There are some problems with this explanation, though. The most obvious is that such a large collision would have also imparted considerable linear momentum on Uranus, which would have caused its eccentricity to be far greater than what's observed today. It is possible that the eccentricity would have decreased to today's levels over the past few billion years since the great collision, but the currently understood mechanisms would not account for a large enough decrease.

The idea Professor Stevenson and I developed held that near the origin of the solar system, Uranus was much hotter and had a somwhat larger angular momentum about its axis, so that it actually looked more like a disk than a sphere. With such a planet, it turns out that a mars-sized planetoid could have come close enough to Uranus to plough through a lot of gas near the outer edge of the flattened Uranus. The drag from this first encounter could have slowed the planetoid enough to make it gravitationally bound in Uranus' field. Thus, after this first encounter, the planetoid would have gradually spiraled into Uranus and been absorbed.

The advantage of this explanation as opposed to the great collision picture is that the planetoid would have had to be only mars-sized, instead of earth-sized. Since all the planetoids had similar speeds relative to Uranus, the mass difference would have resulted in a smaller linear momentum imparted on Uranus, but the same angular momentum since in this picture, the impact parameter is much larger than in the great-collision scenario. The smaller linear momentum addition during the collision would have left Uranus with a considerably smaller eccentricity (though still larger than observed today). I should mention that the likelyhood calculations for this scenario rely heavily on the mass-distribution of the early solar system, which is still on rather shaky ground.