We present results of a multitransition CS and (CS)-S-34 study of the M17 SW molecular cloud core. Fully sampled maps were obtained in the CS J = 1 --> 0, 2 --> 1, and 7 --> 6 lines and the (CS)-S-34 J = 2 --> 1 line with 18''-36'' resolution. Velocity channel maps reveal the clumpy emission from the dense gas on scales of approximately 0.2 pc (20''). The overall agreement in the cloud morphology among maps of different CS and (CS)-S-34 transitions suggests that all CS and (CS)-S-34 lines originate in the same dense gas. Excitation and opacity effects probably cause the modest differences between the maps. We carried out a detailed excitation analysis of the multitransition data. The J = 2 --> 1 and J = 7 --> 6 transitions of CS, analyzed with a large velocity gradient (LVG) radiative transfer model, produced 250 pixel maps of the volume density and the CS column density over an area of about 1.8 pc x 2.4 pc. Peaks in the CS and the (CS)-S-34 line temperature maps are maxima in column density, but not in density. The density maps shows a fairly uniform, high density (n angstrom 10(6.7) cm-3) throughout the cloud core. An independent estimate of the gas densities from analysis of the (CS)-S-34 observations confirms the CS results. Along with other evidence, these results imply a clumpy cloud model in which the CS emission arises from structures smaller than our beam. We compared the observed CS maps with a specific clumpy cloud model with 179 clumps decomposed from the (CO)-O-18 J = 2 --> 1 maps (Stutzki & Gusten 1990). Model channel maps of CS were synthesized based on the clump parameters listed in Stutzki & Gusten (1990) and were compared with the observed maps. The gas densities used in the models were derived from the clump column densities (based on the (CO)-O-18 J = 2 --> 1 emission) and sizes. Most of the dominant clumps had densities near 10(5) cm-3. The resulting synthesized map does not reproduce the observed CS J = 7 --> 6 emission along the eastern ridge of the core. By assuming a constant gas density for all clumps, we were able to synthesize CS channel maps which reproduce the observed cloud morphology and the line intensities reasonably well. A mean clump density in the models of about 5 x 10(5) cm-3 (about 5 times higher than the density derived from (CO)-O-18) matches the observed CS line ratios and a CS/H2 abundance ratio of about 4 x 10(-9) fits the observed line intensities of the J = 1 --> 0, 2 --> 1, and 7 --> 6 transitions of CS. The discrepancy between densities derived from (CO)-O-18 and CS can be resolved if the clumps have internal density structure. Either smooth density gradients in clumps with sizes just below our angular resolution or a continuation of high-contrast clumping to still smaller scales could account for the difference. While we cannot rule out either of these pictures, it is noteworthy that the scale of the (CO)-O-18 observations (approximately 0.15 pc) is the largest for which the density discrepancy can be resolved with smooth density gradients in unresolved clumps.
