Do nebular fractionations, evaporative losses, or both, influence chondrule compositions?

We have made observations and performed heating experiments to determine the relative importance of several processes which may have influenced the compositions of chondrules. As heating destroys nuclei, the number density of olivine and pyroxene crystals gives an indication of the extent of melting. We determined number densities of Semarkona type I chondrules and converted them to nominal grain size, for use as a measure of intensity of heating. Bulk compositions of the chondrules show correlations with nominal grain size. Na, K, Fe, Ni, P and S decrease as grain size (degree of melting) increases, and we interpret this as evidence of evaporative loss. The evidence is less clear for Mn, Cr and Si. SiO_2/MgO ratios show very large variations even in fine-grained type I chondrules containing FeS, and we interpret those variations as due to nebular fractionations affecting precursors. Experiments show that Na and S losses increase with higher temperatures and lower cooling rates. It is hard to preserve any sulfide at all, without flash heating. Na, however, can be retained at close to chondritic levels (as in type II chondrules) with flash heating and high cooling rate, provided also that the oxygen fugacity is high. Type II chondrules can retain much more Na than type I under identical thermal conditions, because of higher fO_2 (either due to non-nebular gas or possibly internal buffering by FeO content) and melt structure (higher SiO_2/MgO). Gas reduction experiments show that type II compositions can be converted to IB by Fe loss, but evaporative loss of SiO_2 (so as to approach IA composition) is not achieved without prolonged isothermal heating. Precursors of type I and II chondrules were probably close to chondritic in composition, but with higher Fa in the type II case. They consisted of olivine, pyroxene, plagioclase, Fe (Ni) S and carbon compounds, probably with insignificant metal. Sulfur loss generated much chondrule metal in ordinary chondrites. C is a possible alternative to gas reduction to explain dusty relict grains and the lower olivine Fa in the more melted type I chondrules. We agree with J. N. GROSSMAN and J. T. WASSON (Geochim. Cosmochim. Acta, 47,759,1983) that variations in Mg/Si are due to nebular fractionations and with S. HUANG et al. (Icarus, 122,316,1996) that variations in Na and Fe in type I chondrules are mainly due to evaporative losses