Hundreds of criticality experiments were performed at Oak Ridge National Laboratory in the 1950s. Several sets of these experiments were used to determine the critical properties of 233U and 235U. Here, four sets are analyzed to provide benchmark descriptions for validation of computational tools used by nuclear criticality specialists. All four sets were performed in water-reflected spherical geometry and contained a highly enriched uranyl fluoride solution (93.18% 235U) with the hydrogen-to-fissile ratio of measurements ranging from 35.8 to 1272. The scope of these experiments spans the minimum values of the subcritical mass limit curve. One experiment was never reported in the open literature, and three experiments were performed at elevated temperatures. An uncertainty in the experimental keff was found by sensitivity studies on reported measurement uncertainties, inconsistencies, and omissions in experimental parameters. To be useful for all computer codes, one-dimensional benchmark configurations were determined for all sets of experiments. The descriptions can be used to find bias values for a code/cross-section package. The keff values for similar configurations can then be corrected using the bias values. The sensitivity analysis of the experiments was performed using ONEDANT with 27-group ENDF/B-IV cross sections and MCNP with continuous-energy ENDF/B-V cross-section data. The keff values for both one- and three-dimensional configurations were found using MCNP with ENDF/B-V and ENDF/B-VI cross-section data. The values for keff for the one-dimensional configuration were also found by using ONEDANT and KENO V.a with Hansen-Roach and 27-group ENDF/B-IV cross sections.