The Castleisland Radon Survey follow-up to the discovery of a house with extremely high radon concentrations in County Kerry (SW Ireland)

IOP PUBLISHING JOURNAL OF RADIOLOGICAL PROTECTION J. Radiol. Prot. 27 (2007) 275 285 doi:10.1088/0952-4746/27/3/002 The Castleisland Radon Survey follow-up to the discovery of a house with extremely high radon concentrations in County Kerry (SW Ireland) Catherine Organo 1,3 and Patrick Murphy 2 1 Advisory Services Division, Radiological Protection Institute of Ireland, 3 Clonskeagh Square, Clonskeagh Road, Dublin 14, Republic of Ireland 2 School of Mathematical Sciences (Statistics), University College Dublin, Belfield, Dublin 4, Republic of Ireland E-mail: corgano@rpii.ie Received 11 December 2006, in final form 1 June 2007, accepted for publication 6 June 2007 Published 29 August 2007 Online at stacks.iop.org/jrp/27/275 Abstract In July 2003, a house with a seasonally adjusted annual average radon concentration of 49 000 Bq m 3 was identified near Castleisland in County Kerry (SW Ireland). The possibility that other houses with similar extreme radon concentrations could be present in the surrounding area triggered the setting up of a localised radon survey, the so-called Castleisland Radon Survey (CRS). To this end, approximately 2500 householders living in four 10 10 km 2 grid squares from the Irish grid closest to the town of Castleisland were invited to participate. Four hundred and eighteen householders responded to the invitation (17% response rate) and 383 home results were used for further analysis. In the 400 km 2 encompassing the four studied grid squares, 14% of the homes were found to have a seasonally adjusted annual average radon concentration above the national reference level of 200 Bq m 3 while 2% above 800 Bq m 3. An average radon concentration of 147 Bq m 3 was calculated. This can be compared with the average radon concentration of 98 Bq m 3 calculated for the same four grid squares on the basis of 80 measurements carried out during the Irish National Radon Survey (NRS) which was conducted between 1992 and 1997. The fourth highest radon concentration (6184 Bq m 3 ) and three of the ten highest ever measured in Ireland were all identified during the CRS. This shows that localised and targeted radon surveys are an invaluable tool for the identification of homes at highest risk from high radon concentrations. Two of the four grid squares investigated during the CRS are currently designated as high radon areas (defined as areas where 10% or more of all houses are predicted to exceed 200 Bq m 3 ) as predicted by the NRS. A thorough statistical analysis of the CRS and NRS data was carried 3 Author to whom any correspondence should be addressed. 0952-4746/07/030275+11$30.00 2007 IOP Publishing Ltd Printed in the UK

276 C Organo and P Murphy out and indicated that both datasets could be merged and used to refine the original NRS predictions. The results indicate that two of the four studied grid squares could potentially be redesignated. The practical feasibility and overall benefit of updating the Irish radon map in light of this analysis is described. 1. Introduction In July 2003, a radon measurement carried out over a 3-month period in a house near Castleisland in County Kerry (SW Ireland, figure 1) by the Radiological Protection Institute of Ireland (RPII) revealed a seasonally adjusted annual radon concentration of approximately 49 000 Bq m 3, the highest concentration ever recorded in a house in Ireland. This house is hereafter referred to as the Castleisland house. The householder s wife, a former smoker who had given up smoking some 20 years previously, had died 5 years earlier from lung cancer, and a never smoker himself, the householder was diagnosed with the same disease in March 2003. He had been advised by his consultant to test his house for radon, which is what led to the discovery of the extreme radon concentrations (Organo et al 2004). The story was extensively covered by the local and national media and has alerted the scientific community. A year and a half later, the attention of the public was once again focused on the issue after the householder died of lung cancer. Concerned by the potential existence of more houses with high indoor radon concentrations in the area, the RPII organised a regional radon survey in September 2003, the so-called Castleisland Radon Survey (CRS). An area of 400 km 2 was designated (figure 1), covering four 10 10 km 2 grid squares from the Irish national grid closest to the town of Castleisland (namely, grid squares 0910, 0911, 1010 and 1011). Based on the results of the National Radon Survey (NRS), grid squares 1010 and 1011 are currently classified as high radon areas 4 (Fennell et al 2002). County Kerry is particularly affected by radon. Since the RPII started its national measurement programme in the early 1990s, 192 Irish homes have been identified with an average radon concentration above 1000 Bq m 3, and 45 of these (or 23%) are located in County Kerry. The reason for the prevalence of high radon concentrations in this part of the country is thought to be primarily linked to its geology (GSI 1995). Approximately half of County Kerry, including the Castleisland area, is underlain by carboniferous limestones. This type of rock normally contains low concentrations of natural radionuclides but it is heavily karstified in this area, with extended underground groundwater systems 5 and, more importantly, it is overlain by bands of Namurian shales known to be very rich in uranium in places (O Connor et al 1994). A similar geology exists in County Clare, further north, and this area has been extensively studied by O Connor and his collaborators (1992, 1993). They concluded that while the ultimate source of radon remains conjectural, the greatly enhanced permeability of the karstified limestones was of fundamental importance in providing a means of rapid radon transport into surface soils and subsequently into buildings. 4 Defined as grid squares in which the predicted percentage of dwellings exceeding the national reference level of 200 Bq m 3 is 10% or more. 5 The Carrignafeela and Potaley underground stream system, the Tobermaing sink and rising and Crag Cave are all mentioned by the Geological Survey of Ireland (GSI 1995) as important karst features in the Castleisland area.

The Castleisland Radon Survey 277 0911 Castleisland 1011 1010 0910 Figure 1. Radon map of the southwest of Ireland dividing the country in 10 10 km 2 grid squares. The light and darker shaded squares represent the high radon areas in which the predicted percentage of dwellings exceeding the national reference level is 10%. The studied area around Castleisland, as well as the four grid squares discussed in this study (0910, 0911, 1010 and 1011) are also indicated. (This figure is in colour only in the electronic version)

278 C Organo and P Murphy 2. Survey methodology The householders names and addresses as well as the grid square reference for 2500 homes in the four Castleisland grid squares were obtained from a commercial database. The RPII wrote to all 2500 householders with a letter referring to the existence of the Castleisland house and urged them to have their home tested for radon. Measurements were offered on a fee-paying basis. To those who requested a measurement, two passive CR-39 alpha track detectors were issued from the end of September 2003 onwards, to be placed in a main living area and in an occupied bedroom for a minimum period of 3 months. Participants were asked to complete a questionnaire regarding the year of construction of the house, the type(s) of floor throughout the house and the type of water supply. They were also asked to indicate as accurately as possible the location of their home on a copy of the Ordnance Survey map from the local area (OSI 1996, OSI 2002). This information was subsequently used to assign the measurement result for each house to the correct grid square. Upon completion of the measurements, the detectors returned to the RPII were processed as described by Fennell et al (2002). An estimated whole house annual average radon concentration was calculated for each home by averaging the results from the individual detectors, assuming equal occupancy between both measured rooms and by applying a seasonal correction factor to allow for the time of the year the measurements were made (adapted from Wrixon et al 1988). All householders were individually notified of their measurement results and, where appropriate, recommendations were made regarding the need for remedial action. 3. Results Of the 2500 householders who were initially invited to participate, 418 responded positively (17% response rate) and of these, 400 returned the detectors within the survey timeframe. Based on the information each householder submitted to the RPII, 383 of these 400 houses could be located with certainty within the four grid squares of interest. A specific grid square could not be allocated to a further 30 measurements because the householders either failed to return the map or there was an uncertainty about the exact location of the house. Table 1 shows the CRS results for the four individual grid squares and taken together as one single area. For comparison, the results from the NRS are also included. Table 2 is a synopsis of all the radon measurements carried out to date by the RPII since the early 1990s in the four grid squares. This includes the results of a measurement programme carried out at Kerry County Council s request which took place just after the discovery of the Castleisland house in 89 local authority homes. The results from an additional 95 measurements not belonging to any of the surveys already mentioned are also included. The Castleisland house belongs to this latter dataset of 95 houses. From table 1 it is clear that the arithmetic mean calculated on the basis of the CRS measurements is higher than the one calculated by the NRS for three of the four studied grid squares. Furthermore, in all four grid squares, the CRS geometric mean and the percentage of homes with measured radon concentration in excess of the national reference level of 200 Bq m 3 are also higher than the ones based on NRS measurements. Overall, on the basis of the CRS results, grid square 911 appears to be the worst of the four studied squares (highest arithmetic and geometric means, highest maximum concentration measured and highest number of homes measured with an average radon concentrations in excess of 800 Bq m 3 ). The Castleisland house is also located in this particular grid square. As more houses were measured for the CRS than for the NRS, the RPII recognised that the availability of these new data could improve the knowledge of the prevalence and risk from radon in the Castleisland area. However, the sampling methodologies in both surveys

The Castleisland Radon Survey 279 Table 1. Summary results from the Castleisland Radon survey (CRS) and comparison with the National Radon Survey (NRS) (Fennell et al 2002). Four grid squares together 0910 0911 1010 1011 Number of houses measured NRS 80 26 15 26 13 CRS 383 b 105 b 63 b 92 b 93 b Number of houses measured above 200 Bq m 3a NRS 9 (11.2%) 2 (7.7%) 1 (6.7%) 3 (11.5%) 3 (23.1%) CRS 54 (14.1%) 15 (14.3%) 9 (14.3%) 15 (16.3%) 12 (12.9%) Predicted % of houses above 200 Bq m 3a NRS 9.20 4.96 5.67 14.67 10.11 Number of houses measured above 800 Bq m 3 NRS 0 0 0 0 0 CRS 6 0 3 1 1 Arithmetic mean (Bq m 3 ) NRS 98 84 85 121 94 CRS 147 105 276 115 152 Maximum concentration measured (Bq m 3 ) NRS 661 552 627 661 397 CRS 6184 632 6184 945 4310 Geometric mean (Bq m 3 ) NRS 56 57 45 62 53 CRS 70 65 80 71 74 a 200 Bq m 3 is the national reference level for domestic dwellings in Ireland. b The total number of homes measured in the four grid squares (383) and the sum of the total number of homes measured in each grid square (105+ 63+ 92+ 93) is differentdueto the fact that aspecific grid square could notbe allocated to 30 of the 383 measurements (householders failed to return the map or uncertainty about the exact location of the house). Table 2. Synopsis of all the radon measurements carried out so far by the RPII since the early 1990s in the four Castleisland grid squares taken together (CRS, Castleisland Radon Survey; NRS, National Radon Survey). CRS NRS Kerry County Council Others a All data a Number of houses measured 383 80 89 95 (94) 647 (646) Number of houses measured above 200 Bq m 3 54 9 0 23 (22) 86 (85) Number of houses measured above 800 Bq m 3 6 0 0 7 (6) 13 (12) Arithmetic mean (Bq m 3 ) 147 98 28 821 (249) 223 (139) Max. annual average concentration measured (Bq m 3 ) 6184 661 84 49 000 (3677) 49 000 (6184) Geometric mean (Bq m 3 ) 70 56 25 93 (86) 61 (61) a The numbers in brackets in these two columns do not take the Castleisland house into account. are not identical 6 and a small number of very high radon concentrations were identified during the CRS. Such high concentrations were not found in the NRS and a statistical bias could be introduced into a combined dataset. Because of this, a thorough statistical analysis of both datasets was undertaken. 6 The NRS householders were randomly selected from the electoral register at a time when awareness of radon in Ireland was low (TNS mrbi 2004) and the measurements were offered free of charge. For the CRS, all 2500 householders living in the studied area around Castleisland were invited to participate, they were specifically made aware of the existence of the Castleisland house and were charged a fee for the test. The CRS measurements also had to be seasonally adjusted to provide a best estimate of the annual average concentration while those carried out during the NRS were done over a 12-month period and therefore seasonal correction was not necessary.

280 C Organo and P Murphy 4. Statistical analysis 4.1. Checking for log-normality of the CRS dataset (353 observations) Results of several radon surveys are consistent in showing that radon measurements averaged over a sufficient number of houses in an appropriately sized grid square follow a log-normal distribution. Where it can be shown that this is the case, it is then simple to estimate the proportion of the distribution above a given threshold from the geometric mean and geometric standard deviation (Miles 1998). This type of modelling has been reported in the published literature and indeed forms the basis of the NRS (Fennell et al 2002). In this modelling approach, a constant background concentration needs to be subtracted from the radon measurements before they are log-transformed and thenrsindicated that an appropriate figure for the background radon concentration in Ireland was 6 Bq m 3. The entire CRS dataset (353 measurements whose grid locations could be clearly identified) was modelled following this approach. Normal quantile (QQ) plots of the log-transformed radon concentrations indicated that, when standardised by the removal of a background concentration of 6 Bq m 3, the CRS data followed a log-normal distribution in each of the four grid squares. This analysis also highlighted that some deviations from a perfect fit were caused by a small number of outliers (extreme observations) which did not belong to the same distribution as the remainder of the CRS data. These were identified both from the normal quantile plots and from boxplots 7 for each of the four grid squares, and, following this procedure, nine outliers were identified, three in grid square 0911 (6184, 4156 and 1263 Bq m 3 ), one in grid square 1010 (945 Bq m 3 ) and five in grid square 1011 (4310, 692, 622, 565 and 433 Bq m 3 ). They were removed from the entire dataset of 353 observations, leaving 344 data on which log-normality Kolmogorov Smirnov (KS) tests were then applied (as a whole dataset as well as for each individual grid square). The results indicated that there was no evidence to reject the null hypothesis that the background-corrected data in each dataset followed a log-normal distribution (P-values for each of the KS tests exceeded 0.2). 4.2. Combining the CRS dataset cleaned from outliers (344 observations) and the NRS dataset (80 observations) The CRS measurements were all carried out between the months of September and January. It is a convenient coincidence that the average of the seasonal correction factors used by the RPII 8 for these months happens to be entirely representative of a yearly average. In other words, the CRS measurements do not require any seasonal adjustment. This being the case, it is reasonable to consider combining the CRS data for the four grid squares with those from the NRS. This new combined dataset now consists of 424 observations (344 from the CRS plus 80 from the NRS). Histograms of the CRS, NRS and of the merged datasets (CRS + NRS) showed that the NRS and CRS data were consistent, and this was further established by the normal quantile plots and the KS tests. 7 In the boxplot, outliers are classified as those which lie further than 1.5 interquartile ranges from the upper and lower quartiles. 8 Since the early 1990s, the RPII has used seasonal correction factors adapted from those derived by the National Radiological Protection Board (now known as the Health Protection Agency or HPA) in the UK (Wrixon et al 1988). No assessment has ever been made to confirm their suitability to the Irish situation, mainly because of the assumption that due to their close geographical proximity, both countries have similar seasonal characteristics. This issue is currently being addressed by the RPII.

The Castleisland Radon Survey 281 Table 3. (a) Model predictions with 90% confidence limits based on NRS data (adapted from Fennell et al 2002) and (b) model predictions with 90% confidence limits based on CRS/NRS merged datasets, after removal of outliers. Grid square (a) Number of homes measured Predicted percentage of homes above 200 Bq m 3 : central estimate (lower and upper confidence limit) 0910 26 5.0 (1.1 13.2) 0911 15 5.7 (0.6 18.6) 1010 26 14.7 (6.1 26.0) 1011 13 10.1 (1.5 25.9) Total 80 (b) 0910 131 11.2 (7.6 15.6) 0911 75 8.5 (4.7 13.8) 1010 117 11.9 (8.0 16.6) 1011 101 7.1 (4.1 11.2) Total 424 4.3. Predictions calculated based on the merged datasets (CRS + NRS, 424 observations) Having demonstrated that the radon concentrations from the merged datasets for each of the four grid squares followed a log-normal distribution, we computed predictions for the percentage of homes above the national reference level of 200 Bq m 3. We removed a background value of 6 Bq m 3 from the radon concentrations in each square to obtain data which follow a log-normal distribution. Once these values have been log-transformed, they can be standardised by subtracting the (arithmetic) mean and dividing by the standard deviation of these log-transformed values to obtain data which follow the standard normal distribution. To calculate the proportion of homes exceeding 200 Bq m 3 for a given square, this reference level was also background-corrected to compensate for the fact that a background value of 6 Bq m 3 was subtracted from all the radon concentrations. Standard normal tables were then used to compute the predictions displayed in table 3. 5. Discussion The reason for merging the CRS and NRS datasets was to take advantage of the extra CRS data to improve our knowledge of the prevalence and risk from radon in the Castleisland area. Table 3 shows that this was partly achieved as the confidence intervals of the predictions based on the merged dataset (table 3(b)) are much smaller than the confidence intervals of the original predictions (table 3(a)). Overall, the predictions based on the merged data indicate that the radon problem in grid squares 0910 and 0911 is greater than previously thought, while in grid squares 1010 and 1011, the radon problem seems to be smaller. From the point of view of reviewing the current designation of grid squares 0910 (upgrade to high radon area) and 1011 (no longer a high radon area), it is important to bear in mind that although the confidence intervals have greatly improved, the reviewed predictions are still subject to some sampling errors. The new confidence intervals still have a significant width of approximately 8%. Additionally, 95% confidence intervals for the geometric mean radon concentrations have a width of approximately 20 Bq m 3.

282 C Organo and P Murphy There are also additional points which need to be carefully considered because of their possible influence in the overall sampling process. For example, inviting all the householders in the Castleisland area to take part in the CRS and making them aware of a serious health risk in their living environment could have either resulted in an increase of the participation rate or, conversely, in a coping strategy of denial from their part, consciously or subconsciously deciding to ignore the problem. The CRS had a 17% participation rate compared to 17 36% for the NRS. The fact that the CRS participants were charged a fee for the measurement could also have influenced their decision to test. From that point of view, previous studies give contradictory results (DEFRA 2000, Lee1992). Lee (1992) also showed that despite being told the opposite, some householders always feel that they are not in a radon prone area or they underestimate the risks. Others resent being told what to do in their own home and, more generally, people are more pre-occupied by immediate worries. Another factor which could have influenced the CRS measurements is the fact that compliance with the 1997 amending Building Regulations (Stationery Office 1997) in Ireland requires that all new homes built since 1 July 1998 have a potential means to extract radon from the substructure (usually a sump) and for those new homes built in high radon areas, the fitting of an approved sealed membrane of low radon permeability over the footprint of the building (radon barrier). Homes tested during the NRS were all built before the introduction of these regulations while the CRS was conducted afterwards. The results of three separate radon surveys in three different high radon areas to study the impact of the 1997 amending Building Regulations on radon concentrations in Irish homes indicated a reduction of approximately 30% in the average radon concentration and in the percentage of homes exceeding the national reference level of 200 Bq m 3 for homes which had commenced construction after the introduction of the building regulations compared to those which had not benefited from these preventative measures (Synnott et al 2003, 2004,Organo2006). A similar effect could be expected in two of the four Castleisland grid squares which are designated as high radon areas (1010 and 1011). However, based on the responses to the questionnaires, only 7% of the CRS homes tested had commenced construction after the introduction of the building regulations thereby limiting their effects on the measured radon concentrations. One of the positive findings of the CRS is that this type of localised study is very effective in identifying houses with high radon concentrations. In the studied area, the NRS failed to identify any home with an annual average radon concentration exceeding 800 Bq m 3 (table 1). The CRS identified six such homes, five of them with radon concentrations above 1000 Bq m 3. Since starting its measurement programme in the early 1990s, the RPII has identified 87, 13 and 11 homes with an average radon concentration above 200, 800 and 1000 Bq m 3, respectively, in the four grid squares surrounding Castleisland. Fifty-four of the 87 homes above 200 Bq m 3 (62%), 6 of the 13 above 800 Bq m 3 (46%) and 5 of the 11 above 1000 Bq m 3 (45%) were identified by the CRS. Moreover, three of the ten homes measured with the highest radon concentrations in Ireland were also identified by the CRS. If we include the Castleisland house itself, it also means that four of the ten Irish homes so far identified with the highest radon concentrations are all located in the Castleisland area. National radon surveys are designed to map the geographical distribution of indoor radon concentrations throughout a country to identify those areas at highest risk. They are a necessary exercise for national authorities to facilitate decision-making and to concentrate their financial and staff resources in reducing the health risk from radon. Mapping can be carried out using administrative unit areas but these are irregularly shaped and sized and sampling can be difficult. For geographically based surveys such as the NRS, it is preferable to use unit areas of identical shape and size, ensuring the same number of householders are invited to participate in the survey for each unit area. For the NRS, a 10 10 km 2 grid square was chosen as a compromise between the requirement

The Castleisland Radon Survey 283 Table 4. Radon concentrations measured in the five homes which had remedial work carried out after the CRS. House no. 1 House no. 2 House no. 3 House no. 4 House No. 5 Grid square 911 1010 1010 1011 1011 Annual average concentration 4156 945 524 622 292 measured before remedial work (Bq m 3 ) Annual average concentration 22 80 103 55 55 measured after remedial work (Bq m 3 ) Percentage of reduction achieved 99 91 80 91 81 by the remediation (%) Reduction factor (ratio of initial to post-remediation concentration) 189 12 5 11 5 for detailed mapping and the need for sufficient results within each area to allow a meaningful statistical analysis to be carried out. Statistical advice indicated that a minimum sample size of five dwellings per grid square was appropriate (Fennell et al 2002). Considering the very high level of fluctuations that can be found between radon measurements in neighbouring houses, relying on a small sample size to classify large areas is often a disadvantage because it can lead to large uncertainties in estimating the radon risk at a more local level. A much larger sample size is therefore preferable for a more cost-effective radon strategy, which will also increase the detection rate of homes with high radon concentrations. A second positive finding of the CRS is the effect of remediation on the reduction of the average radon concentration in each grid square and consequently on the dose reduction from radon in these areas. Only five of the 54 householders (or 9%) whose house had been identified during the CRS with an average radon concentration above the national reference level took remedial action: one home in square 911 and two homes in each of the 1010 and 1011 squares. Not only was the remediation extremely successful for each home (table 4), but the reduction in the average radon concentration for each grid square achieved by the remedial work is clear: from 276 to 211 Bq m 3 (23% reduction), 115 to 101 Bq m 3 (12% reduction) and 152 to 143 Bq m 3 (6% reduction) in squares 911, 1010 and 1011, respectively. If all four grid squares are taken together, the average drops from 147 to 130 Bq m 3 (12% reduction). A vast array of literature has been published highlighting the effectiveness and cost versus health benefit of radon remediation. Reduction factors (ratio of initial to post-remediation measurement) well in excess of 100 are frequently achieved by the installation of active radon sumps and maximum reduction factors with this system are usually associated with the highest initial radon concentrations (Synnott et al 2007, Cliff et al 1994) which is also what we observe in our case (table 4). It is now widely accepted that radon sumps overall achieve the highest radon reductions (>80% up to 98 99%; Henschel 1994). Coskeran et al (2002) even mention a reduction of average radon concentrations in excess of 400%. In Ireland, we observe an average reduction of 72% after the installation of sumps (Dawson and Stephens 2005). Other techniques (sealing all the cracks in the floors, increasing the natural ventilation in the building using wall or trickle vents, increasing or forcing the ventilation of the under floor space, positive pressurisation of the building) do not usually perform as effectively and systematically as active sumps (Welsh and Stephen 1994, Ratti et al 1998) and cannot be used as widely, although they are, in many cases, sufficient and some of them much cheaper in reducing moderately high radon concentrations. The overall remediation rate of 9% (5/54 homes) is one disappointing outcome of the survey as it is well below the more recent estimate of the overall Irish remediation rate of 29%

284 C Organo and P Murphy (Dawson and Stephens 2005). Poor remediation rates have previously been reported in Ireland (Ryan and Kelleher 1999), in the UK (Lee and MacDonald 1994,Bradley1996,Coskeranet al 2002) and in the United States (Lee 1992) but in most countries, including Ireland, remediation is a voluntary process and the cost is entirely covered by the householder. When they are asked to list the reasons why they have not taken remedial action, householders invariably mention the cost, they do not believe there is a risk, they find it too difficult to get advice on remediation or they simply do not trust the accuracy of the measurement in the first place (Dawson and Stephens 2005, Lee and MacDonald 1994,Bradley1996, Ryan and Kelleher 1999,Lee1992). This applies despite repeated efforts by national authorities to inform the public about the risk from radon, promote easy access to remediation advice, involve local authorities and develop the remediation industry. As an incentive, some countries have introduced various types of grant schemes to help covering the remediation costs, and in other countries, either advisory or mandatory radon testing is included in the conveyancing process (Scivyer 2005). In 2005, the RPII advised the Irish government to consider a set of general recommendations to improve the identification and remediation rates of Irish homes with radon concentrations above the national reference level (RPII 2005). These include, among others: involving health authorities in the process of disseminating information, including radon measurements and remediation in the conveyancing process and implementing a programme of free radon measurements in all high radon areas. 6. Conclusion Following the identification of a house with extremely high radon concentrations near Castleisland in County Kerry (SW Ireland), a radon survey of the local area was undertaken. Based on the Irish National Radon Survey, two of the four 10 10 km 2 grid squares surveyed are currently classified as high radon areas. Despite identifying houses with radon concentrations as high as 6184 Bq m 3, the fourth highest level ever measured in an Irish home, extreme radon concentrations such as those found in the Castleisland house were not identified during the course of the Castleisland Radon Survey. Not currently classified as a high radon area, the grid square where the Castleisland house was originally identified first appeared to be the most affected of all four studied grid squares (highest arithmetic and geometric means, highest percentage of measured homes exceeding 200 Bq m 3, maximum radon concentration measured during the Castleisland survey, 6184 Bq m 3 ). However, this fact was not confirmed by a statistical review of the percentages of homes predicted to exceed the national reference level of 200 Bq m 3 based on a merged dataset of results from the National Radon Survey and from the Castleisland Radon Survey. On the other hand, the statistical analysis indicated that two of the four studied grid squares could potentially be redesignated. The RPII is currently using the lessons learnt from the Castleisland Radon Survey to investigate the practical feasibility and overall benefit of updating its Radon in Irish Dwellings map accordingly. References Bradley E J 1996 Responses to remediation advice Proc. 9th Int. Congr. of the International Radiation Protection Association (Vienna, April 1996) www.irpa9.com Cliff K D, Naismith S P, Scivyer C and Stephen R 1994 The efficacy and durability of radon remedial measures Radiat. Prot. Dosim. 56 65 9 Coskeran T, Denman A R, Phillips P S and Gillmore G K 2002 A critical comparison of the cost-effectiveness of domestic radon remediation programmes in three counties of England J. Environ. Radioact. 62 129 44

The Castleisland Radon Survey 285 Dawson D and Stephens D 2005 Attitudes to radon remediation among householders with high radon concentrations in their home 4th Irish National Radon Forum (Tralee, Co. Kerry, Nov. 2005) www.rpii.ie/radon/natradonforum. htm DEFRA (Department for Environment Food and Rural Affairs) 2000 Good Practice Guide www.defra.gov.uk/ environment/radioactivity/radon/pdf/radon laguide.pdf Fennell S G, Mackin G M, Madden J S, McGarry A T, Duffy J T, O Colmáin M, Colgan P A and Pollard D 2002 Radon in dwellings the Irish National Radon Survey RPII Report 02/1 (Dublin: Radiological Protection Institute of Ireland) GSI (Geological Survey of Ireland) 1995 Dingle Bay Sheet 20 (Bedrock Geology 1:100 000 Map Series) (Dublin: GSI) Henschel D B 1994 Analysis of radon mitigation techniques used in existing US houses Radiat. Prot. Dosim. 56 21 7 Lee T R 1992 The public s perception of radon Radiat. Prot. Dosim. 42 257 62 Lee T R and MacDonald S 1994 Public responses to indoor pollution from radon Radiat. Prot. Dosim. 56 331 7 Miles J C H 1998 Mapping radon prone areas by lognormal modelling of house radon data Health Phys. 74 370 8 O Connor P J, Pyne J F, McLaughlin J P and Madden J S 1994 Radon exhalative properties, radioelement content and rare earth element composition of Namurian phosphorite deposits, Co. Clare GSI Report Series RS 94/3 (Environmental Geology) (Dublin: Geological Survey of Ireland) O Connor P J et al 1992 Mapping of 222 Rn and 4 He in soil gas over a karstic limestone granite boundary: correlation of high indoor 222 Rn with zones of enhanced permeability Radiat. Prot. Dosim. 45 215 8 O Connor P J et al 1993 Assessment of the geological factors influencing the occurrence of radon hazard in a karstic region GSI Report Series RS 93/2 (Environmental Geology) (Dublin: Geological Survey of Ireland) Organo C 2006 The Kilkenny Building Regulationssurvey Unpublished data (Dublin: Radiological Protection Institute of Ireland) Organo C, Ellard A, Fenton D, Synnott H, O Colmáin M, Prenter S, O Reilly S and Colgan P A 2004 Variations in radon-222 concentrations J. Radiol. Prot. 24 107 20 OSI (Ordnance Survey of Ireland) 1996 Discovery Series Map 72 (Kerry Cork Limerick 1:50 000) 1st edn (Dublin: OSI) OSI (Ordnance Survey of Ireland) 2002 Discovery Series Map 71 (Kerry 1:50 000) 2nd edn (Dublin: OSI) Ratti A, Piardi S and Baglioni A 1998 Building techniques for the control of radon pollution in Italy Radiat. Prot. Dosim. 78 49 54 RPII (Radiological Protection Institute of Ireland) 2005 Draft action plan to identify and remedy Irish houses with radon concentrations above the National reference level Unpublished Report (Dublin: RPII) Ryan D and Kelleher C C 1999 A survey of householders mitigation strategy Eur. J. Public Health 9 62 4 Scivyer C 2005 A guide to buying and selling homes in radon affected areas of Europe ERRICCA 2 (EC Contract No FIRI-CT-2001-20142) Work package 2 Deliverable 3 (BRE Report No 221045, Watford UK) Stationery Office 1997 Building Control Regulations 1997. Statutory Instrument No. 496 of 1997 (Ireland: Stationery Office) Synnott H, Colgan P A, Hanley O and Fenton D 2007 The effectiveness of radon remediation in Irish schools Health Phys. 92 50 7 Synnott H, Pollard D and Colgan P A 2003 Radon levels in domestic dwellings built since the introduction of the 1997 Building Regulations report of a pilot survey in Ennis, County Clare RPII Report 03/2 (Dublin: Radiological Protection Institute of Ireland) Synnott H, Pollard D, Colgan P A and Fenton D 2004 A survey of the impact of amending Building Regulations on radon concentrations in Irish homes Proc. 11th Int. Congr. of the International Radiation Protection Association (Madrid, May 2004) www.irpa11.com TNS mrbi 2004 RPII Awareness Research (Dublin: TNS mrbi) Welsh P and Stephen R 1994 Radon remediation of dwellings with suspended timber floors case studies from the Building Research Establishment (UK) Radiat. Prot. Dosim. 56 45 50 Wrixon A D, Green B M R, Lomas P R, Miles J C H, Cliff K D, Francis E A, Driscoll C M H, James A C and O Riordan M C 1988 Natural Radiation Exposures in UK Dwellings, Chilton NRPB-R190 (London: HMSO)