A simple technique is proposed for the description and analysis of the patterns of error customarily found in tasks requiring subjects to interpolate between explicit or implicit reference values. The error patterns can be characterized by two parameters. One parameter reflects any consistent tendency to make positive or negative errors overall while the other reflects the difference in error between the two halves of the interpolation interval. The style of analysis advocated is quantitatively descriptive of the error patterns and leads to a simplification of subsequent statistical analysis of the results.