A major constraint is described for a possible chemical origin for the "mass-independent" oxygen isotope phenomenon in calcium-aluminum rich inclusions (CAIs) in meteorites at high temperatures (similar to1500-2000 K). A symmetry-based dynamical eta effect is postulated for O atom-monoxide recombination on the surface of growing CAIs. It is the surface analog of the volume-based eta effect occurring in a similar phenomenon for ozone in the gas phase [Y. Q. Gao, W. C. Chen, and R. A. Marcus, J. Chem. Phys. 117, 1536 (2002), and references cited therein]: In the growth of CAI grains an equilibrium is postulated between adsorbed species XO (ads)+O (ads)--> <--XO2* (ads), where XO2* (ads) is a vibrationally excited adsorbed dioxide molecule and X can be Si, Al, Ti, or other metals and can be C for minerals less refractory than the CAIs. The surface of a growing grain has an entropic effect of many order of magnitude on the position of this monoxide-dioxide equilibrium relative to its volume-based position by acting as a concentrator. The volume-based eta effect for ozone in the earlier study is not applicable to gas phase precursors of CAIs, due to the rarity of three-body recombination collisions at very low pressures and because of the high H-2 and H concentration in solar gas, which reduces gaseous O and gaseous dioxides and prevents the latter from acting as storage reservoirs for the two heavier oxygen isotopes. A surface eta effect yields XO2* (ads) that is mass-independently rich in O-17 and O-18, and yields XO (ads)+O (ads) that is mass-independently poor in the two heavier oxygen isotopes. When the XO2* (ads) is deactivated by vibrational energy loss to the grain, it has only one subsequent fate, evaporation, and so undergoes no further isotopic fractionation. After evaporation the XO2 again has only one fate, which is to react rapidly with H and ultimately form O-16-poor H2O. The other species, O (ads)+XO (ads), are O-16 rich and react with Ca (ads) and other adsorbed metal atoms or metallic monoxides to form CAIs. The latter are thereby mass-independently poor in O-17 and O-18. Some O (ads) used to form the minerals are necessarily in excess of the XO (ads), because of the stoichiometry of the mineral, and modify the fractionation pattern. This effect is incorporated into the mechanistic and mathematical scheme. A merit of this chemical mechanism for the oxygen isotope anomaly is that only one oxygen reservoir is required in the solar nebula. It also does not require a sequestering of intermediate products which could undergo isotopic exchange, hence undoing the original isotopic fractionations. The gas phase source of adsorbed O atoms in this environment is either O or H2O. As inferred from data on the evaporation of Mg2SiO4 taken as an example, the source of O (ads) is primarily H2O rather than O and is accompanied by the evolution of H-2. Nonisotopic kinetic experiments can determine more sharply the mechanism of condensed phase growth of these minerals. Laboratory tests are proposed to test the existence of a surface eta effect on the growing CAI surfaces at these high temperatures. (C) 2004 American Institute of Physics.