A revised version of this review appeared in the Journal of Economic History, 63, 2, June 2003. That journal is the copyright holder of record.

 

 

 

Statistics on the Table: The History of Statistical Concepts and Methods. By Stephen M. Stigler. Cambridge: Harvard University Press, 1999.  Pp. ix, 488. $48.95, cloth; $19.95, paper.

 

            This volume serves up a buffet of delectable and nourishing essays in the history of statistics.  Stigler, perhaps the most eminent modern historian of statistics, has here collected twenty-two previously published papers, revised the essays to stand in more nearly appropriate relationships one to another, and defined his historical concerns in a fine introduction.

 

            From the essay that gives the volume its title, a wonderful piece on Karl Pearson and his controversies with Cambridge economists on the use of statistical evidence, to the often quoted essay on Stigler’s Law of Eponymy (“no scientific discovery is ever named after its original discoverer”), the book is historically sophisticated scholarship.  Early papers on Jevons and Edgeworth present the emergent interconnection of the social sciences with data based argumentation, while the five chapters on Galtonian ideas take us through such issues as fingerprints and regression toward the mean.  For instance, Stigler’s Chapter 8 is a tightly reasoned study to defend the proposition that mathematical statistics as a discipline “began” in 1933, and this defense engages arguments about dating in the history of science more generally.

 

            One of my own favorites became Chapter 11, “Apollo Mathematicus”, which begins like Philip David’s The Thread (Boston: Birkhauser, 1983) with the author, Stigler, perusing the catalog from an Oxfordshire dealer in antiquarian books.  From that beginning, through the book of the chapter’s title on mathematical medicine printed in 1695, we have a rollicking tale of Harvey’s methodology to convince the skeptical about the circulation of blood, Pitcairn’s sociology of invention, mathematical theories of secretion, and the cure of fevers. These diverse elements are all finely knit together by a reexamination of the similarities between this complex episode in which mathematics moves into a previously unmathematical area, and all kinds of modern complaints about the problems of mathematizing field x or subdiscipline y.

 

Part 4, on questions of discovery, take up what would appear to be traditional questions like the origins of the ideas associated with Bayes Theorem, maximum likelihood estimation, least squares, and degrees of freedom.  For instance, Chapter 17 on “Gauss and the Invention of Least Squares” revisits, using two new pieces of evidence he develops, earlier studies sorting out the set of claims and counter-claims for priority – among the champions of the Frenchman Legendre, the German Gauss, and the American Adrain -- in the invention of the method of least squares. The issues of discovery, and utilization, of new ideas are treated with sophistication. Indeed, these essays are uniformly informed by a wonderful sensibility: the author has an eye for nuance, irony, and humor, and the wit and energy to see matters through.  That Stigler writes so clearly and brings the reader in so easily makes these essays quite special. 

 

The final Part 5 collects studies about standards.  My favorite here is “The Trial of the Pyx”, which is the name of the British Royal Mint ceremony “to ascertain that the coinage issued by the Mint met the Crown’s specifications” (p.383).  Of course, such an event is a sampling procedure, and the complex ceremony is examined from the perspective of whether its arcane processes are coherent from a modern perspective – does the Trial make statistical sense? Since Newton, among other noteworthy figures, was Master of the Mint, there is a great deal of lovely material that can be developed in telling the story.

 

            Since the various pieces originally appeared in such widely disparate places, Harvard University Press is to be congratulated for putting this special volume together and making these papers available to the general reader.  Anyone who works in an area in which statistics plays an important role will be entertained and delighted by this volume. 

 

 

 

E. Roy Weintraub

Duke University