ABSTRACT
For example, stellar magnetic elds may be deduced from the polarisation of the wings of spectrum lines (see Section 5.2). e noise (errors, uncertainties) in the observations, however, will mean that a range of eld strengths and orientations will t the data equally well. e three-dimensional distribution of stars in a globular cluster must be found from observations of its two-dimensional projection onto the plane of the sky. Errors in the measurements will lead to a variety of distributions being equally good ts to the data. Similarly, the images from radio and other interferometers (see Section 2.5) contain spurious features due to the side lobes of the beams. ese spurious features may be removed from the image if the eects of the side lobes are known. But since there will be uncertainty in both the data and the measurements of the side lobes, there can remain the possibility that features in the nal image are artefacts of the noise, or are incompletely removed side lobe eects.
