A combination of experiment [optical emission and cavity ring-down spectroscopy (CRDS) of electronically excited H atoms] and two-dimensional (2D) modeling has enabled a uniquely detailed characterization of the key properties of the Ar/H2 plasma within a ⩽ 10-kW, twin-nozzle dc arc jet reactor. The modeling provides a detailed description of the initial conditions in the primary torch head and of the subsequent expansion of the plasma into the lower pressure reactor chamber, where it forms a cylindrical plume of activated gas comprising mainly of Ar, Ar+, H, ArH+, and free electrons. Subsequent reactions lead to the formation of H2 and electronically excited atoms, including H(n = 2) and H(n = 3) that radiate photons, giving the plume its characteristic intense emission. The modeling successfully reproduces the measured spatial distributions of H(n>1) atoms, and their variation with H2 flow rate, FH20. Computed H(n = 2) number densities show near-quantitative agreement with CRDS measurements of H(n = 2) absorption via the Balmer-β transition, successfully capturing the observed decrease in H(n = 2) density with increased FH20. Stark broadening of the Balmer-β transition depends upon the local electron density in close proximity to the H(n = 2) atoms. The modeling reveals that, at low FH20, the maxima in the electron and H(n = 2) atom distributions occur in different spatial regions of the plume; direct analysis of the Stark broadening of the Balmer-β line would thus lead to an underestimate of the peak electron density. The present study highlights the necessity of careful intercomparisons between quantitative experimental data and model predictions in the development of a numerical treatment of the arc jet plasma. The kinetic scheme used here succeeds in describing many disparate observations—e.g., electron and H(n = 2) number densities, spatial distributions of optical emission from the plume, the variation of these quantities with added flow of H2 and, when CH4 is added, absolute number densities and temperatures of radicals such as C2 and CH. The remaining limitations of the model are discussed.