For much of the first half of the nineteenth century, the Western River steamboat was the dominant supplier of long-distance transportation services in the Midwest, plying a network of perhaps 10,000 miles of navigable rivers. Moreover, it was a glamorous activity memorialized in the writings of such literary luminaries as Mark Twain and Charles Dickens. It was also an extraordinarily dangerous activity—at least by modern standards. According to one contemporary observer, over 35 percent of the steamboats built before 1849 were lost in accidents (these figures represent corrections by Louis Hunter, Steamboats on the Western Rivers, Cambridge, MA: Harvard University Press, 1949: 272, fn. 2 to the original estimates). Although steamboats occasionally collided, were destroyed by tornadoes, burned to the waterline thanks to a risky combination of chimney sparks and combustible deck cargo, or blew up as a result of bursting boilers, most of these losses were a result of snags, shifting sandbars, and other natural obstructions to passage along the rivers (Hunter, Steamboats, p. 272). Moreover, these losses took a significant toll on steamboat passengers, not to mention crew. Indeed, a contemporary observer reviewing the losses remarked (without a trace of irony!): "we suppose there is not one in one thousand (italics in original) who is either killed or wounded in travelling one thousand miles on our western waters" (Senate Executive Document No. 42, Serial Set 619, v 8, 32 Cong. 1 sess. "Report of the Secretary of the Treasury…" January 21, 1852: 115).

According to the Preface, Professor Paul Paskoff, Professor of History at Louisiana State University, originally intended that Troubled Waters "establish quantitatively the extent of the federal government's failure to make the country's rivers safe for steamboat navigation" (xv). As the quote from the 1852 Secretary of the Treasury's Report suggests, there should be ample evidence in support of this hypothesis. However, what finally emerged is far from that goal. Instead, the author claims that his "analysis indicated a federal program that had largely succeeded in making steamboat navigation safer by removing natural hazards on the Mississippi and its major tributaries" (xv), although the implicit counterfactual is nowhere explored.

The book is divided into two more or less equal, but quite distinct, parts. After an introductory chapter, the first part deals with the politics (but not the political economy) of congressional appropriations for river improvement. Someone desiring to follow up on the political economy of these appropriations will, however, find some of the necessary data on voting behavior, etc. in appendix D. Discussion of the politics here is divided between pre-Jacksonian and post-Jacksonian politics where "Jacksonian" is broadly defined to include the "Young Hickory"—President James K. Polk. Jackson, of course, was famous (or infamous) for his 1830 veto of the Maysville Road, which many scholars have taken as the seminal event limiting federal underwriting of internal improvements and handicapping Henry Clay's "American System." Paskoff, however, convincingly argues to the contrary that federal funding of internal improvement not only continued after the Maysville veto but actually grew. Rather it fell to Jackson's disciple, Polk, to shutdown federal sponsorship of internal improvements.

The second part of the book focuses upon the mechanics of rivers improvements such as snag removal boats and, more importantly, the adaptation and evolution of steamboat design to meet the challenges of the rivers. It is here that I think the author misses his biggest opportunity. Central to this part of the book is a database that Paskoff has assembled primarily from Frederick Way's Way's Packet Directory, 1848–1994. Athens, OH: Ohio University Press, 1994 (a packet was a steamboat that carried both passengers and freight) and Bruce Berman's Encyclopedia of American Shipwrecks (Boston, 1972). The database, described somewhat sketchily in appendix A, has information on where and how each steamboats met their end. It also typically details when and where they were built, their hull dimensions, mode of propulsion, the number of boilers, and other characteristics of their engines as well as the trade on which they were engaged and the names of key ship's officers. In short, the database contains a wealth of quantitative as well as qualitative information and I would strongly encourage the author to deposit this valuable resource with an agency such as the ICPSR so it will be preserved and available to the scholarly public. Valuable as this database potentially is however, it would be a mistake to view it as complete even within the limitations implied by the underlying sources. Some steamboats—for example, the James Hewitt and the Peytona—which are in my sample (part of which appears in Eric Haites, James Mak, and Gary M. Walton, Western River Transportation, The Johns Hopkins University Press, 1975) do not appear in Way's Directory. At first, I took this to be because both were built before 1848, the initial date listed in Way's Directory, but other vessels such as the Chancellor and the Glencoe (both built in 1846) do appear in Way's Directory and all appear in Lytle's List (William M. Lytle, Merchant Steam Vessels of the United States, 1807–1868. "The Lytle list," compiled from official merchant marine documents of the United States and other sources. Mystic, CT: Steamship Historical Society of America, 1952). Moreover, there are post-1848 steamboats in my sample, such as the Anne Livingston and the Huntress, which also do not appear in Way's Directory. Data, of course, also came from other sources, but these are much less complete than Way's Directory.

There must therefore be a question mark about the comprehensiveness and representativeness of the data underlying the analysis. The analysis of these data, such as it is, appears in the two central chapters to part 2: "Factors of Destruction" and "The Success of Public Policy." Each contains 17 figures describing various aspects of the data. Some of these simply chart averages, often by size of boat, over time. Others present scattergrams, often accompanied by a simple OLS regression line and a correlation coefficient, but without any statistics such as a standard errors being reported. There is no formal modeling of relationships or an effort to go beyond a simple regression controlling for other relevant dimensions of the problem. This limits the usefulness of the results and the author's ability to, say, apportion changes in loss rates between the various sources. As a result, I think the author missed a real opportunity to build upon the wealth of testable hypotheses to be found in Hunter's book which launched several professional careers.