The most direct means for calculating the Earth's age is a Pb/Pb isochron age, derived from samples of the Earth and meteorites.
The higher the uranium-to-lead ratio of a rock, the more the Pb-206/Pb-204 and Pb-207/Pb-204 values will change with time.
If the source of the solar system was also uniformly distributed with respect to uranium isotope ratios, then the data points will always fall on a single line.
For example: Also note that the meteorite ages (both when dated mainly by Rb-Sr dating in groups, and by multiple means individually) are in exact agreement with the solar system "model lead age" produced earlier.
Young-Earthers have several methods which they claim to give "upper limits" to the age of the Earth, much lower than the age calculated above (usually in the thousands of years).
Further, the processes of erosion and crustal recycling have apparently destroyed all of the earliest surface.
The oldest rocks which have been found so far (on the Earth) date to about 3.8 to 3.9 billion years ago (by several radiometric dating methods).
While these values do not compute an age for the Earth, they do establish a lower limit (the Earth must be at least as old as any formation on it).
This lower limit is at least concordant with the independently derived figure of 4.55 billion years for the Earth's actual age.
And from the slope of the line we can compute the amount of time which has passed since the pool of matter became separated into individual objects.
See the Isochron Dating FAQ or Faure (1986, chapter 18) for technical detail.
(I believe this argument was originally put forth by Mormon young-Earther Melvin Cook, in a letter to the editor which was published in .) But helium can and does escape from the atmosphere, at rates calculated to be nearly identical to rates of production.