Ludde asked earlier how to disprove the Molecular Clock hypothesis. There has
been a lot of work on looking at the details of the clock (or clocks) in
mammals. One basic method depends on having a convenient outgroup to two
taxa. This is Wilson's relative rate test. Basically, you ask whether
the numbers of substitutions along the two branches leading from the split
between the two taxa are the same. "Same", here, means not significantly
different. You estimate the two branch lengths (a simple piece of arithmetic
if you have the 3 distances between the 3 taxa) and their variances.
When this is applied to mammals it appears that humans are generally slower
than the other great apes which are in turn slower than primates in general
and rodents (rats and mice anyway) look faster than average. Much of this
has been done by WenHsiung Li in Houston. The ironic thing now is that
these rate differences (if they are real) are taken as the exceptions that
prove the rule. The clock is such a useful and widely validated model
that it is simply accepted as a first approximation until you can disprove
it is particular cases.
Another obvious example of how to see rate differences is to make trees of
ribosomal RNA sequences using both mitochondrial and nuclear genes.
No matter how you root the tree, the effect is startling; mitochondria
"go" much faster than equivalent nuclear genes. You can see the difference
just by looking at the tree: you get a stunted little bush of the nuclear
sequences and a big shrub of the mitochondrial ones.
One amazing piece of work on looking in great detail at clocks (among many other
things) comes from work by Walter Fitch (Irvine, CA) and co-workers
who looked at the evolution of influenza viruses. In this case, they
used frozen isolates dating back to early this century. The "fossil" record
is perfect (dates are very accurate) and believe it or not the molecular
clock is almost too good to be true. You get a beautiful straight line
when you plot numbers of substitutions versus time (for some proteins