Arsenic contamination of well water in Bangladesh is a widespread problem, the enormity of which came to light in 1998 at an international conference that determined that tens of millions of people had suffered exposure that put their health at risk. As a result, individual wells were tested to determine whether the water they provide is safe to drink. Now research to be published in the December 15 issue of the journal Environmental Science and Technology suggests that close to a third of the wells were labeled incorrectly.

Although the World Health Organization (WHO) recommends that drinking water not contain more than 10 micrograms of arsenic per liter, the wells in Bangladesh were tested to a standard of 50 micrograms per liter. Those delivering water with arsenic concentrations lower than this were painted green, while those that had concentrations exceeding 50 micrograms of arsenic per liter were painted red. While working in Bangladeshi villages, Dipankar Chakraborti of Japavpur University in Calcutta and his colleagues periodically tested water from some of the painted wells and found that they were painted the wrong color. The researchers thus decided to investigate the accuracy of the field test kits used to check the wells. Analysis of water from 290 wells using both the kits and a laboratory technique known as flow injection hydride generation atomic absorption spectroscopy determined that false negatives--that is, green-painted wells drawing water with arsenic concentrations between 50 and 100 micrograms per liter--were generated up to 68 percent of the time. The team then analyzed 2,866 samples collected from colored wells in 60 villages. Of the wells deemed safe, 7.5 percent actually had arsenic concentrations in excess of the limit, the scientists report. In addition, half of the wells classified as unsafe contained water with less than 50 micrograms per liter of arsenic. "Given the scarcity of uncontaminated water," the authors write, "the mislabeling of 50 percent of safe wells has a major socioeconomic impact."

The inaccuracy of the field tests arises from a number of factors, the scientists say, including human error and detection limits that are too close to the 50 microgram per liter standard to allow for reliable results. Chakraborti posits that implementing a laboratory-based testing program would be safer for both the environment (because field kits require toxic chemicals that must be disposed of) and the people who drink the well water. Although his group's analysis suggests that the cost of laboratory tests could be lower than that of the field kits, the lack of trained personnel and infrastructure remain obstacles.