The variability of swarming due to specific chemical constituents is becoming well documented. Additionally, water availability is highly influential for swarming; Tremblay et al. documented how drying time prior to inoculation significantly effects the basic swarm assay as over-dried plates do not allow cells to swarm. Similarly, over-wet plates allow for spreading without bacterial surface motility. Here we demonstrate that the formation of swarm tendrils observed with Naloxone benzoylhydrazone changes to the growth surface is due to localized differences in quorum sensing and subsequent rhamnolipid production on these surfaces. The formation of tendrils during swarming is primarily dependent upon liquid effects. The production of rhamnolipid lowers the surface tension within the swarm liquid layer and rhamnolipid also likely acts osmotically to increase water content of this liquid layer by drawing water from the agar matrix as in, similar to the properties of other anionic surfactants. This rhamnolipid-rich, low surface tension, swarm liquid layer then spreads in a classic two-phase tendril pattern described first by Marangoni in the 19th Century. Here we show that fractal formation and increased spreading is dependent on a subpopulation of cells that show robust rhlA transcription at the swarm zone edge. Rhamnolipid becomes a dominant factor in motility when cells transcribe rhlA within close proximity of the advancing swarm edge as on soft agar. Our results are largely in agreement with those of Tremblay and De��ziel, who showed decreased expression of rhlA, rhlB, and rhlC in proximity of the swarm edge. Our inspection of a TP 003 rhlA-gfp reporter always showed greatest fluorescence at the swarm center but the fluorescence at the swarm edge varied. Specifically, if rhlA is transcribed only far away from the swarm edge, as seems to occur on harder agar, cells advance more uniformly by other means without forming rhamnolipid-rich tendrils. The variation in rhamnolipid production at the swarm edge is explained by differences in quorum sensing on these surfaces. This makes sense as rhlA and rhlB are transcribed only in the presence of an active ����quorum.���� Regions exhibiting high fluorescence of an early quorum sensing reporter matched those areas indicative of high rhamnolipid. The difference in timing for expression patterns of rsaL and rhlA reporter fluorescence are in agreement with the onset of these early and late quorum sensing genes.