Evolutionary Operation: A Statistical Method for Process Improvement — Box & Draper, 1969

Evolutionary Operation: A Statistical Method for Process Improvement by George E. P. Box and Norman R. Draper is a landmark 1969 text that introduced EVOP — a practical, statistically grounded methodology for continuously improving industrial processes while they remain in full production operation.

The central idea of the book is elegantly simple yet mathematically rigorous: rather than shutting down a manufacturing process to run experiments, EVOP applies small, carefully designed perturbations to process variables during normal production runs. By systematically collecting and analyzing the resulting data using response surface methods and sequential experimental design, engineers can detect and exploit improvements in yield, quality, or efficiency without ever interrupting output. Box and Draper provide the full statistical framework for implementing EVOP, including the design of experimental patterns, the calculation of effects and error estimates, and the decision rules for moving the process toward optimum conditions.

George Box is one of the towering figures of 20th-century statistics. A student of Egon Pearson at University College London and later a colleague of Ronald Fisher, Box made foundational contributions to response surface methodology, time series analysis, Bayesian inference, and the design of experiments. His famous aphorism — "all models are wrong, but some are useful" — captures the pragmatic philosophy that animates this book. Published in 1969, Evolutionary Operation arrived at the height of the quality revolution in manufacturing and became an influential text in industrial statistics and quality engineering. Its ideas prefigure and complement the Total Quality Management and Six Sigma movements that would reshape global manufacturing in the following decades, making this first edition a historically significant document in the literature of applied statistics and industrial process improvement.