9 
Trying to use the turn-over time in this way one en- 
counters a number of problems, The problem that the 
assumption of well-mixed conditions usually is more or 
less remote from reality has already been mentioned, 
Non-stationarity is the second problem, One may eliminate 
the variations Over periods short compared with the turn- 
over time by taking an average over T < To and introducing 
an exchange parameter accounting for the turbulent exchange 
over the smaller time scales, But in practice there are 
often significant variations over periods longer than To 
which one might best account for by taking the most pessi- 
mistic conditions. In section 3 the problem of non-station- 
arity is addressed in a general way, in section 19 the 
variability resulting from seasonal variations is shown 
in the outcome of model calculations. 
Finally, there is a problem in obtaining reliable flux 
estimates, even in a comparatively well-known area as the 
North Sea, The present report gives estimates which for 
certain sub-regions still have a considerable margin of 
uncertainty. 
In this connection the turbulent exchange mentioned above 
is to be considered, Advective fluxes from one sea area to 
the other may, with smaller or larger margins of uncertainty, 
be determined from classical hydrographic data and from 
numerical models. 
However, the turbulent exchange is even more difficult to 
establish. Given advective fluxes, and the distribution of 
a conservative tfracer and its sources, one may come to an 
estimate. This may be done for the North Sea on the basis 
of the salinity distribution and the fresh water input. How- 
ever, in this report such an estimation has not been carried 
out because the prerequisite, a sufficiently reliable picture 
of the advective fluxes, is not established, at least for 
some of the areas,