I've just finished Stephen Baxter'sXeelee: Redemption, which features more of Baxter's vast mega-structures and time-periods. In this one, he uses a ringworld a light-year in radius, spinning at just a whisker short of light-speed, so that its inhabitants experience 1gof centripetal acceleration, while undergoing massive time dilation (a gamma factor of 5 million). It's a simple application of the a=v²/r equation for centripetal acceleration, and it comes out the neat way it does because of the odd coincidence that lightspeed divided by the length of a year is very close to being one gravity of acceleration.

Trouble is, Baxter neglected the effect of relativity on acceleration. We're familiar with the fact that, as you accelerate towards lightspeed, 1gof acceleration in your moving frame translates into considerably less in the rest frame - there's a factor of gamma-cubed involved for linear motion. A similar thing happens for transverse acceleration, of the kind required to steer your starship in a circle - a marginally more benign gamma-squared conversion.

So Baxter's ring has a centripetal acceleration of 1gin the rest frame, but its inhabitants would be pulling25 trilliongin the frame corotating with the ringworld. Oops. We're going to need a bigger ring.

We need a radius of gyration that produces just one 25-trillionth of agat lightspeed, in the rest frame. That turns out to be 2.4e13 light-years - inconveniently large, even by Baxter's standards.

Shame. It was such a nice idea.

Grant Hutchison