range (Kinner and others 1987). The issues associated with these systems are the same as those outlined for storage alone (acceptable percentage removals in spite of variable retention times) and for aeration systems (disinfection byproduct formation upon disinfection, off-gas emissions, and precipitate formation).

Some public supplies (for example, for some large communities) use a combination of surface water and groundwater. In this case, radon-free or very-low-radon water might be available to mix with the groundwater, which would result in reduction in concentration due to dilution. Blending has been documented as effective in some water supplies (Kennedy/Jenks 1991b; Dixon and Lee 1988), but reductions are usually low (20%) (Drago 1998). The use of blending depends on whether the water is monitored for compliance as it enters the distribution system or when it reaches the first tap (consumer). In the former case, both types of water would need to be available at the same location. In addition, blending can be effective as a best available technology only if mixing is complete.

In the 1991 proposed rule, EPA specified reverse osmosis (RO) as a best available technology for uranium, radium, and beta-and photon-emitters, but not for radon. RO systems use semipermeable membranes and pressure to separate dissolved species from water. In Sweden, Boox (1995) used an RO filter to treat water in two homes. The systems were installed to improve the taste of the water, but they concurrently reduced the radon content by 90%. RO systems have not been tested for radon removal in the United States, and their use in Sweden exclusively for treatment of radon is doubtful because of their low capacity and relatively high cost. In addition, RO membranes do not work well if turbidity-causing material or precipitates (for example, iron, manganese, silica, and calcium) foul them in the raw water. A brine that contains the contaminants removed from the water is created and must be disposed of. Because the brine is concentrated, the levels of radionuclides might be high, dictating special disposal (as outlined for GAC and aeration systems).

There have been several studies of this method of radon removal from drinking water. The majority of the decrease in radon results from decay during transit or storage in the closed piping network. Most of the studies have documented losses in the range of 10–20% (Rand and others 1991; Kinner and others 1989; Dixon and Lee 1988). A study in Sweden of four waterworks (Mjönes 1997)