This paper deals with the practical matter of how to apply the Bartlett method of data transformation to time series: The relation between variance and mean over several groups is what has been used to find simple transformations which make the variance independent of the mean, in which case the data are likely to follow a normal distribution with uniform variance. Bartlett’s transformations are shown to be power transformations like the Box-Cox transformations Seasonal and non-seasonal time series which require transformation are considered. It was found that it is better to use the estimated value of the slope directly in the power transformation derived than to approximate to the known and popular logarithmic, square root, inverse, square, inverse of the square root, and inverse of the square transformations.
