This paper is a contribution to experimental meteorology: A sea-breeze front was investigated by aircraft observations and thorough numerical analysis using an unprecedented number of runs crossing the same front within a timespan of 3 1/2h. The 33 runs were flown in a situation of offshore geostrophic wind of 5m/s in 1000 hPa and with the strategy of obtaining information on the four-dimensional field (t = time, x = cross-coastal coordinate, y = coastal-parallel coordinate, z = height): 9 runs in x-direction (and reverse) at different heights to yield x,z-cross-sections of the observed meteorological quantities (specific humidity q, potential temperature θ and the components u, v and w of the wind velocity), assuming a frozen structure in time; the next 7 runs again in x-direction but all at the same level and on the same track to yield x,t-diagrams of the same quantities in order to study the temporal changes compared to those with x and z; the next 10 runs as a zig-zagging flight track crossing the front but drifting in y-direction, all at the same height, in order to obtain the y-dependency; and finally 7 runs for another x,z-cross-sectional analysis, which can be compared to that evaluated from the runs at the beginning of the mission. The paper describes the 4-dimensional dependencies in detail. Pure x-variations at constant z are expressed by VCM low-pass filtered space series (VCM = variance conserving multiresolution, according to Howell and Mahrt, 1994). The x,z-analyses are similar to those in Kraus et al. (1990) and Finkele et al. (1995) verifying these results. The comparison of the x,z-studies gained from the data at the beginning and at the end of the mission show how the sea-breeze frontal area changes its structure. The fluctuations (in time) revealed by the low-pass filtered x,t-runs (same track and the same height) are smaller than the contour intervals chosen in the x,z-cross-sections. This shows, that the single runs, from which the x,z-cross-sections are constructed, reliably and significantly contribute to the interpolated structure. The paper also demonstrates the overall development of the front within the 1/2h of continuous observation. The x,y-fields demonstrate that the y-dependency of the various quantities is generally one order of magnitude smaller than the x-dependency and that the assumption of negligible y-dependency holds in the first order of approximation for a fairly homogenous coast. Convective disturbances of a horizontal scale of 1 to 4 km at the landward side of the front, embedded in the offshore flow and bouncing against the landward propagating sea-breeze front, considerably contribute to variations of the frontal propagation speed and of the frontal shape and also to changes of the parameters with the along-frontal coordinate y.