Georgia. The New England states overall had the highest radon concentrations in water from all sources; state geometric means ranged from 18,500 Bq m-3 in Massachusetts to 88,800 Bq m-3 in Rhode Island. A population-weighted geometric mean for the United States of 6,900 Bq m-3 was reported.
Two major national databases collected by EPA exist for radioactivity in public water supplies. Beginning in November 1980, EPA systematically sampled the 48 contiguous states, focusing on water supplies that served more than 1,000 people (Horton 1983). Radon samples were analyzed with liquid scintillation-counting methods, and samples were targeted to be from as close to the groundwater source as possible and to exclude surface waters. The more than 2,500 public water supplies that were sampled represented 45% of the water consumed by US groundwater consumers. High radon concentrations were found in the waters of the New England states, North Carolina and South Carolina, Georgia, Virginia, Arizona, Colorado, Nevada, Montana, and Wyoming. Individual sample measurements ranged from 0 to over 500,000 Bq m-3, the average was 12,600 Bq m-3 and the geometric mean was 3,700 Bq m-3.
From 1984 to 1986, EPA conducted the National Inorganics and Radionuclides Survey on the basis of 990 randomly distributed samples from the inventory of public water systems in the Federal Reporting Data System (Longtin 1988). The random sample was stratified into four general categories that represented the population served by the system and represented finished water in the distribution system, generally sampled at the tap. Radon was measured with liquid scintillation-counting methods. Longtin (1988) calculated a population-weighted average radon concentration of 9,200 Bq m-3 but did not calculate unweighted statistics. Our committee examined the unweighted data; of the 990 records, 275 had censored observations of less than 3,700 Bq m-3. Values ranged from below the detection level to 949,000 Bq m-3. The distribution of the concentrations was assumed to be log normal and statistics were estimated with the method of maximal likelihood, using SAS and LIFEREG, which accounts for censored data. The geometric mean radon concentration was estimated at 7,500 Bq m-3, the average 20,000 Bq m-3 and the geometric standard deviation 4.06.
A comparison of the two data sets with the data of Hess and colleagues (1985) is shown in figure 2.4. The distributions appear similar in most respects. The 9,000 measurements of Hess and others included the 2,700 measurements of Horton (1983) and some state studies but did not include the Longtin (1988) data. The Hess data have higher percentages of readings in the highest concentration categories than either of the other two data sets. The Horton data have the highest percentage of radon measurements less than 18,500 Bq m-3.
Ambient radon concentration is the concentration of radon in the atmosphere. The outdoor concentration of radon varies with distance and height from