AS 1301.402:2019 Methods of test for pulp and paper
7.5  Review of system suitability
The test results of the different systems shall be compared to draw conclusions on their level of agreement. As a guide, if the difference between the highest and lowest mean is less than 3 %, agreement would be satisfactory for most purposes, but this depends on the test and level of monitoring required. If agreement is not satisfactory, take action to identify and correct whatever equipment, operator or test atmosphere deficiency is causing the difference. Do not intentionally alter any of the mechanical variables away from aim value in order to improve agreement. This step does not apply when only a single test system is being monitored.
7.6  Generate statistically relevant number of test results
When satisfactory agreement has been achieved, testing by each selected test system shall be carried out on a minimum of n specimens (one replicate test per specimen), where n is given by the square of 2.58 times the coefficient of variation obtained from the pooled standard deviation calculated in Clause 7.7. The total number of tests (n) may include tests done in Clause 7.4 provided, in the case of more than one test system being used, the tests shall be among those judged to be in satisfactory agreement at Clause 7.5. The means of n tests shall be accurate to within 1 % at the 99 % confidence level.
7.7 Calculation of means and standard deviations
A mean and standard deviation shall be calculated for n tests by each system. The means obtained for the test systems shall be selected. The average of the means shall be the best estimate of the mean true value and for the purpose of this Standard is called the assigned value. The pooled standard deviation shall be calculated by averaging the squares of the standard deviations of the selected test systems and taking the square root. The pooled standard deviation shall be the best estimate of the standard deviation of the true values of replicate tests.