Despite many decades of study, models of discharges in molecular gases still lack accurate data on many key collisional processes, both in the gas phase and on surfaces, necessary for reliable predictive modeling. Good data is lacking for electron-impact dissociation, surface recombination, reactions with metastables, gas heating mechanisms, energy transfer, and surface thermal accommodation. This is true even for such “simple” and ubiquitous gases as O2 and N2. Diagnostic techniques have advanced significantly, with an emphasis on accurate absolute density measurements, space and time resolution, and energy distribution functions (translational, rotational and vibrational). However, most studies focus on measuring only a few of the modeled parameters for a given system, and often in plasma configurations optimized for applications rather than for model/experiment comparison. Particularly problematic is the presence of large gradients (in temperature, density, composition) as well as poorly-controlled surface conditions. Our approach therefore is to use a well-characterized, stable and uniform discharge chosen to be simple to model, namely a DC positive column in pure O2, and to perform a comprehensive set of measurements of all accessible parameters using multiple (and sometimes overlapping) diagnostics. The model/experiment comparison is therefore highly constrained, allowing the origin of discrepancies to be probed in unprecedented detail. Measurements were made of the absolute densities and kinetics of O2 in the X, a and b states, and O 3P atoms, using a combination of synchrotron vacuum ultraviolet absorption, cavity ringdown spectroscopy and optical emission (visible and infrared). These results show the importance of gas heating and surface reactions, as well as charged and neutral gas phase reactive collisions.