The removal of iron from the <0.45m fraction of Tamar River water on addition of both calcium chloride solution and sea water was studied using a continuous autoanalytical system. The addition of increasing concentrations of either calcium ions or sea water caused increased iron removal, but at all of the concentrations studied a fraction of iron was found to remain within the <0.45jim fraction (termed the residual or unreactive fraction). Storage time was shown to have a marked effect on the residual concentration. Kinetic analysis of the experimental results showed that when a residual fraction was taken into account, the data could be described by either a first or second order kinetic model. The first order model gave an approximately linear increase in rate constant with increasing concentrations of calcium ions or sea water, ranging from 1.02 xlO to 6.14x10 s for the addition of calcium ions and from 3.97 xlO to 6.95 xlO s for the addition of sea water. The second orderrate constant also showed an increase with increasing calcium concentration, from 3.95 xlO to 3.02 xlO 1 mol s . When the first and second order profiles were studied and the fit of the model to the experimental residual values was considered, the first order was shown to be a better descriptor of the observed removal of microcolloidal iron. A comparative study was made of iron removal under different conditions of simulated estuarine mixing, following the procedures of Fox & Wofsy {Geochim. Cosmochim. Acta, 1983, vol 47 p211), Mayer (Geochim. Cosmochim. Acta, 1982, vol 46 p2527), Hunter & Leonard (Geochim. Cosmochim. Acta, 1988, vol 52 pi 123) - all using 'discrete sampling during mixing and Duffy (PhD Thesis, University of Southampton) - using continuous sampling with more rapid mixing. Variations among the 'discrete' methods are small relative to those between them and the method of Duffy (1985) e.g. 65 % for the 'continuous' method as compared with 36%, at a salinity of 5, and the first order rate constant was seen to be about an order of magnitude greater for the'continuous'method e.g. 3.97 xlO s as compared with 1.97 xlO s , at a salinity of 5. Further experiments looking specifically at the effect of stirring rate on a 'discrete' sampling method concurred with these observations. In experiments carried out at a salinity of 8, the iron removal increased from 44% to 55% and the first order rate constant increased from 2.96 xlO to 3.58x10 s as a result of a relative increase in stirring rate from 1 to 5. The fact that the first order rate constant for iron removal varied with the energy of the system, even in the least energetic system studied, and that under all conditions the first order model provided the best descriptor of the observed process, showed that the mechanism of microcolloidal aggregation under all these conditions could not be accounted for by Brownian Motion. Shear must therefore be a significant factor and it was concluded that none of the systems studied could afford a test of kinetic order under conditions where Brownian Motion dominantly accounts for particle collisions. The experiments, however, are still of relevance to environmental conditions, since estuarine mixing is greatly influenced by turbulence due to tidal energy and wind stress. Phosphate behaviour was studied concurrently with that of iron and both the percentage removal and kinetic rate constants showed comparable dependence on calcium ion / sea water concentration, although the percentage phosphate removal was consistently lower than that observed for iron. Kinetic rate constants ranged from 0.0019 to 0.0179 s for the first order model and 0.29 xlO to 5.3 xlO' 1 mol" s for second order. When the removal profiles were studied and the fit of the first and second order models to the experimental residual values was considered, the first order model was again shown to be a better descriptor of the observed removal. Phosphate behaviour remained essentially unaffected by changes in experimental methodologies or stirring rate. The factor which most affected iron (stirring) had no major systematic effect on phosphate and therefore it must be concluded that the results indicate a co-removal of phosphate during iron colloid aggregation rather than removal of a common colloidal population containing both constituents.