A5. | | The sample mean value, *x,*based on only a few samples may be a poor estimate of the true (population) mean. Therefore, a no remedial action decision based on *x*less than [the Soil Guideline Value] *G*may not be adequately health protective when *x*is computed from only a small number of samples. Clearly it is desirable to state with a given level on (sic) confidence (say 95th percentile) that the population mean is less than the Soil Guideline Value G. |

A6. | | The necessary calculation involves five simple steps as follows: |

| (i) | calculate the arithmetic sample mean, *x* |

| (ii) | calculate the (unbiased) sample standard deviation, s. |

| (iii) | select an appropriate t value from standard tables. Table A 1.1 gives t values for a 95th percentile confidence limit. t values for other confidence limits are given in Table A 1.2. It should be noted that when using Table A 1.2, the number of degrees of freedom is one less than the number of samples, i.e. for n=8, ν=7. |

| (iv) | calculate the upper 95th percentile bound of sample as: US_{95}=*x*+(*t*.*s*)/*n* |

| (v) | compare the upper bound value *(US*_{95}) with the Soil Guideline Value (G). If the upper bound value is less than G, then the mean value test has been passed, and the site may be considered not to present a significant possibility of significant harm to human health in the context of Part IIA. Conversely, if the test is failed, then the assessor should consider whether it is appropriate to take more samples (because the number on which the test has been based is very low), or to make a determination as contaminated land under Part IIA, taking into account the other requirements of the regime such as the presence of a significant pollutant linkage. |