Comments on methods and results in: Sherman et al., "Segregation analysis of balanced pericentric inversions in pedigree data".

The analysis by Sherman et al. (1986), its basis and results have been examined. The analysis relies on general methods, which may give acceptable results under the special conditions considered by the authors, but will usually produce more or less misleading results. The program POINTER (Lalouel & Morton 1981) is shown to be based partly on a chain of irrelevant arguments for the actual context. The so-called "conventional ascertainment rules" (Morton et al. 1983) are shown to produce misleading results in cases where their assumptions are not satisfied. The mean risk for unbalanced offspring is underestimated because of an erroneous ascertainment correction. The segregation frequencies are found to be different in three national samples, contrary to the claim by Sherman et al. Only a small proportion of all information available in the data has been utilized. Alternative and more appropriate models, hypotheses and procedures have been suggested. The frequent use of packages with computer programmes of standard statistical procedures in nonstandard situations with data from collaborative studies in human cytogenetics is discussed.