The toxicity of A1 to aquatic biota varies with speciation. Driscoll (1980) has developed a scheme for determining three A1 fractions using the ferron method (Smith, 1971) to detect A1. We have compared the results obtained with Driscoll's method with those obtained when pyrocatechol violet is the complexing agent. The agreement is generally satisfactory. The latter method is more sensitive and less influenced by interferences. Results for water samples, mainly from two streams in southern Norway, are discussed. For most samples the labile monomeric Al-fraction dominates, the major species being either A13 + or AIF 2 +. Possible equilibria between A13 + and some minerals and the temperature changes in these equilibria, are discussed.Water, Air, and Soil Pollution 23 (1984) 81-95. 0049-6979/84/0231-0081502,25.

Questions Field‐based ecosystem mapping is prone to observer bias, typically resulting in a mismatch between maps made by different mappers, that is, inconsistency. Experimental studies testing the influence of site, mapping scale, and differences in experience level on inconsistency in field‐based ecosystem mapping are lacking. Here, we study how inconsistencies in field‐based ecosystem maps depend on these factors. Location Iškoras and Guollemuorsuolu, northeastern Norway, and Landsvik and Lygra, western Norway. Methods In a balanced experiment, four sites were field‐mapped wall‐to‐wall to scales 1:5000 and 1:20,000 by 12 mappers, representing three experience levels. Thematic inconsistency was calculated by overlay analysis of map pairs from the same site, mapped to the same scale. We tested for significant differences between sites, scales, and experience‐level groups. Principal components analysis was used in an analysis of additional map inconsistencies and their relationships with site, scale and differences in experience level and time consumption were analysed with redundancy analysis. Results On average, thematic inconsistency was 51%. The most important predictor for thematic inconsistency, and for all map inconsistencies, was site. Scale and its interaction with site predicted map inconsistencies, but only the latter were important for thematic inconsistency. The only experience‐level group that differed significantly from the mean thematic inconsistency was that of the most experienced mappers, with nine percentage points. Experience had no significant effect on map inconsistency as a whole. Conclusion Thematic inconsistency was high for all but the dominant thematic units, with potentially adverse consequences for mapping ecosystems that are fragmented or have low coverage. Interactions between site and mapping system properties are considered the main reasons why no relationships between scale and thematic inconsistency were observed. More controlled experiments are needed to quantify the effect of other factors on inconsistency in field‐based mapping.