On Tue, 22 Sep 2009 01:11:10 -0700 (PDT), Dawlish
wrote:

I doubt whether it is more
than 1SD away from the mean and it is certainly not 2SD, which would
include 95% of the population (in this case all the measurements of
Antarctic Sea ice area from the graph) and would be regarded as
statistically significant.

The only slight caveat to that is the usual one of knowing/estimating
what the underlying distribution of the population might be. Perhaps
it might be at least tolerably close to a normal distribution, but
maybe someone has done some checks.

Personally, I was always somehow more comfortable with non-parametric
stats for this sort of study - again I imagine that someone might have
used non-parametric methods to look at eg ice data - but I really
can't remember too much about how to apply such methods.

JGD