Shields & Searle (1978) pointed out that sulfur is an important secondary coolant in metal-rich HII regions, and that measurements of [S III] could help constrain photoionization models for such nebulae. Although it is well established that S/O in HII regions is generally near the solar system value, log S/O -1.7 (Anders & Grevesse 1989), the question of systematic variations with O/H remains unclear and controversial from several recent studies. Shaver et al. (1983), Vílchez et al. (1988) and Díaz et al. (1991) derived an apparent decline in S/O by approximately 0.5 dex from log O/H = -3.7 to log O/H = -2.6 in the Galaxy, M33, and M51. On the other hand, Garnett (1989) and Torres-Peimbert et al. (1989) discern no systematic variation in S/O from the solar system value over the range -4.8 < log O/H < -3.3 in dwarf irregular galaxies and M101. Meanwhile, measurements of IR [S III] lines (Pipher et al. 1984, Simpson et al. 1995) indicate that the sulfur abundance gradient in the Milky Way is much steeper than the gradient derived by Shaver et al. (1983), and is essentially identical to the oxygen gradient. Note that Garnett (1989) and Vílchez et al. (1988) disagree by a factor of two in the S abundance for the region NGC 588 in M33, a critical low-abundance point in the spiral galaxy sample. A new observation of this HII region would be very valuable.
Our new [S III] measurements for the NGC 2403 sample provide a new homogeneous set of data to address this question. Our derived S/O abundance ratios are displayed in Figure 20; the top panel shows the S/O with full ionization corrections for S (Section 3.3); the lower panel shows S+ + S+2 alone. The ionization corrections we apply for S are minor, but they do introduce a small systematic effect. Figure 20 shows that there is little variation in S/O across the disk of NGC 2403. Formally, the observed gradient in S/O is barely significant, and the total range in S/O in NGC 2403 is only about 0.2 dex between 1 and 6 kpc, essentially the same as the observational scatter.
This would seem to indicate a universal value for S/O consistent with the solar value. However, much of the evidence for systematic variation in S/O is based on measurements of HII regions more metal-rich than those in the NGC 2403 sample; this remains a critical area for future study.