The diffuse ionized gas (DIG), also referred to as the Reynolds layer, is a nearly fully ionized gas with a mean midplane density of and a scale height of . This layer has an average filling factor of with the possibility that the filling factor increases from in the galactic plane to at (Kulkarni & Heiles 1987; Reynolds 1991, 1993). The average mean density of a clump in this layer is while the mean density of a clump in the Galactic plane is .
Table 1 summarizes the observed properties in the DIG. The majority of these measurements have been made near the Galactic midplane within of the Sun. Observed limits of emission from the DIG indicate that the hydrogen is at least ionized, and that ionization is probable (Reynolds 1989). The energy required to keep the DIG ionized is , corresponding to hydrogen ionizing photons per per sec in the Galactic disk ([Reynolds 1992]). This corresponds to of the Lyman continuum photons of Galactic OB stars or of the kinetic energy injected by Galactic supernovae into the interstellar medium (ISM) ([Reynolds 1993]). The electron temperature of the DIG has been constrained to be in the range ([Reynolds 1985a]), with an average of . Reynolds & Tufte (1995) found an upper limit of which indicates that the ionizing spectrum incident on the DIG is relatively soft and corresponds to a helium-to-hydrogen ionization fraction of . This is consistent with radio recombination line measurements of H and He toward the inner Galaxy which produce (Heiles et al.\ 1996). The observed ranges of and require a much harder ionizing spectrum if photoionization is to explain these line ratios ([Reynolds 1985a]).
It appears that the physical conditions in the DIG cannot be explained by simple photoionization models and that other physical processes may be important in the DIG. It has been determined that the DIG in NGC 891 also cannot be explained by simple photoionization models ([Rand 1997]). Previous models developed by Mathis (1986) and Domgörgen & Mathis (1994) have attempted to explain the conditions in the DIG but have not been entirely successful (Reynolds & Tufte 1995; [Rand 1997]). Raymond (1976) and Shull & McKee (1979) have developed models in which shocks in the DIG predict the observed line ratios. The shock models, however, produce results that are inconsistent with the observed linewidths ([Reynolds 1985a], 1985b).