You are here

The Lady Tasting Tea: How Statistics Revolutionized Science in the Twentieth Century

David Salsburg
Owl Books
Publication Date: 
Number of Pages: 
BLL Rating: 

The Basic Library List Committee considers this book essential for undergraduate mathematics libraries.

[Reviewed by
Marc Mehlman
, on

The title of David Salsburg's book refers to a lady who claimed that tea tasted differently depending upon whether milk or tea was introduced to the teacup first. The lady's claim was disputed. However, Ronald Fisher (later to become Sir Ronald Fisher) devised a test to determine the likelihood of the lady's claim being correct. I leave the outcome of this test to those who will read this book. It is definitely worth reading.

The Lady Tasting Tea appeals to both the professional and layman. While some knowledge of statistics is helpful, it is not necessary for an informed reading of this book. Statistical milestones are examined from both a historical and intellectual perspective, in a manner that the general public will understand. Also included are the personal details, opinions, life views and circumstances of many of the more famous statisticians.

Salsburg believes that the public is not fully aware of the degree to which recent developments in statistics impact the way we perceive the world. He correctly points out that the twentieth century saw the fading of a deterministic outlook and the rise of a statistical/probabilistic way of looking at the world. This ongoing revolution is not only in the physical sciences, it also touches the social sciences and even the humanities. Though profound, it is a quiet revolution that has been unnoticed by many.

The book consists of a series of often interlocking histories of superior minds grappling with applied problems. Often one finds that behind an eloquent solution to a historical problem is a hidden, convoluted history of various statisticians disagreeing about proper methods, followed by new inputs from fresh minds and, finally, with the help of time, a rough consensus among the experts. Salsburg shows that this processes is as often driven by the personalities and dogmas of the individuals involved as by their intellectual abilities. One such example of such a process is referred to in the title of the book. This is Salsburg's introduction to hypothesis testing (along with an interesting dispute between Fisher and Neyman). The book is easy to read since it takes a look at various threads of statistical history one at a time.

Salsburg covers the interesting history of Karl Pearson, Sir Francis Galton, William Gosset, Sir Ronald Fisher and Jerzy Neyman. This is, of course, mandatory in any statistical history. Particularly nice is Salsburg's presentation of the history behind the theory of distributions of extremes, illustrating the contributions of Leonard Tippet, Ronald Fisher, and Emil Gumbel. The Soviet school of mathematics and statistics is covered with interesting sections on Arnold Kolmogorov (axioms of probability) and Chester Bliss (probit analysis). Nor are the tremendous contributions of Indian statisticians ignored. Touching on contributions from America to Tasmania, the history of nonparametric statistics is presented. Other streams of statistics are also covered.

Salsburg places each contribution/contributor in the context of its/his times and situation. Salsburg notes that the anti-intellectual atmospheres inside the Nazi and Soviet regimes had a stifling effect on the quest for statistical understanding. In the case of Germany, many of her greatest minds were forced to flee the anti-Semitic terror. In the Soviet Union, communist dogma prevented the brilliant work of many of its scientists from being fully appreciated. The corporate culture of the 1950's caused W. E. Deming's ideas on quality control to find more fertile ground in Japan than in the United States.

In a relatively short book like this the author must make choices as to whom to include and whom to emphasize. Indeed, Salsburg discusses the rationale for his choices in an appendix, predicating inclusion on their relevance to the various historical movements in statistics. With no agenda other than this, Salsburg arrives at a collection of people quite diverse in their worldly outlook, origins and nationalities. Perhaps the only exception to his "rule of inclusion" is found in the special effort Salsburg makes to illustrate the contribution of women statisticians, a contribution that Salsburg believes deserves more attention.

This book is not an encyclopedic history of statistics. Neither is this book a dry scholarly tome. Salsburg has chosen to give voice to his own opinions and theories concerning many scientists in his book along with reports of their times and surroundings. The book does not censor the personal side of its protagonists, nor does it hide disagreements between many great men and women who also sometimes possessed great egos. One learns details such as the circumstances of A. Kolmogorov's birth, or why Karl Pearson changed the spelling of his first name from Carl (in honor of Karl Marx). Such details, often well known by word of mouth to those in the field, add much to the book. Naturally, not everyone will agree with everything Salsburg presents. For instance, while Salsburg points out that statistics has been a fertile birthing place for many wonderful mathematical insights and theories, one senses that Salsburg has little use for mathematics, even if it is beautiful mathematics, if it has no direct application to reality. Salsburg is no Platonist in awe of an independent, idealized truth. Truth to him is in what statistics does and describes, in the muck of empiricism, not in the remote, abstract mathematics behind the statistics. Whatever the reader's ultimate position on this issue, he will benefit greatly from the insights and opinions that Salsburg shares.

The Lady Tasting Tea laudably is both a thoroughly enjoyable read and a text that one can learn a lot from.









Marc Mehlman ( is Associate Professor of Mathematics at the University of New Haven. His interests are in probability and statistics.

The table of contents is not available.