A determination of the resonance escape probability of thorium as thorium nitrate in aqueous solution has been made as a function of thorium concentration. The physical system used was an aluminum box surrounded by successive layers of cadmium, paraffin and borated paraffin to keep out neutrons scattered by objects in the laboratory. Neutrons were obtained from a Cockcroft-Walton type accelerator by the D(d,n)He³ reaction. The drive-in target was located at the center of one of the faces of the aluminum box. Neutron density was measured at nine spatial positions in the direction of the neutron beam with a bare boron trifluoride detector. The area under a curve of neutron density versus spatial position was obtained for various concentrations of absorber. The above process was carried out for the thorium solution and for a “mock solution," whose cross-section was similar to that of thorium except that it had no resonances in the thorium resonance region. By taking ratios of the neutron densities (area under curves of neutron density versus spatial position) in the thorium solution to the neutron density in the mock solution, it was possible to determine the resonance escape probability of neutrons in a homogeneous, aqueous solution of thorium nitrate. It is shown that, for the absorber concentrations used in the experiment, the resonance escape probability for an infinite geometry may be obtained by the above ratio method. The difference between a finite system and an infinite one is exhibited as leakage of neutrons from the system in the finite case. If one can compare neutron densities for systems which are large enough so that leakage is negligible or for systems with corresponding leakage rates, the effect of leakage can be overcome and the resonance escape probability for the infinite geometry obtained. Before taking the above ratios of neutron densities, it was necessary to compensate for the spectral shift of the thermal flux in the two solutions. After such a correction, the resonance escape probability so obtained shows good correlation with the results of the Monte Carlo prediction for this system. Over the range covered by the experiment (0 - 1 x 10²¹ atoms of thorium per cubic centimeter), experimental results agree with Monte Carlo predictions to within one percent. Counting statistics were good, with 10⁶ counts normally taken per spatial position. The curves from which the value of neutron density were determined were formed by nine points, each of which represented at least 10⁶ counts. Reproducibility of the neutron density at a point was of the order of one percent. Various changes made in the analysis of the data have caused corresponding changes in the values obtained for the resonance escape probability of less than one percent. These facts all indicate that the uncertainty in the experimental determination is of the order of one percent. Calculations of effective resonance integrals from the experimentally determined values of the resonance escape probability show good agreement with published measurements on other systems. / Ph. D.