Recent research has indicated that dissolved organic matter (DOM) may plan an important role in the ability of natural waters to complex metals. This research was conducted because the quantitative nature of this role is uncertain. Gas-liquid chromatography was used to study the hydrolyzable amino acids at twelve sampling sites on the Williamson River at monthly intervals for two years. The relative abundances showed little spacial or temporal variation. The two-year averages for total amino acids ranged from about 0.5 (mu)M to about 8 (mu)M. A separation technique was used to show that (GREATERTHEQ) 96% of the dissolved amino acids were associated with aquatic humus. Since it was found that amino acids contributed less than 1% to humic carbon and since a published report found that carbohydrates contributed less than 2% to humic carbon, this research provided the necessary data to conclude that DOM in the Williamson River is essentially aquatic humus. Humus complexation capacity is often operationally defined as amount of metal bound per unit weight of humus. This research has shown that the titrimetric methods commonly used to obtain this parameter underestimate its magnitude. However, it was shown that these methods can be combined with acidic functional group analyses to determine upper and lower limit for this parameter. For Williamson River humus, the range was 7.2 - 15.4 (mu)mols copper per mg humic carbon. Titrations of humus into a copper-oxalate metal-ion buffer enabled the determination of the copper-humus binding "constant" at humas : copper ratios found in the Williamson River, (LESSTHEQ) 4300. The binding "constant" was a variable and a function of pH. At a humus: copper ratio of 4300, the values of the function at pH 5.0, 5.5, 6.0, and 6.5 were: 3.0 x 10('6), 8.9 x 10('6), 3.0 x 10('7), and 1.7 x 10('8). Current models of metal-humus complexation, were shown to be inappropriate via rigorous mathematical examination and via application to computer-simulated titrations. A model, in which it is assumed that the concentrations of binding sites in humus are normally distributed with respect to the log of the metal binding constant for each site, is proposed. Application of this model to simulated titrations and to experimental data proved it to be superior to other current models.