In a study of arsenic concentrations in public-supply wells in the New England Coastal Basins, concentrations at or above 0.005 mg/L (milligrams per liter) were detected in more samples of water from wells completed in bedrock (25 percent of all samples) than in water from wells completed in stratified drift (7.5 percent of all samples). Iron and manganese were detected (at concentrations of 0.05 and 0.03 mg/L, respectively) at approximately the same frequency in water from wells in both types of aquifers.

Concentrations of arsenic in public-supply wells drilled in bedrock (in the National Water-Quality Assessment Program New England Coastal Basins study unit) vary with the bedrock lithology. Broad groups of lithogeochemical units generalized from bedrock lithologic units shown on state geologic maps were used in the statistical analyses. Concentrations of arsenic in water from public-supply wells in metasedimentary bedrock units that contain slightly to moderately calcareous and calcsilicate rocks (lithogeochemical group M_c) were significantly higher than the concentrations in five other groups of bedrock units in the study unit. Arsenic was detected, at or above 0.005 mg/L, in water from 44 percent of the wells in the lithogeochemical group M c and in water from less than 28 percent of wells in the five other groups. Additionally, arsenic concentrations in ground water were the lowest in the metasedimentary rocks that are characterized as variably sulfidic (group M_s ). Generally, concentrations of arsenic were low in water from bedrock wells in the felsic igneous rocks (group I_f ) though locally some bedrock wells in granitic rocks are known to have ground water with high arsenic concentrations, especially in New Hampshire.

The concentrations of arsenic in ground water also correlate with land-use data; significantly higher concentrations are found in areas identified as agricultural land use than in undeveloped areas. There is, however, more agricultural land in areas overlying the metasedimentary rocks of lithogeochemical groups M_c and the minimally-deformed clastic sediments of group M_md than in areas overlying other lithogeochemical groups. This correlation complicates the interpretation of sources of arsenic to ground water in bedrock. A test of this association revealed that relations between arsenic concentrations and the metasedimentary rocks of group M_c are not weakened when data associated with agricultural land use is removed; the reverse is true, however, if the data associated with the group M_c are removed from the analysis.

The occurrence and variability of arsenic in water from bedrock supply wells could be related to several factors. These include (1) the distribution and chemical form of arsenic in soils and rocks that are part of the ground-water-flow system, (2) the characteristics that influence the solubility and transport of arsenic in ground water, (3) the differing degrees of vulnerability of ground-water supplies to surface contamination, and (4) the spatial associations between land use, geology, and ground-water-flow patterns. Strong relations between agricultural land use and the metasedimentary rocks of group M_c complicate the interpretation of arsenic source to water in these bedrock aquifers. This is due in part to the past use of arsenical pesticides; additionally, few whole-rock geochemical data are available for the rock types in the lithogeochemical groups of aquifers that contain ground water with elevated concentrations of arsenic. Without such data, identifying specific bedrock types as arsenic sources is not possible. In southern Maine and south-central New Hampshire, and in northern Massachusetts, the few available whole-rock analyses suggest, at least for these local areas, a connection between known bedrock chemistry and ground-water arsenic levels.

Although the lithogeochemical group and land-use category variables individually describe much of the variance in the concentrations of arsenic in ground water, the lithogeochemical relation is statistically stronger than the land-use relation. Low concentrations of arsenic in water from bedrock public-supply wells are associated with the metasedimentary rocks of group M_s (characterized as variably sulfidic). This association could reflect a variety of factors and suggests that simple dissolution of arsenic-bearing iron phases, such as sulfides, may not explain concentrations of arsenic in water in this bedrock aquifer group. Whole-rock geochemical data and more complete water-chemistry data, as well as studies of historical variation of arsenic concentrations (time-line studies), and site-specific studies, will be critical in addressing the arsenic source issue.