Lyell's percentage method of Stratigraphic correlation is interesting and useful. However, the method has been used at times to support correlation on a more detailed level than data warrant. By treating the number of extant species as successes and the number of extinct species as failures, samples may be treated as following the binomial distribution. The percentage of Recent species in the sample thus becomes the sample mean. The mean and variance of the sample may then be used to calculate a confidence interval and thereby to determine the limits of correlation at the desired confidence level. In this way samples can be referred to the proper epoch and the reliability of the reference can be stated. Treating samples as representatives of binomial populations also permits direct comparisons between two or more samples; the technique holds whether the sample mean is computed on the basis of percentage of extant species or on the basis of percentage in common with some standard interval from the geologic column.
Simpson (1953) and Umbgrove (1946) have shown the importance of epoch by epoch comparisons of percentages of extant species in widely different systematic groups. In addition, it should prove fruitful to compare evolution rates of similar lineages from different geographic localities or faunal realms. A method of testing for significant differences of mean numbers of extant species for these two cases is made possible by applying the binomial distribution to percentage data.