Abstract
This paper describes the underlying principles of a computer code to automatically analyse a gamma-ray spectrum obtained with a Ge(Li) detector. The significant peaks present in the spectrum are detected by observing changes in the first derivative which is measured after random variations in count-rate have been minimised by the application of a smoothing function. Pseudo-peaks due to backscattering events or Compton edge effects are eliminated, and the intensities and energies of the component gamma-rays are calculated by fitting a Gaussian function to each peak in turn. The analysis of overlapping peaks is considered and conditions chosen to optimise their detection. It is shown that if a catalogue of gamma-ray energies is available to the computer then it is possible to automatically identify the component isotopes and evaluate their activities.