Due to inaccuracies of our
knowledge of the geometrical positions of the equipment in the hall and of the
magnitude and configuration of the magnetic field, the acceptance correction is
imperfect.
It is less reliable on the edges.
Typically, when looking at either
or
spectra from a single setting with sufficiently fine
binning one sees points on the edges of acceptance that deviate sharply from the
overall pattern. These points must not be included in the fiducial cut area.
A sharp non-constancy of
, inconsistent between different settings, would indicate that a
wrong
was used.
Acceptance correction to the or
spectra includes
information about shape of the
distribution. Therefore any
uncertainty in the
shape results in an uncertainty of the slope.
In quantifying the uncertainty, the first step was to derive the error
propagation factor to convert the uncertainty of the
width
into uncertainty of the inverse slope
. I describe the
with a Gaussian whose width I denote by
.
MC was run with two widths of input
:
and
.
The slopes were extracted in the two cases and compared,
the error-propagation factor was found to be