I already sent Warren a paper that Melosh sent me that addresses this issue to some extent. This is the basis of Melosh's conclusion that a significant % of such ejecta may eventually make it into interstellar space. There has been a huge amount of work on the early orbital evolution of the solar system. These simulations seem to indicate that the evolution of many of the planet types -- including a "Jupiter", an asteroid belt, Earth-size rocky planets like Venus & Earth, isn't such a rare thing. You are right; close encounters with Jupiter & the Sun send a lot of early planet wannabes packing into interstellar space. Some end up in extremely elliptical orbits, with periods in the hundreds and possibly thousands of years. There are now many, many simulations of the first 150 million years of our solar system because -- among other reasons -- it apparently took that long for the Earth's Moon to get into position and for most things to settle down into more-or-less the current configuration. Michele Trenti has done a lot of work on gravitational simulations: http://casa.colorado.edu/~trenti/home/highz.html http://www.ast.cam.ac.uk/~trenti/ At 08:50 AM 4/11/2012, Warren Smith wrote:

To eject stuff from not just earth, but in fact whole solar system, you'd probably need help from "slingshot" effects around planets like Jupiter. Is it really true system ejection is common? How do you know it is, if it is?

There are observation-based claims that there are a ton of "rogue planets" (more than the number of stars) wandering round the galaxy, which were ejected from their systems by such effects. So it seems plausible 10% of stuff inserted into the solar system in a "random orbit" is ultimately ejected. But trying to justify such an estimate mathematically or computationally might be very difficult, since the ejection process could take 100M years.