Serious errors in the statistical analysis conducted in a paper on the aurochs Bos primigenius (Lasota-Moskalewska and Kobryń 1990) are revealed in this article. A linear coding method transforming raw measurements into dimensionless numbers ranging from 0 to 100 was applied by the authors. Pooling the coded data on different traits is criticised and so is their application to any statistical tests. Correct tests are proposed.