*3.1. Distribution of 220Rn Concentration*

The mean annual 220Rn concentrations were obtained for 899 houses, the number of houses monitored being reduced from the original 940 houses as was the case for 222Rn [23]. The annual arithmetic mean, and the median were calculated and values less than the DL (<4 Bq m−3) were included in each quarter value. In addition, if a negative value was obtained due to statistical variation as a result of background subtraction, this value was assigned as a zero. The histogram for the mean annual indoor 220Rn concentrations is presented in Figure 1. The mean annual 220Rn concentration was found to vary from <4 to 383 Bq m−3. The arithmetic mean, the median, the geometric mean and the geometric standard deviation were 20.1 ± 36.8, 9.6, 10.0 Bq m−<sup>3</sup> and 3.2, respectively. The 222Rn concentrations varied from 3.1 to 208 Bq m−3. The arithmetic mean, the median, the geometric mean and the geometric standard deviation were 15.5 ± 13.5, 11.7, 12.7 Bq m−<sup>3</sup> and 1.78, respectively [23]. As a comparison, Kim et al. reported that the geometric mean for 220Rn concentrations in Korea was 10.7 Bq m−3. The log-normal cumulative frequency distribution for the indoor 220Rn concentrations is shown in Figure 2. The 220Rn concentration distribution would appear to be close to a log-normal distribution. The distribution of the mean annual indoor 220Rn concentrations was accepted as a log-normal distribution based on the Kolmogorov–Smirnov test at a significance level of 95%.

**Figure 1.** Histogram for indoor 220Rn concentrations.

#### *3.2. Seasonal Variation*

The indoor 220Rn concentration data for each season are presented in Table 1. Negative values in this dataset were eliminated for calculation of the geometric mean. A significant seasonal variation in the 220Rn concentrations for the four seasons was not found. According to Kim et al. [15] and Stjanovska et al. [16], a temporal pattern in the 220Rn concentration data was observed with values in the winter and spring seasons being higher than those in the summer and autumn. Martinez et al. [17] found that the highest concentrations for Mexico City were in the autumn season and the lowest concentrations were in summer.

In the present study, slight differences were noted in the 220Rn concentrations depending on the periods of exposure. The lowest 220Rn concentrations for all types of houses were observed in the winter season (October–December). However, a different relationship was noted for the 222Rn concentrations, namely, that the 222Rn concentrations tended to be higher in winter compared to the other seasons [23]. This was probably because the residents used domestic heaters to maintain a comfortable room temperature in winter, and

consequently there would have been increased ventilation rates due to the contribution of convection and/or stack effect in the rooms.

**Figure 2.** Cumulative frequency distribution for indoor 220Rn concentrations. This figure has been prepared using the mean annual 220Rn concentrations in excess of zero Bq m−3.


**Table 1.** 220Rn concentrations measured in the different seasons.

AM: Arithmetic mean; SD: Standard deviation; GM: Geometric mean; GSD: Geometric standard deviation (dimensionless).

The variation of the 220Rn concentration in the rooms was slightly different from that of 222Rn, which may reflect the differences in the half-lives and sources of 220Rn, despite the fact that there were large fluctuations in the standard deviations for the seasonal variations of 220Rn concentrations. The reason why the indoor 220Rn concentrations did not display a variation similar to 222Rn is unclear at this time.

#### *3.3. Nature of Housing*

Indoor 220Rn concentrations were categorized in terms of the structural features of the housing. The annual mean, the standard deviation, and the geometric mean for the indoor 220Rn concentrations together with number of houses monitored are given in Table 2. The arithmetic and geometric mean concentrations for wooden and concrete-based houses have higher values than those of other structures, although there were large fluctuations in the data. The maximum value was found for a wooden house with a mud wall, the highest 220Rn concentration being 383 Bq m−3. The cause of the high 220Rn concentration of wooden houses is that they have relatively high ratio of the mud wall in comparison to other house structure types. Table 3 lists the ratio of the mud wall in each housing type. Accordingly, the 220Rn concentrations in wooden houses are higher than those for other housing types.


**Table 2.** The mean annual 220Rn concentration for each type of house.

AM: Arithmetic mean; SD: Standard deviation; GM: Geometric mean.

**Table 3.** Ratio of mud wall in each structure type.


With respect to the 220Rn concentrations by region, the overall ratios for wooden houses with mud walls in the Hokkaido—Tohoku, Kanto and Kyushu—Okinawa areas are lower than for those in other areas of Japan. Therefore, the 220Rn concentrations in these former areas also tends to be lower than the values found in the other areas.
