The Laki (Skaftar Fires) fissure eruption in southern Iceland lasted for eight months during 1783 to 1784, and produced one of the largest basaltic lava flows in historic times (14.7 +/- 1.0 km3). In addition, neighboring Grimsvotn central volcano was frequently active during the period from May 1783 to May 1785. The combined activity is interpreted as having been the result of a two-year-long volcano-tectonic episode on the Grimsvotn volcanic system. Contemporary descriptions of the explosive activity make it possible to relate the tephra stratigraphy to the progress of the eruption on a weekly basis and show that activity on the fissures propagated to the NE with time, towards Grimsvotn. The eruption at Laki began on 8 June with a brief explosive event on a short fissure, and lava rapidly began to flow into the Skafta river gorge. It reached the lowlands, 35 km away, four days later and continued to flow, with variable discharge, until 7 February 1784. Approximately 90% of the lava was emplaced in the first five months of activity. The 27-km-long vent complex is composed of ten en echelon fissures distributed on both sides of the much older Laki hyaloclastite mountain. The surface expression of each fissure is a continuous row of vents consisting of scoria cones, spatter cones, and tuff cones. Six tephra fall units are positively identified; two units are completely composed of phreatomagmatic tephra derived from two tuff cones and the others are Strombolian deposits. The volume of tephra, including ash fall that extended to mainland Europe, is 0.4 km3 dense rock equivalent volume, or 2.6% of the total erupted volume. Interpretation of contemporary descriptions of tephra falls, combined with the preserved stratigraphy, allow the identification of ten eruptive episodes during the eight months of activity on the Laki fissures. These eruptive episodes are inferred to have resulted from the unsteady flow of magma in the feeder system. In addition, at least eight eruption episodes occurred at Grimsvotn in 1783 to 1785, five in 1783, two in 1784, and one in 1785. Each episode at Laki began with a seismic swarm of increasing intensity that led to the formation of a new fissure, the opening of which was followed by short-lived phreatomagmatic activity caused by the high water table around the eruption site. Activity usually changed to violent Strombolian or sub-Plinian, followed by Hawaiian fire fountaining and effusive activity as the availability of groundwater dwindled. Thus, the explosive activity associated with the opening of each fissure was largely controlled by external water-magma interactions. Maximum effusion rates, occurring in the first two episodes, are estimated to have been 8.5 x 10(3) and 8.7 x 10(3) m3 s-1 from fissures totaling 2.2 and 2.8 km in length, respectively, and, in general, discharge gradually decreased over time. The highest rates are equivalent to 5.6 x 10(3) and 4.5 x 10(3) kg s-1 per meter length of fissure, values that could conceivably be similar to those that produced some flood basalt lava flows. Maximum fire fountain heights are estimated to have varied from 800 m to 1400 m and convecting eruption columns above the vents rose to a maximum altitude of about 15 km. The release of sulfur gases during fountaining produced an acid haze (aerosol) which spread widely and had a considerable environmental, and possibly climatic, impact on the Northern Hemisphere.
