On low-gradient floodplains across the globe, small, leveed, sinuous but non-meandering channels connect floodplain lakes to the mainstem river. These "tie" channels operate bi-directionally, creating a jet of sediment laden water into the lakes during periods of rising river stage and draining the lakes when the head gradient between river and lake reverses. Tie channels play a critical role in both floodplain sediment dispersal and ecology, but remain largely unknown and unstudied in the geological literature. Fundamental questions of how tie or any deltaic channel form and what processes control their morphology remain poorly explained by the body of literature relating jets to channel formation. I describe morphological, sedimentological and process data collected on tie channels along three river systems varying in size and climatic settings. Morphologically, when scaled by the river size, channels on all three systems are identical. Based on these field studies, I conducted scaled laboratory experiments of tie channels which provide one of the few quantitative hydro- and morphodynamic tests of widely applied jet theory. These experiments show the hydrodynamics of the experimental tie channels to be consistent with existing jet theory at distances greater than eight channel widths from the channel outlet. Closer to the outlet, where levee formation occurs, existing jet theory fails to accurately predict jet behavior. Four timescales for sediment transport control both the rate of development and morphology of levees formed in the laboratory. These timescales represent transport by: (1) streamwise advection, (2) cross stream diffusion, (3) large-scale turbulence on the outflow margins, and (4) vertical settling through the water column. From these timescales I develop a pair of scaling parameters to describe expected channel morphologies at river mouth outlets uninfluenced by waves or tides. Based on these scaling parameters, I present a conceptual distribution of river mouth morphologies in which tie channels represent a unique end-member morphology.