A set of water level time series collected along a transect through a sedimentary island aquifer was used to test the utility of various simple models of ocean tidal propagation in bounded one-dimensional aquifers. Fourier spectra were calculated for the ocean tidal modes and compared with spectra measured in wells along the island transect. Other sources of fluctuation could be neglected. An observed spatial bias in the well responses (attenuations and lags) could not be modeled by a homogeneous aquifer theory. A theory involving composite heterogeneity accounted well for the spatial bias, yielding estimates of aquifer transmissivities and storage coefficients indicating a fivefold difference in hydraulic diffusivity along the transect. A lack of well locations toward one end of the transect reduced the statistical significance of this result, with correlations between regression parameters evident. At the same time, a second bias was seen involving the ratio of signal amplitude and lag with penetration distance into the aquifer, as observed in prior tidal studies. A brief set of numerical experiments showed that horizontal layering in aquifer properties was the most probable cause of this propagation bias. Application of these results to the island data set supported a conceptual stratigraphic model of a highly conductive, sloping stratum underlying a less conductive, superficial sand layer. This model is inconsistent with well logs along the island transect but is supported by additional off-transect well logs. It was concluded that one-dimensional tidal propagation models may be useful in inverse characterization of aquifers with macroscale hydrogeological structures, and that the analysis of measured propagation bias has the potential to yield extra information on aquifer properties in the vertical direction.