BERNARD Bailyn taught Harvard's introductory graduate research seminar in Early American History for the last time in the fall term of academic year 1992–1993, at the close of which he retired in compliance with Harvard's then-mandatory policy. The following September he returned to teach an expanded version of the course, reworked as a general introduction to historical methods, under a new rubric. He would offer that seminar every fall term from 1993 through 1999, before he stopped (at age seventy-seven) to concentrate on administering and teaching The International Seminar on the History of the Atlantic World, the highly successful, influential post-doctoral summer institute that he launched in 1997 and superintended through 2010.Graduate courses in historical method had not customarily been taught at Harvard, where the two research seminars required of all first-year students had been thought a sufficient introduction to professional research and writing. These had varied greatly in approach and content, however, and when the English historian Mark Kishlansky became History's director of graduate studies, he proposed a common introductory methods seminar as part of a general reform in the curriculum. Such a first-semester experience would socialize incoming students to professional training, sharpen their research techniques and critical-thinking skills, and (not least) introduce them to one another across the boundaries that divided their intended fields of study.1 Such a course was formidable undertaking, but Bailyn's scholarly stature and well-known desire to remain engaged in graduate instruction made it “a no-brainer” for the department chair, William Kirby, “to ask him to teach after retirement.” Bailyn “seemed to relish” the challenge and a new course, History 2910, was born.2 The title Bailyn chose, The Practice of History, signaled that its focus would be, as The Oxford English Dictionary had it, on “the actual application … of an idea, belief, or method, as opposed to the theory or principles of it; performance, execution; achievement.”3As the practical, ephemeral documents that they are, syllabi are not prime candidates for archival preservation, though they often turn up at random in the course notes students retain. One such is the description of History 2910 as offered in Fall 1996, which survives in the papers of Professor Brian DeLay of the University of California, Berkeley, and is reproduced here. Although it is not, in itself, a particularly informative document, when examined closely in the context of notes that DeLay jotted down over the course of the semester, it reveals much about Bailyn's approach to seminar teaching and opens a window into his thinking about the craft and art of History.At the organizational meeting on September 16, 1996 (the first Monday of the term; the class met on Tuesday afternoons thereafter), Bailyn distributed the nearest thing the class would have to a syllabus: a list of assigned readings grouped under ten Roman numerals.4 He explained that he had arranged assignments thematically and that he wanted the students to approach the texts critically, deciding for themselves what those themes were. They were to look for the questions the authors posed and the arguments they made, seeking in each selection to answer the question, “How is the craft practiced [here]?”5 He then briefly outlined his expectations for the two course papers, stated their due dates (November 5 and January 2), announced his office hours, and no doubt asked the class members to introduce themselves to one another. Then he dismissed the twenty or so students to read the first assignment and prepare to discuss it on October 1.Bailyn had explained little about the design of the course and nothing at all about his assumptions concerning their preparation. The document he passed out was no help: it seems unlikely any of the students had ever encountered a sparer, more opaque, less revealing course description. Anyone not taking careful notes at the first meeting would have had nothing to remind him or her of the seminar's goals or content; no record of the instructor's office location, phone number, email address, office hours, or even the spelling of his name; no indication of the day and hour or location of the seminar's meetings; no specifications or due dates for its written work: nothing. A close examination of the readings, moreover, disclosed no obvious scheme. Within each week's assignments the selections were ordered neither alphabetically, nor chronologically, nor topically. History 2910 was a riddle, wrapped in a mystery.Two weeks later, when class convened for its first discussion, the students understood that Bailyn had further cloaked his intentions in an enigma. The 300 or so pages of readings grouped under Roman numeral I consisted of selections from four historical masterworks, but what possible connection was there between them? The selections from Fernand Braudel's great work, The Mediterranean and the Mediterranean World in the Age of Philip II, led the list and were presumably the first thing the students were to read, but what they were to deduce from a poetically-written preface, a vast table of contents, and an unspecified portion of each of the book's three parts was hard to imagine.6 Attentive readers, remembering from the first meeting that they were to decide what the theme of each week was, doubtless noted a resonance between the “world” of Braudel's title and the content of the other selections, and inferred that the first week's theme must have been the depiction of historical milieux. They were right in that; but the realization went only part way to encompass the intention behind the readings.Bailyn regarded three of the works assigned as well-wrought social history narratives. The first half of Peter Brown's synthesis, The World of Late Antiquity, offered a brilliant stratospheric overview of how Roman imperial unity fragmented over the course of the third century, then recovered cultural coherence over the next hundred years with the rise of Christianity as a mass religion, while the empire's center of gravity shifted east toward Constantinople.7 If Brown's work represented narrative on the grand scale, Emmanuel Le Roy Ladurie's technical analysis in Montaillou stood out as a feat of synecdoche, constructing a medieval milieu in miniature from evidence gathered by the heresy-hunting Bishop of Pamiers in the early fourteenth century: a portrait of the material, social, economic, and mental world of Occitan villagers built literally from the ground up, evoking the textures of daily life, lust, and spirituality with astonishing verisimilitude.8 David Cecil's highly-polished prologue to his life of the British prime minister William Lamb, second Viscount Melbourne (1779–1848), presented an example of an older, impressionistic mode of social history—a broad-stroke portrait of the life style of late eighteenth-century Whig aristocratic families as elegant as the Adam-style country houses that symbolized their dominance and displayed their taste.9By contrast, Bailyn regarded the Braudel selection, which he had trickily placed first on the list, as deeply problematic. As none of the students in the seminar could reasonably have known, nearly a half-century earlier he had argued that The Mediterranean was a fatally flawed work. Despite Braudel's ambition to depict the great sea and its littoral as a single, deeply-contextualized world, Bailyn had written, he had asked no unifying question; thus he failed to integrate as a single story the events described on the three functionally independent time-scales (geological and climatic, demographic, and political) he treated sequentially in the book's three parts.10 Behind the week's overt theme of historical worlds, therefore, lay a hidden message, which would be key to the students’ understanding of their task in every remaining meeting. Evaluating the assigned selections by comparison against a common standard (in the case of Session I, the degree to which each author achieved narrative unity) and then determining what qualities distinguished the successful from the less successful works, were the first steps to understanding the practice of history as Bailyn himself did.In the weeks that followed, the assigned works unfolded around themes instructive of some dimension of the historian's craft. Sessions II through V principally concerned literary issues, with each discussion challenging students to look critically at disparate texts drawn together around whatever themes they could discern. The overarching concern that connected these four sessions was the relationship between analytical historical writings—studies—and historical narratives—stories. The nearer authors came to unifying analysis and narrative, the closer they came to writing history as a form of literary art.Session II (October 7) dealt with contextualization and the attendant problems of selection and “normalization,” or the suspension of moral judgment, in fully-contextualized histories. As no student could possibly have known, the week's readings were all works that Bailyn had commented on, the previous year, in his Charles Joseph La Trobe Memorial Lecture in North American History, at La Trobe University in Melbourne.11 It was no accident that Herbert Butterfield's short volume, The Whig Interpretation of History (1931), was the first selection listed. Bailyn had made that “resonant,” “remarkable little book” the touchstone for his lecture, because in arguing for value-neutral history—a history written without foreshadowing the present or abridging complex and ambiguous developments by drawing straight lines from past conflicts to modern conditions and values—Butterfield “probed the core problems of contextualism more deeply than anyone had done before.” Fully-contextualized history, Bailyn had said in that lecture, is “the deepest history,” because it “penetrate[s] beyond manifest events into their … settings, into the substructures and surroundings from which they emerge.” It also poses “difficult and subtle” challenges for historians who must fully and sympathetically enter the minds and world-views of “people not only distant but alien from themselves,” in such a way as to retain “the capacity for moral judgments that do not warp the narrative and the conviction that change, growth, decline—evanescence—is what history is all about.”12A student who succeeded in extracting that insight from Butterfield's evocative, allusive, and sinuous prose would presumably have been able to see the intended connections between the texts that followed: A.J.P. Taylor's famously controversial argument that Germany annexed Austria in March 1938 not as a stage in Hitler's master-plan to create a Thousand-Year Reich but rather as an unintended consequence of how European powers practiced diplomacy in the aftermath of the Great War and the Great Depression;13 Brendan Bradshaw's furious denunciation of Butterfield as the inspirer of revisionist Irish historians in the 1980s who denied that the Ulster rising of 1641 anticipated a centuries-long Irish national liberation struggle against English dominion;14 and selections from one of the five massive volumes of Australian history that had been published for the nation's Bicentennial commemoration series, books marked by their refusal to offer any overarching narrative and instead to present history in “slices,” deep analytical descriptions of the continent at fifty-year intervals from 1788 to the modern era.15 The surviving evidence suggests that if such a student was present in the 1996 seminar, she or he sat mum in the discussion of October 7. Brian DeLay's class notes for the following meeting, on October 15, begin with Bailyn's summary of “conclusions from last discussion,” a strong indication that in his view the previous discussion had been less than conclusive.That summary, which must have taken at least ten minutes to deliver, suggested that the class had not grasped three crucial points from the readings for Session II. First, context is essential to all historical writing, but tricky: in its absence there is no way to interpret evidence without distortion, but if overdone contextualization can overstate elements of stability and rob narratives of dynamism. Second, while historical analysis necessarily requires the selection of evidence, to select based on one's knowledge of outcomes that no contemporary witness could have foreseen virtually guarantees anachronism. Only strict attention to context—what participants knew or believed—can counterbalance the historian's knowledge of how events worked out, a knowledge that no participant could have had, or acted upon. Finally, while contextualization is crucial, it is possible for historians to understand the motives of historical actors so thoroughly that they risk blinding themselves to—or worse, excusing—abhorrent beliefs and actions. In this way, the value-neutral history that contextualization promotes can drain narratives of their moral content, or even normalize such horrors as slavery or genocide.16The readings in Session III concerned objectivity, subjectivity, and the question of authorial stance. Bailyn believed deeply that any direct intrusion of an author's modern-day concerns into a depiction of the past carries the double risk of anachronism and self-referential indulgence, potentially distracting readers from the historical period and characters described in the work at hand. Because authors cannot divorce their writings from their own time and perspective, the first reading selection of the week was “Everyman His Own Historian,” Carl Becker's statement of the case for historical relativism. Bailyn had reservations about Becker's scholarship but not about this classic argument; in the La Trobe Lecture he embraced it, like virtually all modern historians—“none of whom … believe naively that historians can attain perfect objectivity … free from the prejudices, assumptions, and biases of one's own time, place, and personality.”17 Yet Bailyn also believed that relativism could be taken too far, and the point of Session III was to make clear that historians’ limited ability to know or describe wie es eigentlich gewesen ist was no license for subjectivity in their writings.Samuel Eliot Morison's biography of Christopher Columbus was a classic work, aspects of which Bailyn admired—he assigned it in Session X as a positive example of the recovery of the mental world of a medieval mariner—but in this set of readings it appeared as an example of self-indulgence, flawed by Morison's frequent, casual insertion of imagined dialogues and expostulations unattested by any contemporary source.18 By injecting direct discourse into the narrative, Bailyn believed, Morison had tried to communicate a sense of real life, but instead built “a severe anachronism” into his story, because imagined speech is necessarily “a reflection of the way [the historian might] … romantically think some people would have talked” in the past.19 Morison had failed to see that in spinning such yarns he put himself, not Columbus, at the center of the narrative.It is unclear if Bailyn found equal fault with John Demos for his more deliberate use of internal monologues, imagined scenes, and invented dialogue in passages carefully signaled by italicized text, in The Redeemed Captive: A Family Story from Early America.20 Demos constructed the elusive story of Eunice Williams's life—from her experience as an English puritan child taken captive in the Deerfield raid of February 29, 1704, through her conversion to Roman Catholicism and her transculturation as a Mohawk wife and mother living at Kahnawake near Montreal, to her continued efforts to sustain connections with her New England kin over more than seven decades—almost entirely from carefully contextualized evidentiary fragments. Bailyn called these inventions “subjective and self-referential” yet also made clear that he admired Demos's research and literary ambition. His final question to the class, “Is it romantic scholarship?”21 suggests that he did not see Demos's experiment in narrative as the kind of methodologically untethered, irresponsibly subjective post-modernism that Gertrude Himmelfarb derided in the concluding chapter of On Looking into the Abyss: Untimely Thoughts on Culture and Society, the fourth selection of the week. The burden of Himmelfarb's polemic, dripping with scorn, was that literary theorizing had spawned nihilist history: “a negation of the discipline” that was tantamount to “intellectual and moral suicide.”22Two decades earlier Bailyn had criticized historians of the Annales school for their devotion to literary theory—his 1977 review article, “French Historical Method,” had been positively scorching—but his position in Session III's discussion seems to have been more bemused than outraged, merely counseling the students to exercise caution and self-restraint.23 History and historical fiction alike could engage readers’ imaginations with the past, but anyone who hoped to write history responsibly needed to understand the difference between the two. Resisting the temptation to fictionalize was principally a matter of understanding the principles of good craftsmanship and remaining committed to practicing them; pace Himmelfarb, one did not need to believe that the fate of Western Civilization hung in the balance.Session IV introduced the narrative mode as an exercise in literary technique and structure in selections that moved from Antiquity to Whig political history in the nineteenth century, to presidential biography, to modern social and cultural history. Tacitus's account of Nero's debauched reign from the murder of his mother in the year 59 through his divorce of Octavia and his subsequent marriage to Poppea (62 CE), demonstrated the limitations of the Annals’ strict chronological discipline: all events appeared on an equal footing, without a plot to illuminate their interconnections or their significance.24 Macaulay's sweeping story of Monmouth's Rebellion solved the problems of chronology and significance by presenting events and characters in a plot structured like a novel's, working forward through time toward an inescapable end. Such moral teleology made it the very archetype of Whig history, which Butterfield had warned against.25 David McCullough's scrupulous chronicle of the first weeks of Truman's presidency, a whirlwind of events that carried the reader from Potsdam through Hiroshima, Nagasaki, and the surrender of Japan, took the opposite tack by telling the story in close-up views, as seen from Truman's perspective. Where Macaulay's plot implied inevitability, McCullough's emphasized contingency, making clear the minute-to-minute uncertainty of it all, solving the problem of significance by mapping events along the axis of Truman's growing confidence in his own judgment as he confronted crisis after crisis in a world transformed by war.26Against these three stories, structured by strict chronologies and political occurrences, the chapters selected from Religion and the Decline of Magic stood apart as an example of social and cultural history, the narrative character of which was less intuitively obvious.27 Bailyn's point, however, was that Keith Thomas's great work did indeed tell a story, and in so doing demonstrated how historical groups and cultures can be both rigorously analyzed and compellingly described, even in the absence of conventionally recognizable events. Thomas's success required recognition of the long coexistence of formal religion with popular magic and intellectual occultism, and subtle attention to the external causes that contributed to the gradual waning of magical practices, building a narrative with few milestone moments, many qualifications, and an ambiguous ending that probably came long after Thomas himself thought.28Religion and the Decline of Magic posed challenges at the conceptual level both because it dealt with changes in mentalité of which contemporary witnesses were unaware, and because its central question—Why did magic decline?—did not yield an answer as clearly falsifiable by the evidence as the questions asked in constructing political, diplomatic, military, and similarly event-driven narratives.Whether or not the class members understood it at the time, ultimately Bailyn intended in Session IV to introduce them to history as a form of literary art. As he pointed out in concluding comments, the lives of individual human beings, and of the societies and polities that they comprise, are not lived as stories but only understood as such in retrospect, when the surviving evidence can be arranged in sequential form. Only then can the actions human beings took and the consequences that followed from them—as well as from the accidents and unanticipated results that also influenced outcomes—be understood in terms of events with causes and consequences, and described as stories with coherent meanings. To qualify as history such stories must be as falsifiable as any formal argument, which is to say that they should be rigorously measured against the evidence for accuracy, consistency, and coherence. Insofar as these narratives are products of the human imagination, they are creative representations of the past; if written with sufficient verisimilitude and evocativeness, they can even be understood as works of art. Close attention to writing, therefore, was a matter of supreme importance.29Session V, building on the examples of narratives written in various modes, considered the importance of prose style—specifically, of authorial voice—in conveying arguments or stories in such a way as to engage the intended audience. The message here was clear: to communicate contingency, character, action, argument, and meaning authors must attend to the connections between substance, style, and tone, not merely in an abstract way, but as a practical means of meeting readers’ expectations and needs. That recognizing these connections was a crucial skill in interpreting primary-source narratives was the point of the first excerpt, from Lord Denning's Report, the official account of Parliament's inquiry into the 1963 Profumo Scandal. Its overall tone was dry, spare, and legalistic: the voice of an eminent judge (Denning was master of the rolls in the Court of Appeal) who had been summoned by his sovereign to inform the nation's lawmakers on a matter of great national concern. Against this sober backdrop the salacious modifiers that Denning consistently used to describe the drug use and sexual activities of Christine Keeler and Stephen Ward stood out vividly, leaving no doubt of Denning's own views of who were the most culpable figures in this sordid episode.30Hugh Trevor-Roper and John Clive attended to the expectations of their intended audiences with equal care, and Bailyn evidently intended the selections from these accomplished stylists to be read as a matched set. Both wrote with a notably light touch. The insider humor and elaborately mannered tone of Trevor-Roper's Inaugural Lecture as Regius Professor to a ceremonial assemblage of Oxford dons, like Clive's witty, relaxed, conversational discussion of the literary styles of Macaulay, Gibbon, and other eminent narrators for the amusement and delight of educated readers, showed pitch-perfect attention to their audiences’ expectations.31 Finally, a chapter from Religion and the Rise of Capitalism demonstrated how well-chosen metaphors could reinforce a subtle argument—at least for readers familiar enough with the Authorized Version of the Bible to hear the resonances—in R.H. Tawney's invocation of scriptural references and images as he teased out the ironies of puritan divines’ writings on commerce and social ethics.32The message of Session V was ultimately both simple and demanding. Effective historical prose required the writer to match form and content while writing in a way that meshed with readers’ expectations, understanding, and taste. An engaging style obviously demanded clarity, but that was only the beginning. Authorial voice was a matter of careful contrivance—artfulness—practiced in the interest of making the writer's work endure.Session V may have been a little more tense than the previous ones because on that day, November 5th, students handed in their first papers. The assignment, which Bailyn had given them orally the week before, was to write a ten-page essay reflecting their “view or sense of the topics,” and especially of “the questions and issues” the seminar had dealt with to date. Beyond that notably vague prescription, he had stipulated only that the paper must “be [a] structured, organized essay”—presumably, one with an argument—and that the writer should do no research beyond the sources already read. Footnotes were not required.33Did the students who confronted this task understand that the themes of the first five weeks—comparison as a fundamental critical technique, the need for careful attention to context, the tension between relativism and objectivity, the varieties of narrative, and the significance of style—were the essential components of historical thinking, conceptualization, and writing as Bailyn understood them? Perhaps. At the least they must have sensed that something significant was intended because they had reached the midpoint of a course about whose second half Bailyn had given no hint. Only in retrospect, much later, would it be possible for them to understand that producing those papers anticipated, in miniature, the writing of a dissertation prospectus: to take their present, imperfect knowledge of the thicket they had been hacking their way through for weeks and use it to make an educated guess about the shadowy woods that stretched beyond, a forest the dimensions of which they had only begun to fathom.The selections for Sessions VI through X centered on works that incorporated recent advances in quantitative and interdisciplinary history, in the expectation that the students would evaluate them according to the criteria of problem definition, argument, narration, and style they had come to understand in the first half of the course. Session VI examined the ways in which a powerful medium of ethnographic analysis—Geertzian thick description—could offer striking insights when applied to the interpretation of modern cultures but prove problematic as a mode of explaining past cultures. In large part this simply was because historians cannot directly observe their subjects’ interactions or interview them as witnesses. In the absence of such investigative tools, speculative interpretations can multiply without limit.34 Irresolvable debates result—as shown by divergent attempts to explain the death of Captain James Cook in 1779—where it is impossible to formulate a falsifiable historical question.35In Bailyn's view, Rhys Isaac's celebrated work, The Transformation of Virginia, 1740–1790, embodied several intractable problems of the ethnographic approach to history. Chief among these was that Isaac had structured his account around set-piece explications that anticipated the episodic slices of Australian history the class had read for Session II. This produced not a unified narrative but a sequential set of static studies, which in Isaac's case were further constrained by the elite provenance of his sources: an account that failed to illuminate the dynamic interactions of religion, class, and political culture in the Revolutionary-era Chesapeake.36 Bailyn evidently found the problem of stasis particularly acute when writers attempted to apply thick description to closely focused historical cases, a common feature of microhistorical studies where issues of larger significance remained obscure.37The formulation of historical problems of a different order was the topic of Session VII, which dealt with the origins of chattel slavery in North America within the wider context of the early modern Atlantic world. This week's assignments demonstrated the narrative carrying capacity of social history and the usefulness of ethnographic concepts when carefully applied. Here the centerpiece was the contrast between two groundbreaking mid-century works, Frank Tannenbaum's Slave and Citizen and Oscar and Mary Handlin's William and Mary Quarterly article, “Origins of the Southern Labor System.”38 Bailyn clearly intended to emphasize how, in the absence of a focusing question and close attention to chronology, Tannenbaum's attempt to investigate global practices of slavery in the Early Modern period fell apart in taxonomic confusion and inconclusive comparisons. In contrast, the Handlins’ incisive analysis of anomalies in the evolving meaning of the words servant and slave allowed them to describe the emergence of chattel slavery and the hardening of racism in the North American colonies with temporal precision, conceptual sophistication, and scrupulous attention to change. The progress of the field over the next generation, and most of all the benefit of examining shifts in the institutional and cultural character of slavery and its regional variations, was the point of the week's third selection. Ira Berlin's 1980 article, “Time, Space, and the Evolution of Afro-American Society in British Mainland North America,” was notable for its careful use of the ethnographic concept of creolization, not as a taxonomic label but rather as an analytic tool, capable of disclosing dynamic changes through time.39The five readings of Session VIII were all either pioneering exercises in, or reflections upon, quantitative analysis in history. Four articles dated from 1965 or 1966 and one from 1979; between them they reflected two sharply differing approaches to quantification. The point of the session was to show how very different those ap