Experiments were conducted under canonical nebular conditions to see whether the chemical compositions of the various chondrule types can be derived from a single CI-like starting material by open-system melting and evaporation. Experimental charges, produced at 1580 degreesC and PH2 of 1.31 x 10(-5) atm over I to 18 hours, consisted of only two phases, porphyritic olivine crystals in glass. Sulfur, metallic-iron and alkalis were completely evaporated in the first minutes of the experiments and subsequently the main evaporating liquid oxides were FeO and SiO2. Olivines from short runs (2-4 hours) have compositions of Fo(83)-Fo(89), as in Type IIA chondrules, while longer experimental runs (12-18 hours) produce similar toFo(99) olivine, similar to Type IA chondrules. The concentration of CaO in both olivine (up to 0.6 wt.%) and glass, and their Mg#, increased with increasing heating duration. Natural chondrules also show increasing CaO with decreasing S, alkalis, FeO and SiO2. The similarities in bulk chemistry, mineralogy and textures between Type IIA and IA chondrules and the experimental charges demonstrate that these chondrules could have formed by the evaporation of CI precursors. The formation of silica-rich chondrules (IIB and IB) by evaporation requires a more pyroxene-rich precursor. Based on the FeO evaporation rates measured here, Type IIA and IA chondrules, were heated for at least similar to0.5 and similar to3.5 h, respectively, if formed at 1580 degreesC and P-H2 of 1.31 x 10(-5) atm. Type II chondrules may have experienced higher cooling-rates and less evaporation than Type I. The experimental charges experienced free evaporation and exhibited heavy isotopic enrichments in silicon, as well as zero concentrations of S, Na and K, which are not observed in natural chondrules. However, experiments on potassium-rich melts at the same pressure but in closed capsules showed less evaporation of K, and less K isotopic mass fractionation, than expected as a function of decreasing cooling rate. Thus the environment in which chondrules formed is as important as the kinetic processes they experienced. If chondrule formation occurred under conditions in which evaporated gases remained in the vicinity of the residual melts, the extent of evaporation would be reduced and back reaction between the gas and the melt could contribute to the suppression of isotopic mass fractionation. Hence chondrule formation could have involved evaporative loss without Rayleigh fractionation. Volatile-rich Type II and volatile-poor Type I chondrules may have formed in domains with high and low chondrule concentrations, and high partial pressures of lithophile elements, respectively. Copyright (C) 2004 Elsevier Ltd.