To measure the exchange between the Atlantic and Mediterranean through the Strait of Gibraltar, an army of current meter moorings was deployed for a year in the Strait during 1985-86. A novel aspect of these measurements is the inclusion of conductivity as well as temperature and pressure sensors on each current meter so that the salinity of the flows could be monitored continuously. These salinity measurements determine the water mass characteristics of the flows crossing the sill; they allow definition of the 37 psu isohaline as the interface between in flowing fresher Atlantic water and outflowing saltier Mediterranean water; and they enable time series to be developed for the depth of this interface, for the upper layer inflow, and for the lower layer outflow. From these measurements, the time-averaged outflow of Mediterranean water is estimated to be -0.68Sv (1 Sv = 1 x 10(6) m3 s-1) and the outflow salinity transport, defined to be the outflow times the salinity excess above a basic Atlantic water salinity of 36.1psu, is estimated to be-1.50 x 10(3)m3s-1(1Sv x 1psu = 1 x 10(3)m3s-1) equivalent to a net evaporation over the Mediterranean basin of 52cm y-1. Extrapolated measurements of the inflow from current meters generally deployed below 100m depth yield an estimate for the time-averaged inflow of 0.93 Sv, which is believed to be unrealistically high in view of the better measured outflow and net evaporation. Thus, a more realistic estimate of the inflow is 0.72Sv, equal to the sum of the outflow and net evaporation as required by the mass budget for the Mediterranean Sea. Such estimates of the exchange are smaller by almost a factor of 2 than previous values for the exchange by LACOMBE and RICHEZ (1982). The exchange across the Gibraltar sill is found to be due in nearly equal parts to the mean currents and to the tidal fluctuations. The mean currents are smaller than had been expected reaching a peak value of only about -60cm s-1 in the deep outflow over the sill. The tidal exchange is due to a strong correlation over the tidal period between the depth of the interface and the strength of the inflowing currents. For the M2-tide at the sill, the amplitude of the interface depth is 51 m and the amplitude of the tidal currents is 1.2m s-1; furthermore, the inflow and interface depth have similar phases. As a consequence, the upper layer is deep on the inflowing tide so that a large slug of Atlantic water crosses the sill into the Mediterranean, on the outflowing tide, the interface is shallow so that a large slug of Mediterranean water crosses the sill into the Atlantic. Similar processes occur for the S2, O1 and K1 tides, though the amplitudes are smaller. In this manner, tidal oscillations lead to a time-averaged exchange of water masses across the Gibraltar sill. The inflow and outflow, defined to be the instantaneous transports above and below the 37psu isohaline interface, exhibit M2-tidal amplitudes of 23 Sv and 1.3 Sv respectively. Thus, the tides are large enough to reverse the mean upper layer inflow and lower layer outflow. Daily averaged inflow and outflow transports exhibit low-frequency fluctuations with standard deviations of 0.37Sv and 0.22Sv respectively. Such low frequency fluctuations have been shown previously to be associated with barotropic flows through the Strait of Gibraltar compensating for sea level variations over the Mediterranean due to atmospheric pressure fluctuations (CANDELA, WINANT and BRYDEN, 1989). Finally, from these measurements there appear to be little fortnightly or annual period fluctuations in the exchange through the Strait of Gibraltar,
