However, as pointed out in the Perspective by Daniel Kleppner in the same issue, with an error budget that includes perturbation of the order of 10^-18 s, such precision now has to consider the effects of General Relativity.
When precision is pushed to new levels, ever more subtle effects must be taken into account. For instance, the error budget includes a small contribution, 1 mult 10-18, due to an uncertainty in the gravitational potential of the two clocks. This corresponds to a difference in their altitudes of 1 cm. This heralds one of the most interesting aspects of time keeping with optical clocks: The effects of general relativity that mix time with gravity are starting to approach a point that will require rethinking the basic concept of "keeping time."
The "two clocks" referred here are the two ion clocks used in the T. Rosenband et al. experiments - Al+ ion and Hg+ ion clocks. Having clocks that are now sensitive enough to detect effects of GR would certainly open up a whole possibility of testing GR even more.
 T. Rosenband et al., Science v.319, p.1808 (2008).
 A. D. Ludlow et al., Science v.319, p.1805 (2008).