The History of Statistics: The Measurement of Uncertainty before 1900 by Stephen M. Stigler

Abstract

TECHNOLOGY AND CULTURE Book Reviews 299 The second half of the book, on Marx’s salvation “story” told using an underlying organic metaphor, is better handled—though I believe the machine metaphor was never far away, as shown by Wessell’s constant reference to “structure,” “base,” and cause regarding Marx. The same mythic insight seems to have influenced Weber, again pro­ posing a theory of enablement. Ifonly Marx’s “workmen” would throw over the bosses’ yolk, if only everyone would act “rationally,” a New Jerusalem would appear on earth. . . . Or so the “theory” indicates. Wessell has missed a trick over the ancient Greeks to fail to note that two others of his principal topics are related. By going back to basics (Greek: supporting foundations), not only can mythos and Logos be seen to be alike in ways, so can “theory” and “theater”: each stems from the Greek theasthai, to view. Thus Aeschylus may have used poetry (something made) and mimesis to tell his “stories,” and he may have tried to grasp the sym-“pathy” of playgoers, whereas scholars use other devices to articulate their themes. But—all produce scripts; all put on a type of show for others; all try to reduce their messages to some type of “truth”; sight has always been the dominant human sense involved (scientists “observe” things). But Marx claimed to compose wissenschaftlich accounts. This means his work should be assessed not as stage drama, but rather against the evidence available; and, by this criterion, in my judgment, his characterization of modern industry fails. The particular mode of failure concerns the “audience”-substitute in scholarship. “The audience was wonderful,” thespians often claim in veiled self­ esteem. Without those human beings (those who go to “see a play”), no stage drama will generate a sense of pathos; so its theatrical rep­ resentation, its attempt at successful mimesis, cannot work. Marx has had his thousands of discussants for his tale of dramatic conflict (p. 40), his Tale of Two Cities over time. Some have cheered and some have booed. As scholarship, however, Capital makes good show biz; and Wessell’s book helps to show why. Michael Fores Mr. Fores was with the Department of Industry in London and has written journal articles on various sightings of Industrial Revolution. The History of Statistics: The Measurement of Uncertainty before 1900. By Stephen M. Stigler. Cambridge, Mass.: Belknap Press of Harvard University Press, 1986. Pp. xvi + 410; illustrations, figures, tables, appendixes, bibliography, index. $25.00. This is an important book. It is the first modern general history of statistical theory. It begins with the first uses of mathematics and 300 Book Reviews TECHNOLOGY AND CULTURE probability theory in analyzing the combination of observations in astronomy and geodesy and traces the evolution and eventual syn­ thesis of the methods of “least squares” and “inverse probability.” It goes on to describe the partially successful 19th-century efforts to extend those methods into the social sciences. It concludes with the late-19th-century rise of the “British school” of statistical theory— men such as Francis Galton, Francis Ysidro Edgeworth, Karl Pearson, and George Udny Yule, who decisively extended “the measurement of uncertainty,” especially in studies of heredity. The book ends with Yule’s first recognizably modern social science regression analysis in the closing years of the 19th century. A constant theme is the growing sense, still incomplete by 1900, of the generality of statistical theory. A single statistical technique—least squares, for instance—can find uses in fields as different as studies of the orbits of comets, analyses of the causes of pauperism, and (to take a post-1900, technological example) missile guidance. Our contemporary understanding of that generality carries with it risks for the historian, especially one who, like Stephen Stigler, is also a practicing statistician. It is too easy to see the past as the gradual “discovery” of that general order, and to treat the specific concerns of the forerunners of modern statistics as merely “context”—inter­ esting but incidental backdrop. Stigler is largely successful in avoiding this Whiggish teleology. Even where he is not, his superbly crafted book is informative. It is also clear, lively, and readable, despite the difficulties of...