Since 
is 
distributed about its minimum with
respect to A (at a fixed point), the probability of 
can be
evaluated by performing a one-tailed 
test on the statistic.
Given the best-fit value 
, the detection significance is evaluated
by noting that that A = 0 falls on the probability P confidence contour,
such that
The significance of the detection is then the deviate 
of a
one-tailed normal distribution (one-tailed as negative A
are regarded as un-physical) such that
For small values of 
, P can be readily found using a NAG routine
and converted into the a normal deviate. However, for very small P (or
equivalently, large 
) the NAG routine has overflow problems. The
following asymptotic expansion is useful,
