The article by James S. Farris, Mari Källersjö, Arnold G. Kluge, and Carol Bult addresses the issue of assessing incongruence between sources of phylogenetic evidence. Traditionally, such assessments have been largely intuitive or based on arbitrary interpretations of quantitative indices. The authors propose a statistical test to overcome these limitations by constructing a null hypothesis of congruence and obtaining its null distribution through random partitioning. The test involves selecting a measure of incongruence, such as the incongruence length difference ($D_{XY}$), which quantifies the disagreement between two matrices. The null distribution is determined by summing over all possible partitions of the characters into matrices of the original sizes. The significance of the observed $D$ value is assessed by comparing it to the distribution obtained from randomly selected partitions. The authors introduce the incongruence length difference ($D_{XY}$) as a measure that isolates the source of incongruence, avoiding the issue of equal contributions from well-supported and disputed groups. This measure is particularly suitable for measuring incongruence between matrices because it focuses on the strength of evidence. The test is implemented in a program called arn, which can handle both two-sample and multi-sample tests. The program has been applied in several studies, demonstrating its effectiveness in assessing incongruence between different types of data, such as morphological and molecular data. The discussion section addresses criticisms and clarifies the limitations of the test, emphasizing that it is effective in detecting incongruence between matrices, even when one matrix is much larger than the other. The authors also provide examples to illustrate the test's capabilities and limitations.The article by James S. Farris, Mari Källersjö, Arnold G. Kluge, and Carol Bult addresses the issue of assessing incongruence between sources of phylogenetic evidence. Traditionally, such assessments have been largely intuitive or based on arbitrary interpretations of quantitative indices. The authors propose a statistical test to overcome these limitations by constructing a null hypothesis of congruence and obtaining its null distribution through random partitioning. The test involves selecting a measure of incongruence, such as the incongruence length difference ($D_{XY}$), which quantifies the disagreement between two matrices. The null distribution is determined by summing over all possible partitions of the characters into matrices of the original sizes. The significance of the observed $D$ value is assessed by comparing it to the distribution obtained from randomly selected partitions. The authors introduce the incongruence length difference ($D_{XY}$) as a measure that isolates the source of incongruence, avoiding the issue of equal contributions from well-supported and disputed groups. This measure is particularly suitable for measuring incongruence between matrices because it focuses on the strength of evidence. The test is implemented in a program called arn, which can handle both two-sample and multi-sample tests. The program has been applied in several studies, demonstrating its effectiveness in assessing incongruence between different types of data, such as morphological and molecular data. The discussion section addresses criticisms and clarifies the limitations of the test, emphasizing that it is effective in detecting incongruence between matrices, even when one matrix is much larger than the other. The authors also provide examples to illustrate the test's capabilities and limitations.