来岩浆演化的完整系列，符合分离结晶作用的演化趋势。     

我国五大连池、集宁碱性玄武岩矿物中的岩浆包裹体

本文研究了五大连池火山近代富钾玄武岩及集宁新生代碱性玄武岩矿物中的岩浆包裹体。所研究的岩浆包裹体属演化型岩浆包裹体。其主矿物经受了缓慢地冷却和结晶。我们可利用电子探针分析均一后的包裹体以获得初始化学成分。岩浆包裹体的测温测压结果表明,集宁碱性玄武岩中的橄榄石斑晶是在1295—1310℃和3.1—3.5kbar下结晶;五大连池火山富钾玄武岩的矿物斑晶是在岩浆缓慢上升和冷却过程中结晶。这表现为自橄榄石向单斜辉石,岩浆包裹体的均一温度(1308→1178℃)和CO_2压力(0.9→0.4kbar)逐渐下降。     

印度尼西亚克拉卡托火山安山质熔岩斑晶矿物中的岩浆包裹体

印度尼西亚克拉卡托火山安山岩的各类矿物斑晶中舍大量岩浆包裹体。根据岩浆包裹体均一温度和化学成分，可以建立如下斑晶矿物结晶顺序(由早至晚)：斜长石中心相(An83—72)、橄榄石→斜长石过渡相(An61)、单斜辉石、钛磁铁矿→斜长石边缘相(An54-42)。斜长石斑晶中心部分在5kb压力的岩浆深处结晶。高压环境和高钛含量，有利于早期斜长石斑晶所含岩浆包裹体中发生硅酸盐液体不混溶作用。这是板块构造消减带中安山质熔岩的特点。     

加勒比海圣卢西亚岛英安质熔岩的温压地球化学

加勒比海圣卢西亚岛最新喷发(<1百万年)的英安质熔岩属钙-碱性火山岩系。呈岩穹、岩钟状火山侵出体及猛烈喷发的浮岩相产出。根据矿物学及岩浆包裹体温压地球化学研究,英安质浮岩富含挥发组分,各类矿物斑晶在岩浆结晶分异晚期,于800-900℃、0.6-1.8千巴(P_(H_2O))条件下近于同时结晶析出;英安岩侵出体形成于猛烈爆发之后,气体大量逸出,挥发分含量较低,各类矿物斑晶于880℃、1千巴(P_(H_2O))条件下近于同时结晶析出。     

从神话、传说与文字记录中探索中国活火山喷发的信息

火山喷发文字记录资料的考证可以给出历史上火山喷发的最为准确的时间限定，对喷发过程、灾害效应也都可给出极为详细、准确的措施。长白山天池火山喷发的满语资料是除了汉语、朝语之处另一种极为重要的资料来源，目前已从有关满语神话传说中得到了若干有重要意义的火山与火山学信息。本文列出了3类、14条天池火山喷发的神话与传说，从中可以得知数千年以来天池火山发生过多次猛烈的与温和的周期性的喷发。天池火山喷发与火山泥石流、喷发后洪水泛滥有密切关系，泛滥物波及到距离天池火山460km以过的嫩江流域及珲春一带。在中国近代史上还有若干次火山喷发记录，经考察较为确定的1951年昆仑山西部于田县卡尔达西火山喷发是一次地下岩浆上侵、地表汽爆炸与喷发的火山喷发事件。     

火山岩中熔体包裹体研究进展

火山岩斑晶矿物中保存完好的熔体包裹体含有原始熔浆的成分和挥发分含量,带有原始岩浆的信息,是恢复火山喷发前岩浆化学成分和了解岩浆作用过程及源区特征的最好工具。本文概述了熔体包裹体研究的发展历程和国内外研究进展及应用,总结了目前对熔体包裹体的分析测试仪器,还对熔体包裹体均一实验和熔体挥发分及脱气作用展开了充分讨论。随着现代测试仪器精度的不断提高和测试应用范围的扩大,熔体包裹体研究从传统的高温热台研究转向地球化学研究和应用领域,可以更好地解释岩浆过程中的地质现象、了解岩浆演化,目前熔体包裹体的研究已进入更快发展阶段。     

Compositions of magmas and carbonate–silicate liquid immiscibility in the Vulture alkaline igneous complex, Italy

This paper presents a study of melt and fluid inclusions in minerals of an olivine-leucite phonolitic nephelinite bomb from the Monticchio Lake Formation, Vulture. The rock contains 50 vol.% clinopyroxene, 12% leucite, 10% alkali feldspars, 8% hauyne/sodalite, 7.5% nepheline, 4.5% apatite, 3.2% olivine, 2% opaques, 2.6% plagioclase, and < 1% amphibole. We distinguished three generations of clinopyroxene differing in composition and morphology. All the phenocrysts bear primary and secondary melt and fluid inclusions, which recorded successive stages of melt evolution. The most primitive melts were found in the most magnesian olivine and the earliest clinopyroxene phenocrysts. The melts are near primary mantle liquids and are rich in Ca, Mg and incompatible and volatile elements. Thermometric experiments with the melt inclusions suggested that melt crystallization began at temperatures of about 1200 °C. Because of the partial leakage of all primary fluid inclusions, the pressure of crystallization is constrained only to minimum of 3.5 kbar. Combined silicate–carbonate melt inclusions were found in apatite phenocrysts. They are indicative of carbonate–silicate liquid immiscibility, which occurred during magma evolution. Large hydrous secondary melt inclusions were found in olivine and clinopyroxene. The inclusions in the phenocrysts recorded an open-system magma evolution during its rise towards the surface including crystallization, degassing, oxidation, and liquid immiscibility processes.     

Mixing and differentiation in the Oruanui rhyolitic magma, Taupo, New Zealand: evidence from volatiles and trace elements in melt inclusions

Large pyroclastic rhyolites are snapshots of evolving magma bodies, and preserved in their eruptive pyroclasts is a record of evolution up to the time of eruption. Here we focus on the conditions and processes in the Oruanui magma that erupted at 26.5 ka from Taupo Volcano, New Zealand. The 530 km³ (void-free) of material erupted in the Oruanui event is comparable in size to the Bishop Tuff in California, but differs in that rhyolitic pumice and glass compositions, although variable, did not change systematically with eruption order. We measured the concentrations of H₂O, CO₂ and major and trace elements in zoned phenocrysts and melt inclusions from individual pumice clasts covering the range from early to late erupted units. We also used cathodoluminescence imaging to infer growth histories of quartz phenocrysts. For quartz-hosted inclusions, we studied both fully enclosed melt inclusions and reentrants (connecting to host melt through a small opening). The textures and compositions of inclusions and phenocrysts reflect complex pre-eruptive processes of incomplete assimilation/partial melting, crystallization differentiation, magma mixing and gas saturation. ‘Restitic’ quartz occurs in seven of eight pumice clasts studied. Variations in dissolved H₂O and CO₂ in quartz-hosted melt inclusions reflect gas saturation in the Oruanui magma and crystallization depths of ∼3.5–7 km. Based on variations of dissolved H₂O and CO₂ in reentrants, the amount of exsolved gas at the beginning of eruption increased with depth, corresponding to decreasing density with depth. Pre-eruptive mixing of magma with varying gas content implies variations in magma bulk density that would have driven convective mixing. Electronic Supplementary Material Supplementary material is available for this article at and is accessible for authorized users.     

Melt Inclusions in Olivine Phenocrysts: Using Diffusive Re-equilibration to Determine the Cooling History of a Crystal, with Implications for the Origin of Olivine-phyric Volcanic Rocks

A technique is described for determining the cooling historyof olivine phenocrysts. The technique is based on the analysisof the diffusive re-equilibration of melt inclusions trappedby olivine phenocrysts during crystallization. The mechanismof re-equilibration involves diffusion of Fe from and Mg intothe initial volume of the inclusion. The technique applies toa single crystal, and thus the cooling history of differentphenocrysts in a single erupted magma can be established. Weshow that melt inclusions in high-Fo olivine phenocrysts frommantle-derived magmas are typically partially re-equilibratedwith their hosts at temperatures below trapping. Our analysisdemonstrates that at a reasonable combination of factors suchas (1) cooling interval before eruption (<350°C), (2)eruption temperatures (>1000°C), and (3) inclusion size(<70 µm in radius), partial re-equilibration of upto 85% occurs within 3–5 months, corresponding to coolingrates faster than 1–2°/day. Short residence timesof high-Fo phenocrysts suggest that if eruption does not happenwithin a few months after a primitive magma begins cooling andcrystallization, olivines that crystallize from it are unlikelyto be erupted as phenocrysts. This can be explained by efficientseparation of olivine crystals from the melt, and their rapidincorporation into the cumulate layer of the chamber. Theseresults also suggest that in most cases erupted high-Fo olivinephenocrysts retain their original composition, and thus compositionsof melt inclusions in erupted high-Fo olivine phenocrysts donot suffer changes that cannot be reversed. Short residencetimes also imply that large unzoned cores of high-Fo phenocrystscannot reflect diffusive re-equilibration of originally zonedphenocrysts. The unzoned cores are a result of fast efficientaccumulation of olivines from the crystallizing magma, i.e.olivines are separated from the magma faster than melt changesits composition. Thus, the main source of high-Fo crystals inthe erupted magmas is the cumulate layers of the magmatic system.In other words, olivine-phyric rocks represent mixtures of anevolved transporting magma (which forms the groundmass of therock) with crystals that were formed during crystallizationof more primitive melt(s). Unlike high-Fo olivine phenocrysts,the evolved magma may reside in the magmatic system for a longtime. This reconciles long magma residence times estimated fromthe compositions of rocks with short residence times of high-Foolivine phenocrysts. KEY WORDS: melt inclusions; olivine; picrites; residence time; diffusion     

Understanding mafic-felsic magma interactions in a subvolcanic magma chamber using rapakivi feldspar: A case study from the Bathani volcano-sedimentary sequence,eastern India

The Ghansura Rhyolite Dome of the Bathani volcano-sedimentary sequence in eastern India originated from a subvolcanic felsic magma chamber that was intruded by volatile-rich basaltic magma during its evolution leading to the formation of a porphyritic andesite. The porphyritic andesite consists of rapakivi feldspars, which are characterized by phenocrysts of alkali feldspar mantled by plagioclase rims. Results presented in this work suggest that intimate mixing of the mafic and felsic magmas produced a homogeneous hybrid magma of intermediate composition. The mixing of the hot volatile-rich mafic magma with the relatively colder felsic magma halted undercooling in the subvolcanic felsic system and produced a hybrid magma rich in volatiles. Under such conditions, selective crystals in the hybrid magma underwent textural coarsening or Ostwald ripening. Rapid crystallization of anhydrous phases, like feldspars, increased the melt water content in the hybrid magma. Eventually, volatile saturation in the hybrid magma was reached that led to the sudden release of volatiles. The sudden release of volatiles or devolatilization event led to resorption of alkali feldspar phenocrysts and stabilizing plagioclase, some of which precipitated around the resorbed phenocrysts to produce rapakivi feldspars.     

Pre- and syn-eruptive degassing and crystallisation processes of the 2010 and 2006 eruptions of Merapi volcano,Indonesia

The 2010 eruption of Merapi (VEI 4) was the volcano’s largest since 1872. In contrast to the prolonged and effusive dome-forming eruptions typical of Merapi’s recent activity, the 2010 eruption began explosively, before a new dome was rapidly emplaced. This new dome was subsequently destroyed by explosions, generating pyroclastic density currents (PDCs), predominantly consisting of dark coloured, dense blocks of basaltic andesite dome lava. A shift towards open-vent conditions in the later stages of the eruption culminated in multiple explosions and the generation of PDCs with conspicuous grey scoria and white pumice clasts resulting from sub-plinian convective column collapse. This paper presents geochemical data for melt inclusions and their clinopyroxene hosts extracted from dense dome lava, grey scoria and white pumice generated during the peak of the 2010 eruption. These are compared with clinopyroxene-hosted melt inclusions from scoriaceous dome fragments from the prolonged dome-forming 2006 eruption, to elucidate any relationship between pre-eruptive degassing and crystallisation processes and eruptive style. Secondary ion mass spectrometry analysis of volatiles (H₂O, CO₂) and light lithophile elements (Li, B, Be) is augmented by electron microprobe analysis of major elements and volatiles (Cl, S, F) in melt inclusions and groundmass glass. Geobarometric analysis shows that the clinopyroxene phenocrysts crystallised at depths of up to 20 km, with the greatest calculated depths associated with phenocrysts from the white pumice. Based on their volatile contents, melt inclusions have re-equilibrated during shallower storage and/or ascent, at depths of ~0.6–9.7 km, where the Merapi magma system is interpreted to be highly interconnected and not formed of discrete magma reservoirs. Melt inclusions enriched in Li show uniform “buffered” Cl concentrations, indicating the presence of an exsolved brine phase. Boron-enriched inclusions also support the presence of a brine phase, which helped to stabilise B in the melt. Calculations based on S concentrations in melt inclusions and groundmass glass require a degassing melt volume of 0.36 km³ in order to produce the mass of SO₂ emitted during the 2010 eruption. This volume is approximately an order of magnitude higher than the erupted magma (DRE) volume. The transition between the contrasting eruptive styles in 2010 and 2006 is linked to changes in magmatic flux and changes in degassing style, with the explosive activity in 2010 driven by an influx of deep magma, which overwhelmed the shallower magma system and ascended rapidly, accompanied by closed-system degassing.     

熔体包裹体挥发分应用的研究进展

挥发分(例如H2O、CO2、F、Cl和S)是地幔的重要组成部分,虽然它们在地幔中的含量很低,但是在地幔熔融和熔体演化、地幔不均一、地幔流变学、地幔地震特性和电导率等研究方面具有重要作用。对矿物熔体包裹体和玻璃挥发分的研究已经成为当前的研究热点。其中,熔体包裹体研究凭借其独特的优势成为研究地幔和岩浆挥发分组成的重要手段。熔体包裹体直接捕获了矿物形成时岩浆中的成分,且由于寄主矿物的存在使得熔体包裹体能够保持独立演化而不受外界环境影响,因此能够较为完整地保存岩浆中的挥发分信息。同时,研究熔体包裹体中的挥发分是恢复岩浆喷发前挥发分含量最直接的途径。如果通过现代分析方法(如扫描电镜、电子探针和离子探针等)对熔体包裹体进行详细的岩相学观察以及对后期可能影响熔体包裹体原始挥发组分的作用(地壳混染、岩浆去气、扩散和水化作用)进行评估,并结合实验研究熔体包裹体被捕获后发生的变化而对数据进行矫正,那熔体包裹体对研究岩浆体系中的挥发分将大有可为。基于此,本文系统介绍了熔体包裹体挥发分研究的现状及主要研究内容,主要包括熔体包裹体挥发分的测试方法、挥发分在岩浆中的溶解度、判断挥发分数据可靠性和挥发分研究的经典应用等4个方面。     

Magma degassing during 7600 14C Kurile Lake caldera-forming eruption and its climatic impact

The main goal of this investigation is estimating volume of volatile emission, atmospheric and climatic impact of the Kurile Lake caldera-forming eruption, one of the Earth’s largest Holocene explosive eruptions. The volatile content of magma before the eruption was estimated by comparing H₂O, S, Cl and F contents in natural quenched glassy melt inclusions trapped by plagioclase phenocrysts. The volatile content of igneous rocks after eruption was estimated by comparing concentrations of degassed matrix glasses. As a result of KO-eruption not more than (3.7–4.2) × 10¹² kg of water, (4.3–4.9) × 10¹⁰ kg of chlorine, (8.6–9.8) × 10⁹ kg of fluorine and (2.6–2.9) × 10¹⁰ kg of sulphur were injected into the atmosphere. This eruption had to produce an important climatic impact.     

The time scales of magma mixing and mingling involving primitive melts and melt–mush interaction at mid-ocean ridges

We have studied the chemical zoning of plagioclase phenocrysts from the slow-spreading Mid-Atlantic Ridge and the intermediate-spreading rate Costa Rica Rift to obtain the time scales of magmatic processes beneath these ridges. The anorthite content, Mg, and Sr in plagioclase phenocrysts from the Mid-Atlantic Ridge can be interpreted as recording initial crystallisation from a primitive magma (~11 wt% MgO) in an open system. This was followed by crystal accumulation in a mush zone and later entrainment of crystals into the erupted magma. The initial magma crystallised plagioclase more anorthitic than those in equilibrium with any erupted basalt. Evidence that the crystals accumulated in a mush zone comes from both: (1) plagioclase rims that were in equilibrium with a Sr-poor melt requiring extreme differentiation; and (2) different crystals found in the same thin section having different histories. Diffusion modelling shows that crystal residence times in the mush were <140 years, whereas the interval between mush disaggregation and eruption was ≤1.5 years. Zoning of anorthite content and Mg in plagioclase phenocrysts from the Costa Rica Rift show that they partially or completely equilibrated with a MgO-rich melt (>11 wt%). Partial equilibration in some crystals can be modelled as starting <1 year prior to eruption but for others longer times are required for complete equilibration. This variety of times is most readily explained if the mixing occurred in a mush zone. None of the plagioclase phenocrysts from the Costa Rica Rift that we studied have Mg contents in equilibrium with their host basalt even at their rims, requiring mixing into a much more evolved magma within days of eruption. In combination these observations suggest that at both intermediate- and slow-spreading ridges: (1) the chemical environment to which crystals are exposed changes on annual to decadal time scales; (2) plagioclase crystals record the existence of melts unlike those erupted; and (3) disaggregation of crystal mush zones appears to precede eruption, providing an efficient mechanism by which evolved interstitial melt can be mixed into erupted basalts.     

Melt inclusions and crystal-liquid separation in rhyolitic magma of the Bishop Tuff

Melt inclusions in quartz phenocrysts from a single clast of pumice near the base of the plinian pumice fall of the Bishop Tuff were studied to test ideas concerning separation of melt and crystals in silicic magmas. Ten analyzed inclusions from the pumice clast are of high silica rhyolite composition with very low contents of the highly compatible elements Ba, Sr, and Eu, consistent with extensive fractionation. The concentrations of U, La, Ce, Mg, and Ca of these ten melt inclusions vary considerably as determined by ion microprobe. Petrologic considerations indicate that uranium is an incompatible element with a maximum bulk partition coefficient D of about 0.2 and that the evolution of the uranium content of the melt was controlled by crystallization of the magma. A minimum of 33 wt% perfect fractional crystallization is required to explain the observed range in uranium. However, only 17 wt% crystals occurred in the pumice clast. The greater calculated fraction of crystals requires significant separation of crystals and melt before the eruption of the plinian pumice fall in spite of the fact that crystal mixing (settling, etc.) did not occur in the Bishop magma.     

Re-equilibration of melt inclusions trapped by magnesian olivine phenocrysts from subduction-related magmas: petrological implications

We describe and model a potential re-equilibration process that can affect compositions of melt inclusions in magnesian olivine phenocrysts. This process, referred to as “Fe-loss”, can operate during natural pre-eruptive cooling of host magma and results in lower FeO^t and higher MgO contents within the initially trapped volume of inclusion. The extent of Fe-loss is enhanced by large temperature intervals of magma cooling before eruption. The compositions of homogenised melt inclusions in olivine phenocrysts from several subduction-related suites demonstrate that (1) Fe-loss is a common process, (2) the maximum observed degree of re-equilibration varies between suites, and (3) within a single sample, variable degrees of re-equilibration can be recorded by melt inclusions trapped in olivine phenocrysts of identical composition. Our modelling also demonstrates that the re-equilibration process is fast going to completion, in the largest inclusions in the most magnesian phenocrysts it is completed within 2 years. The results we obtained indicate that the possibility of Fe-loss must be considered when estimating compositions of parental subduction-related magmas from naturally quenched glassy melt inclusions in magnesian olivine phenocrysts. Compositions calculated from glassy inclusions affected by Fe-loss will inherit not only erroneously low FeO^t contents, but also low MgO due to the inherited higher Mg##of the residual melt in re-equilibrated inclusions. We also demonstrate that due to the higher MgO contents of homogenised melt inclusions affected by Fe-loss, homogenisation temperatures achieved in heating experiments will be higher than original trapping temperatures. The extent of overheating will increase depending on the degree of re-equilibration, and can reach up to 50 °C in cases where complete re-equilibration occurs over a cooling interval of 200 °C. Received: 2 November 1998 / Accepted: 27 September 1999