A pressure-driven three-dimensional turbulent boundary layer flow upstream and around a junction vortex was experimentally studied and is offered for use as a benchmark flow for testing and evaluating the predictive ability of state-of-the-art three-dimensional turbulent boundary layer codes. The pressure-driven flow and junction vortex system was generated by a streamlined cylinder placed normal to a flat surface. Measurements of wall static pressure, wall shear stress, mean velocity, and Reynolds stress tensor field are reported at several stations in the three-dimensional turbulent boundary layer region. Documentation of the flow edge conditions is provided as well as upstream initial conditions along a plane with measured mean velocity and Reynolds stress tensor to permit the testing of intermediate and higher order turbulence models. Measurements of wall shear stress magnitude were made with a Preston tube and the wall shear stress directions were taken from an oil streak flow visualization. These results are compared with earlier direct force wall shear measurements of both magnitude and direction. Mean velocity magnitude and direction were measured with a single hot film probe. Measurements of the complete Reynolds stress tensor were carried out with three hot film x-array probes. Supporting work includes a wind tunnel calibration which examined the sensitivity and effects of spanwise nonuniformities and a two-dimensional momentum integral calculation along the tunnel center plane; the development of a calibration technique to determine individual sensor yaw characteristics in more complex probe geometries; and a generalized response analysis for a sensor with arbitrary orientation to the flow which allows for the use of an arbitrary yaw cooling law, allows for modest amounts of probe misalignment and yields a precise definition of matched sensors, geometric guidelines for constructing x-array probes, and a general mean velocity correction for turbulence where several existing formulas are compared. In addition, two popular cooling laws are studied, comparisons are made with other response equations, and an extensive discussion of the errors associated with the matched sensor approximations is given. Comparisons are made of several mean velocity measurements using different probes and redundant normal and shear stresses measured by the different x-array film probes, a single wire, and single film probe are compared. / Ph. D.