Toward the end of the Gulf War, Iraqi troops damaged several hundred oil wells in Kuwait setting many of them on fire. Measurements made in March 1991, a few weeks after most of the fires had started (Johnson et al., 1991), were used to estimate the total burn rate and the emission rates of individual pollutants. Measurements of the principal carbon species in the plume, obtained from flask samples collected at the same time as continuous measurements of SO2 have been used to derive an ''effective'' sulphur content of the smoke of 2.4%, almost a third lower than the previous estimate. This sulphur content of 2.4% combined with the capping history of the fires has been used to revise the earlier estimates and provide more detailed information on the speciation of the emissions. It is now estimated that 139¿106 t of crude oil were burnt during an 8-month period, resulting in the release of 112¿106 t of carbon in carbon dioxide, 3¿106 t of carbon in soot, 1.6¿106 t of carbon in carbon monoxide, 1.3¿106 t of carbon in nonmethane hydrocarbons, 0.11¿106 t of nitrogen in nitrogen oxides, and 3.11¿106 t of sulphur in sulphur dioxide. In addition to measurements made close to the source of the plume, one flight successfully sampled a plume some 600 km from the fires which had experienced significant photochemical aging. These observations provided a unique data set with which to estimate the rate at which hydrocarbon pollutants in the plume degrade and to infer the hydroxyl radical concentrations which cause that degradation.

Most of the aliphatic hydrocarbon concentrations determined from flask samples collected at a range of distances from the Kuwait source conform to a simple loss process proportional to hydrocarbon hydroxyl reactivity and imply a diurnally averaged hydroxyl radical concentration within the plume of 1¿106 molecules cm-3. Finally, it is shown that, although theoretically, hydrocarbon concentrations can be combined to predict the difference ratio of hydrocarbon reactivity, that is, (k1-k2)/(k3-k4), in practice experimental error can give rise to apparent discrepancies between theory and observations. Earlier studies found similar discrepancies and have speculated that these arise from differences in ''photochemical age'' for different hydrocarbons, a parameter those studies could not determine. We have shown similar discrepancies but have been able to determine photochemical ages for several hydrocarbons and have shown that the hydrocarbon reductions are consistent with a single photochemical age (3% precision), precluding the earlier explanations. However, even though the photochemical lifetime may be determined very precisely, consideration of the computational stability of the procedures involved suggest that unavoidable experimental precision may lead to factors of 2 errors in the determination of the difference ratio, which is more than sufficient to explain the known discrepancies. ¿ American Geophysical Union 1995