Flood basalts erupted at three distinct times during formation of the Hartford basin. Such tripartite activity is believed to be inherent to most continental rift-associated magmatism. In the Hartford basin, lavas erupted from three northeast-trending dikes. First, the Talcott basalt erupted from the eastern Higganum dike; 138 ka later the Holyoke basalt erupted from the central Buttress dike; and 345 ka later still, the Hampden basalt erupted from the western Bridgeport dike. The time between eruptions is estimated from Milankovitch-type cycles in lake sediments between the flows. It is proposed that magmas were generated by decompression melting of the adiabatically rising mantle beneath the basin. All three basalts rose through a common conduit in the lower lithosphere, but the upper part of this dike was repeatedly beheaded and shifted eastward by crustal extension that occurred along an eastward-dipping detachment surface that passed beneath the basin. Extension rates determined from dike displacements range from 4 to 9 cm/yr, being greatest where the basin is widest.
The compositional changes and decreases in eruption temperature of successive basalts are thought to result from adiabatic partial melting of the source region as it continued to rise beneath the basin. The 18 °C decrease in eruption temperature (experimentally determined) between the Talcott and Holyoke basalts can be attributed to the Holyoke magma being formed by ~4% partial melting of the mantle. Similarly, the Hampden basalt would be formed by 2.2% partial melting to give a 9 °C decrease. If 1% melt must remain in the mantle, the fractions of partial melt that formed these two basalts are in the same proportions as the estimated erupted volumes of these basalts.
Each of the basalts has a calculated magma density that is significantly less than average crustal densities. Consequently, magma would not have ponded at depth but would have risen rapidly toward the surface. Because the magmas were dry, the partial melting that began in the mantle continued all the way to the surface. During ascent, therefore, the decreasing crystal content of the magma caused the viscosity to decrease exponentially and the bulk density to decrease linearly. Moreover, the volume expansion on melting added to the buoyant force causing the magma to rise. These factors combined to give high magmatic fluxes, which account for the enormous volumes of single eruptive units.
Because of the rapid ascent of the magmas, little intratelluric differentiation could occur. The major compositional differences between successive basalts must therefore reflect differences in the proportions of melt to solid in the magmas rising from the source or to phase changes resulting from decompression of the ascending source region. Resorption of phenocrysts and filter pressing during ascent modified magma compositions sufficiently to produce rocks that plot near multiple saturation boundaries. The magmas were also contaminated by low-melting fractions of crustal rocks during turbulent emplacement through wide dikes.