Subject: How do you measure the inaccuracy of the most accurate clock?

The new cesium fountain clock is supposed to be the most accurate time 
keeping device we have--accurate to one sec. in 20 million years. But if 
we can measure how inaccurate it is, doesn't that imply that somewhere 
there is a device or standard that is even MORE accurate (or even 
perfectly accurate?) against which we measure the inaccuaracy of the 
cesium clock? It almost sounds like a paradox to me--how do they know?

Re: How do you measure the inaccuracy of the most accurate clock?