In northern New England, 43 percent of the population’s drinking water is supplied by private wells in bedrock aquifers, which often contain variable amounts of arsenic. Arsenic in drinking water is of interest for human health studies, however, time-varying arsenic concentrations can complicate estimates of the population’s exposure to arsenic in drinking water. Arsenic in water from these bedrock aquifers is associated with specific geologic formations and to a lesser extent with land use. Concentrations are low to moderate, commonly less than 50 parts per billion (ppb) but occasionally much higher. Variability of arsenic levels from 193 public water supply systems (wells or groups of wells) in Maine, New Hampshire, and Vermont was evaluated; this variation in reported arsenic concentrations may include temporal trends and laboratory analytical variability. Water-quality data for these public-supply systems are from 1989 – 2001 (median number of years per system is 4.5; median number of samples is 4).

Single-well public systems have similar arsenic concentrations to private domestic wells and more closely represent actual aquifer conditions whereas multiple-well public systems represent a blend of waters and can result in dilution of contaminant concentrations through mixing. For the 156 systems with single bedrock wells, 116 had low (10 ppb or less) median arsenic concentrations; 36 had moderate (> 10 to less than or equal to 50 ppb) median concentrations; and 4 had high (> 50 ppb) median concentrations. Arsenic concentrations varied over time by more than 5 ppb for 23 percent of the single-well systems with low median concentrations, by more than 20 ppb for 40 percent of wells with moderate median concentrations, and by more than 50 ppb for 50 percent of wells with high median concentrations. Multiple-well systems had greater variation in arsenic concentrations than single well systems.

These findings have implications for epidemiological studies where historical measurements of exposure to inorganic arsenic are needed. The data suggest that one-time sampling of drinking water from public bedrock wells may sufficiently represent arsenic levels in low-arsenic wells over time periods of up to 10 years. However, concentrations may become increasingly variable for moderate- to high-arsenic wells.