In assessing the time a human operator takes to perform a task (whether the object be to compare the tasks or operator performance) there are many factors which must be taken into consideration. For experimental purposes we assume relationships between these various factors, but how do our assumptions affect our results? For instance, can times be dealt with as if they were normally distributed? A further problem; when comparing two or more sets of limes do we take into account whether or not the data has the same variance?