On the generalised friedman test