This presidential address was delivered at the 98th annual meeting of the American Historical Association, held in San Francisco, December 28–30, 1983. Published in the American Historical Review 89, no. 1 (February 1984): 1–9.

Depth, Span, and Relevance

The discipline of history has broadened enormously in the postwar decades, but historians have not. We teach the history of Africa and Asia, but specialists in American history know no more about the history of Africa than their predecessors did in 1940. We have specialists in black history, women’s history, and historical demography, but people outside these specialties pay little attention to their work. Where the field of history grew broader and richer, the training of historians grew narrower. The proportion of new Ph.D.’s that can easily teach the standard “Western Civ” course is smaller in the 1980s than it was in the 1950s. The new Asianists and Africanists know next to nothing about European or American history. Americanists know less European history than they did thirty years ago. At the level of course offerings, the old surveys in European and American history lost popularity. Departments offered a greater number of specialized courses, while history enrollments declined over all.

In recent years, the idea of a broad survey has begun to recover—with some uncertainty about what it ought to be about. A rejuvenated Western Civ is one possibility. A new world history with real concern for the history of non-Western world is another. Ironically, one reason the world history movement has not gained momentum is that few historians have the background necessary to take it on. Many world history courses are “team taught,” a reasonable solution to begin with. But the very fact that they are is a sad admission. Even those who have recognized the need have also realized how few of their colleagues have the breadth of knowledge required—even for an introductory undergraduate course.

Nor is our failure to help graduate students gain a world-historical perspective just of concern to the history departments that train them. What we teach passes to a broader public, and members of that public make political decisions that are crucial for us all. From the heights of power in the White House, we find portrayed a simplistic, tripartite division of the world into ourselves, our enemies, and the rest—who do not count, even though they form the vast majority of the world’s population. Historians did not do this all by themselves, of course; the rest of the educational system carries as much responsibility. Nor is everyone in the federal government as badly informed as Mr. Reagan’s circle of advisers. But, if one of our responsibilities as historians is to explain how the world came to be as it is, either our answers are not very good, or they are not communicated to the national leadership. In fact, both are the case. The government has enormous resources in short-term intelligence data, but the national leadership lacks long-term understanding of historical change. Without that, its evaluation of the short-term evidence has the fallacious quality we see week in and week out. Perhaps this problem is no worse than it was when an earlier generation of leaders led us into the Vietnam war, but official American reactions to affairs in Central America and talk about preparing to “prevail” in a nuclear war suggest deterioration at the top levels.

In one sense, all this criticism amounts to is the statement that a liberal education is better than ignorance. Historians have been on the right side of that one all along. Their professional failing has been something else—to forget that one of the prime values of a liberal education is breadth, not narrow specialization. Even before the explosion of new kinds of historical knowledge, historical competence required a balance between deep mastery of a particular field and a span of knowledge over other fields of history. Depth was necessary to discover and validate the evidence. Span was necessary to know what kind of evidence to look for—and to make some sense of it, once discovered.

We find the tendency toward specialization not only in graduate requirements and course offerings but also in our professional associations. We have an American Historical Association, and the Association also has seventy-five affiliated societies, each with a special concern with some particular field of history. Many other professional associations, like the Medieval Academy of America and the Organization of American Historians, are not even affiliated, although we work together in many different ways. The clear fact seems to be that many, perhaps most, historians value their personal contacts with fellow-specialists more than they value their contacts with historians in general. This is somewhat understandable. Historians can talk with many different specialists in their own schools, colleges, and universities. But to find someone in precisely the same subfield or specialty, they have to look elsewhere. A common complaint about the annual meeting of the AHA is that too few sessions are concerned with these particular research interests. The dissatisfied drift away from the AHA toward more specialized associations. This tendency is strongest outside old-line, Western history. Most Latin Americanists, Africanists, and Asianists have stronger ties to their own subfields than they have to history as a whole. This in turn makes it hard for a program committee to organize panels in these fields, thus increasing the sense of alienation.1 Many Asianists already regard the AHA as just another specialized organization for European and American history.

In time, these trends could be serious for the Association, but they are far more serious for the profession. They are the expression, not the cause, of intellectual splintering that has been going on for decades. The old-time, main-line fields like European and American history have been, to put it mildly, unenthusiastic about the growth of non-Western history. The Americanists were—and are—the most parochial, almost by definition, because they study the past of their own society. But Asianists and Africanists have often reacted by cultivating their own kind of parochialism—sometimes seeking refuge in area-studies programs. These, in turn, have been far more successful in opening interdisciplinary communication within the area than they have been in keeping open interarea communication within history.

But the fundamental problem is still overspecialization. It is just as ubiquitous and just as deplorable in almost any other field of knowledge. But what to do about it? One way to begin looking for answers is to step back and consider what the study of history might be expected to accomplish. Every historian will have his own answer, but many of these answers can be grouped roughly under three headings—in rising order of general importance. Each heading can be set off as a question: (1) How did we come to be as we are? (2) How did the world come to be as it is? And (3) how and why do human societies change over time?

The “we” in the first of these questions can be somewhat variable, from family history on to local history and on up to national history or even the history of Western civilization, which remains ethnocentric despite its broad scale. History, considered as one’s own past, is both the most common and the most problematic approach. Self-knowledge is no doubt a good thing, but self-knowledge by itself is also a form of selfishness that can be dangerous to social health. In the nineteenth century and too far into the twentieth, history was consciously one-sided; it was not supposed to be even-handed but designed instead to promote patriotism and glorify the nation. These tendencies reached a kind of apogee with the overblown patriotic fervor of the First World War. They have declined somewhat since, but only slowly and not in all parts of the world.

I have already suggested the second category under the question, How did the world come to be as it is? This, too, is a form of self-knowledge, useful information to guide all kinds of decisions that have to be made by people both inside and outside the circles of power. This kind of history is just now beginning to challenge the more self-centered variety for a place in high school and college curricula. But the contest is no longer a simple choice between the traditional Western Civ and another, broader survey of world history. It involves complex problems at several levels in the educational process. One problem is what to teach students who will take only one or two history courses, at most. A second is what to offer to undergraduate history majors who will not become professional historians. A third is what span of knowledge to demand of graduate students in addition to the depth of specialized knowledge they will need for their research.

The choice of formats for historical study is pretty much what it has been for some time: survey courses like Western Civ or U.S. history, backed by a second level of more specialized courses set in the familiar time-span segments—Germany from Barbarosa to Bismarck, the American South from 1860 to 1876, the history of China to 1910. Off in the wings, a potential world history survey is ready to compete for the present position of Western Civ, but that competition is still uncertain. Meanwhile, we continue with time-span segments. We have to trust that students will be able to put them together into some kind of synthesis. That may sometimes happen, but it leaves the burden with the student. It may be a useful challenge, but only for students capable of creative learning—and most are not.

Perhaps we, the professional historians, should attempt to meet the challenge by trying to create our own syntheses of historical knowledge. For myself, I doubt that a world-history survey course will be even as satisfactory as Western Civ was in its day. What we need is a new kind of course that will have the perspective of world history—will ask and try to answer the question, How did the world come to be as it is?—by treating topics selectively, with examples detailed enough to be comprehensible, rather than by surveying the entire panorama too superficially to be worth remembering. Different teachers will no doubt see the world from different points of view, but the important thing is to seek a genuine world perspective. Historians have not yet tried this approach very much, but Eric Wolf’s Europe and the People without History (1982) can stand as a useful example, by an anthropologist, of the kind of broad synthesis we should be doing ourselves.

If we move in some such direction, history may well appear relevant again to far more students than we now reach. But this will not be easy, and our graduate education is largely to blame. We train people to do research on narrow subjects. We make them acquire the kind of background knowledge that appears necessary for such research. They emerge with a Ph.D. and teaching competence adequate to a narrow time‑space framework. With a good deal of self-education, most will work their way outward enough to teach Western Civ or the American survey. Graduate schools, with very few exceptions, do nothing at all to prepare history teachers to handle the kind of courses students need in order to understand the world they live in.

Which brings me to the third kind of history—the kind that asks how and why human societies change over time. The main difference between this question and the second, on world-centered history, is that the answers here require an understanding of human beings in general. But human beings in general are too amorphous to be investigated directly. Obviously, we have to begin with some part of the whole—but what part? This question is nearly identical to the one Arnold Toynbee posed a half-century ago in the first volume of his A Study of History: What is the correct, objective “intelligible field of historical study?” His answer, of course, was that it could not be the European nation-state, the framework historians used far too commonly in his time, as in ours. He opted instead for units he called societies, of which the familiar Western civilization is the clearest example.2 He was right in wanting some field of study other than the nation-state, but his identification of “civilizations” as the prime actors on the historical stage led to new problems. Civilizations are hard to identify. Their borders in time and space are shifting and uncertain, and we lack clear criteria to demarcate one from another. Toynbee’s own choice of religion as the marker of civilization was, moreover, not universally accepted—or acceptable.

Rather than seeking a single “intelligible field of historical study,” we need, more prudently, to go part way along the road Toynbee mapped out, without stopping precisely where he did. Toynbee believed that each configuration of historical events can be separated fairly clearly in time and space from other, different configurations. And he dealt at length with the limits of his civilizations in space and in time. We can each go through the same exercise and reach different conclusions by centering on a problem to be solved. Just as a “civilization” has its limits in time and space, so too each historical problem has a universe of data necessary to its solution.

Social scientists sometimes talk about a relevant aggregate. An individual, for example, can simultaneously form part of many different groupings—by family, social class, income, ethnic background, race, and so on. The task is to find which of these groupings is the significant aggregate for the problem at hand. One of the worst mistakes social scientists ever made was to assume that race is the most important determinant of human action in society or in history. That mistake led to the nineteenth-century rise of pseudo-scientific racism. Yet, to solve historical problems, we do first have to identify the relevant aggregates and discover their limits in time and space. Sometimes the correct answers are so obvious as to present no difficulty. At other times, discovering the correct aggregate requires rare breadth of knowledge, depth of insight, and plain luck. The point, however, is that the relevant body of data to be examined is not a free choice. It is dictated, ultimately, by the problem to be solved. Failure to identify it correctly can lead to errors that range from minor misunderstandings to completely wrong conclusions.

Let me take a few examples from the history of the Atlantic basin—a region badly treated by historians until recent decades, mainly because it has been partitioned among specialists in European, North American, Latin American, and African history. None of these groups paid much attention to the work of the others. The result was a range of misunderstanding from trivial to deadly. On the trivial side, I discovered, when working on the history of the Atlantic slave trade, that most Americans thought then (and probably still think) that a large majority of the slaves transported from Africa to the New World arrived in the present territory of the United States. In fact, those that came to the United States were only about 6 percent of the whole. I have no idea what teachers of U.S. history actually tell their students, but I do not think that they set out consciously to misinform. More likely, most simply stay within the limits of the assigned aggregate, the history of the United States—and thereby leave out any mention of the other 94 percent of the Atlantic slave trade. And students, unfortunately, take silence to mean absence. The result is a major misunderstanding of the role of the United States in the larger history of the Atlantic basin—and of the migrations that have formed so much of its history over the past four centuries.

A far more serious—indeed, deadly—misunderstanding underlay the early nineteenth-century idea of some philanthropic (and not-so-philanthropic) Americans that it would be both humane and convenient if freedmen of African descent could return to Africa. Recent history, as it was then understood, showed that people of European descent died of disease on the African coast at astronomical rates—in an environment in which adult Africans appeared to be reasonably healthy. This “lesson of history” suggested that, since black Americans looked like Africans and had African ancestors, Afro-Americans would be safe from African diseases. But it was wrong. The relevant factor was not race but childhood disease environment. The retransported settlers from America died at rates nearly as high as those of North Europeans. In many cases, the move to Liberia was an unintended death sentence.3

A third example comes from the demographic history of the American side of the ocean and concerns the comparative demographic patterns of slave populations in the U.S. South, the Caribbean islands, and Brazil. The error began with the failure of U.S. historians to look beyond the political boundaries of the United States. Historians of the South paid no special attention to the mortality and fertility rates of the antebellum slave population. In general, the numbers looked a lot like those for the white population in the same regions—with a slightly lower rate of population growth, as one might expect of a people with a lower standard of living than free people had. In this narrow framework of North American history, there seemed little to explain. Meanwhile, the Latin Americanists and Caribbeanists knew that slave populations in the tropics has such high death rates and low birth rates that population increase was rarely possible. Rates of net natural decrease could run to 2 or 3 percent, occasionally even more. But that, too, seemed explicable for the region. The white populations also suffered a net natural decrease, at least among those who were newly arrived from Europe. Only in the past two decades or so have historians of the United States looked further afield and found that the demography of the slave population here was highly unusual. What first appeared to be no problem at all now called for explanation—once it was set in the larger aggregate of New World slave demography.

These three examples, all taken from Atlantic demographic history, show an aggregate misidentified on grounds of race or from overly narrow regional specialization. Other conventions of historical discourse, even more commonplace, can lead to similar misunderstandings. One of the most deceptive is our conventional approach to divisions of time and space. Reigns and dynasties, centuries and decades are convenient short cuts for dealing with chronology, but these man-made markers can also take on the appearance of reality. Just think of the thousands of students who still associate Queen Elizabeth I with Shakespeare’s plays, simply because we tolerate the label “Elizabethan drama.” Some escape is possible by petty distortion, like beginning the nineteenth century in 1815 and ending it with the outbreak of the First World War. But the conventions sometimes get out of control, like the former habit of dating the Industrial Revolution in England from 1760 to 1820—partly because those dates mark off neat decades and partly because they are the regnal dates of George III, who, of course, had nothing to do with it.

Mapping conventions are far more serious in their unintended—sometimes their intended—influence on historical thought. The Mercator projection is a prime example. Even though we know distortion is necessary in order to show the surface of a sphere on a flat piece of paper, we become conditioned to accept the convention as reality. As a result, most well‑informed people “know” that the European subcontinent of the Eurasian landmass is considerably larger than the Indian subcontinent, when in reality they are nearly the same size. This misperception arises not only because the Mercator projection enlarges all northern territories but also because we use the Urals as the conventional eastern boundary of Europe, when a line drawn from the White Sea to the Black Sea makes much more sense geographically.4 For similar reasons, most Americans imagine the Indonesian archipelago as about the same size as the Antilles. In fact, Indonesia from east to west stretches a good deal farther than the distance from Maine to California.

The conventional hemispheres cause still more serious distortions. Anyone who bothers to think about such things knows that the potential number of hemispheres is infinite, simply because the earth can be viewed from an infinite number of points in space. Knowing our ethnocentric traditions, we might expect conventional hemispheres in school atlases to center on the United States, and the newer books often show some such thing. But the older convention respects the earth’s rotation and shows the north pole at the top and the south pole at the bottom. Hemispheres are thus viewed from above the equator. If the United States is then placed in the middle between east and west, the result is a conventional hemisphere that actually centers approximately on the Galapagos Islands off the coast of Ecuador. This is, of course, the hemisphere that served as a basis for “hemispheric solidarity.” It conveniently puts the Asians and the Europeans halfway around the world—and on another page. Another hemisphere, centering on Omaha, would show that parts of Siberia and northwest Europe are comparatively close neighbors. Even parts of Africa would be included, while Argentina would not appear at all.

The hemispheric misconception is obviously associated with the political hemispherism of the Monroe Doctrine, the Good Neighbor Policy, and American isolationism before the Second World War. But even historians are still led astray. A recent book on the North American fur trade argues that a particular combination of native American religious ideas joined the commercial impact of European fur traders to produce the depletion of fur-bearing animals in North America between the early seventeenth and the early nineteenth century.5 Both the author and his opponents have argued their case without seriously considering that the depletion of fur-bearing animals belongs not to the Galapagos-centered hemisphere but to a polar-centered hemisphere. Over these same centuries, the phenomenon stretched from Finland east to the Saint Lawrence. The few miles of the Bering Straits made no real difference to the Russian fur traders, who followed their prey into Alaskan waters and on south to California. The obvious aggregate is the depletion of fur-bearing animals in northern latitudes in these centuries. The rule of parsimony suggests that similar events, wherever they occur, have similar causes. It can hardly be legitimate to leave out the culture and psychology of the Native Siberians, which might have been quite different from those of Native Americans—even though the Siberians killed their animals with about the same speed and timing as the Native Americans did.

In this case, then, the relative aggregate was split in two by geographical convention. Much the same is sometimes done for political reasons. The treatment of Canadian history in U.S. schools and universities provides a striking example. In the broadest sweep of world history over the past four hundred years, the most important thing that happened in the history of northern North America was its repopulation by settlers from Europe. These settlers founded new societies of overseas Europeans, beginning about the same time in Canada and the United States. This pattern of blanket settlement by Europeans, also prevalent in southern South America, was clearly different from two other kinds of settlement over these same centuries: the settlement of some Spanish and Portuguese among the surviving Indians of the tropics, especially in the highlands, and the settlement of the predominantly African slave populations in the tropical lowlands.

It can be argued that the fullest understanding of New World history requires a comparative study of what went on in all three of these zones. What is hard to justify is dividing the zone of blanket settlement in two by a political boundary. For school and university teaching, it seems self-evident that Americans need to know how Canada came to be there and how the Canadian historical experience differs from our own. In fact, Canadian history adds a whole new dimension and range of understanding to U.S. history seen in this comparative context. Comparison is valuable for illuminating differences as well as similarities, as the example of slave demography demonstrates. Regionalism in politics has been important in the political life of both Canada and the United States, but it has taken different forms. Both countries have a federal constitution—but one is parliamentary and the other presidential. Both countries have experienced severe regional conflicts—but one led to a ma or civil war and the other did not. One country has faced a severe problem of linguistic nationalism while the other has not. Yet our old emphasis on political history, with a healthy carry-over from the patriotic goals of an earlier historical tradition, makes it possible for historians on both sides of the border to deny the immense explanatory value that a broader history of North America would have had.

Examples can be multiplied. Nor is the problem limited to history as an intellectual discipline. The underlying problem is the proliferation of all knowledge in this century. The historians’ solution, in common with that of other disciplines, has been to multiply the fields of specialization. To some degree, that is unavoidable. But its very unavoidability imposes an obligation. We must try even harder to balance the depth of our own specializations against a wider span of historical knowledge—to make sure we are asking the most important questions and seeking answers in the framework of the relevant aggregates. I can close by recalling two time-honored aphorisms that are still worth remembering:

Some of those gaps in our knowledge belong there.
An elegant answer to an irrelevant question is still irrelevant.

Philip D. Curtin (May 22, 1922–June 4, 2009) was professor emeritus at Johns Hopkins University. He specialized in Africa and the Atlantic slave trade.



  1. The journal of the Association experiences the same difficulties as the Program Committee. The Review cannot publish what it does not receive, and on several occasions in the last few years the editors have had to plan special issues and call for papers in those fields in order to receive articles in Asian, Latin American, and African history. []
  2. Toynbee, A Study in History , 1 (2d edn., London, 1935): 17–181. []
  3. Tom W. Shick, “A Quantitative Analysis of Liberian Colonization from 1820 to 1834, with Special Reference to Mortality,” Journal of African History, 12 (1972): 45–59. []
  4. The area of Europe, exclusive of Iceland and any part of the Soviet Union, is approximately 1,820,000 square miles; the area of the Indian subcontinent, exclusive of Ceylon and Burma, is roughly 1,680,000 square miles. Including that part of the Soviet Union west of 40 degrees east longitude (a line drawn essentially from Archangel to Rostov) would add roughly 675,000 square miles to the area of Europe; including Burma (often considered geographically a part of the Indian subcontinent) would add approximately 260,000 square miles. []
  5. Calvin Martin, The Keepers of the Game: Indian-Animal Relationships and the Fur Trade (Berkeley and Los Angeles, 1978). Also see Shepard Krech III, ed., Indians, Animals, and the Fur Trade: A Critique of Keepers of the Game (Athens, Ga., 1981). []