Abstract
A method is developed for the extraction, from SQUID magnetometer data, of a value for the total magnetic moment of an extended sample. Background drift associated with the superconducting magnet is usually a serious problem in these experiments and the authors show, using computer simulations, that their method deals successfully with this. Reasonably good estimates are required of the length of the sample and of the number of readings over which the background drift is correlated. The method uses a matched filter to carry out deconvolution in the presence of correlated noise. The approach is illustrated for a wire sample of Ni-Cr alloy in extended and compressed forms. Results are analysed for pure water and oxidized human blood, giving volume susceptibility values of (-9.060+or-0.003)*10-6 and (-7.390+or-0.003)*10-6 respectively.