Did Franklin Delano Roosevelt escalate confliict with Japan and Germany before Pearl Harbor, or did he attempt to avoid war? To what extent did U.S. public opinion influence these decisions? And, crucially, how do we know? Scholars offer diametrically opposed analyses of this historical case, bearing directly on international relations theories regarding the effects of democracy on war and foreign policy. In this debate and the broader security studies field, scholars increasingly employ published and archival primary sources. Because researchers lack a clear template for descriptive and causal inference with documentary evidence, though, such work is indeterminate and ultimately unpersuasive. How can political scientists approach archives and primary documents more effectively and efficiently? Above all, case studies need stronger research designs and clearer source selection strategies, not just more authoritative documents. A critical review of the sources cited in recent scholarship in the debate leading to the United States' entry into World War II, and a replication analysis of a key portion of the documentary record, underscores this need for improved research design and buttresses eight guidelines for the selection and analysis of textual evidence in case study research.


International relations scholars increasingly conduct case study analysis using primary documents and archival research rather than relying solely on secondary sources.1 “Primary sources are essential,” as Deborah Larson argues, to produce “fine grained decision-making analyses” that mitigate the biases of hindsight and uncritical dependence on existing scholarly works; assess causal mechanisms, counterfactuals, and alternative arguments; and account for a complex array of variables, including psychological and organizational factors in foreign policy.2 Documentary evidence now features on both sides of important debates in which the persuasiveness of contending theories hinges on the assessment of specific historical cases.

Despite several recent advances in qualitative methods, scholars interested in adjudicating those debates face several unresolved problems for successful descriptive and causal inference from textual sources. As Alexander George and Andrew Bennett warn, “The possibility of erroneous interpretation of the significance of archival documents is enormous”; they lament the absence of research that “provides an adequate discussion of the problems of weighing the evidentiary worth of archival materials.”3 Larson, similarly, observes that critics “complain that case studies are unscientific because history can say anything you want,” but “there is no guidebook for political scientists on how to use primary sources.”4 Two important contributions, by Marc Trachtenberg and Scott Frisch et al., focus, respectively, on assessing and interpreting individual documents' content in the context of public sources and scholarly literature, and on measurement and data collection (particularly for future quantitative analysis).5 Both volumes are especially good on the inductive component of archival research (how documents help reframe questions, theories, and assumptions) and on the practical one (the nuts and bolts of visiting an archive and working with documents and staff). Nevertheless, George and Bennett's point stands: political scientists need more guidance on how (and how not) to use archival materials and other primary documents for inferences and hypothesis testing.

The central problem is that for all the careful articulation of what Trachtenberg calls “the architecture” of a debate—that is, the concrete empirical questions that would yield leverage on broader theoretical arguments—there is virtually no explicit attention to what I would call the “architecture of the sources.”6 Even when both sides of a debate accumulate textual evidence, they generally have not specified how this selection of documents relates to other available sources and to the empirical questions they seek to address. This opacity with respect to source selection handicaps authors' and readers' ability to make successful inferences from those sources. Scholars regularly dispute interpretations of individual texts, but the main problem that I address in this article is a structural one: the multiple-stage selection process by which some primary sources but not others make it into the debate to be analyzed. Building on work by Cameron Thies and Ian Lustick,7 I consider the selection processes that restrict and categorize the sample of documents observed and cited by the researcher. In addition, I discuss sources of bias, and recommendations for improved effectiveness, in developing strategies for inference from a necessarily limited body of primary documents.

To showcase the promise and limitations of research with primary documents in security studies, I turn to the scholarly debate over President Franklin D. Roosevelt (FDR) and the U.S. entry into World War II (henceforth, the 1941 debate). This is a particularly rich and rewarding set of texts for students of international security and qualitative research methods, for four reasons. First, the debate features cogent yet starkly divergent historical claims. Did FDR intentionally provoke Japan and Germany in the months before the attack on Pearl Harbor, or not, and did his policy accord with or circumvent American public opinion? A “deception” argument (developed in Marc Trachtenberg's book The Craft of International History, advanced by John Schuessler in International Security and in his book Deceit on the Road to War, and deepened further by Trachtenberg in H-Diplo) suggests that FDR covertly escalated hostility with Japan and Germany before Pearl Harbor to produce American public support for a war he saw as inevitable and necessary.8 A “democracy” counterargument, articulated most thoroughly by Dan Reiter in Security Studies, maintains that FDR downplayed potential crises in the Atlantic, pursued deterrence in the Pacific, and confined U.S. foreign policy within limits set by public opinion.9

Second, the 1941 debate speaks directly to a major theoretical divide in the field of international relations between core strands of realism and liberalism regarding the impact of democratic institutions and public opinion on foreign policy and the causes and outcomes of war.10 If a U.S. president could manipulate or bypass public opposition in provoking or hurrying entry of the United States into a major war, and could survive politically, then this would deal a significant blow to democratic (and American) exceptionalism and would provide strong support for the international relations realist worldview. Although one case will not resolve one of the foundational debates in international relations scholarship, much is clearly at stake.

Third, FDR's presidency is a particularly valuable case for examining inferenee and evidence in security studies because of Roosevelt's enigmatic leadership style, compounded by the absence of presidential diaries and memoirs. When presidents provide ostensibly authoritative collections of their own words, one might feel confident (perhaps, overconfident) that one has direct access to their objectives and motivations. With FDR, though, biographers acknowledge confronting an unavoidable problem of uncertainty.11 This analytical challenge should discipline scholars of FDR's foreign policy to argue inferentially and to reflect on the logic of our claims and evidence, developing best practices that can be applied even in seemingly more straightforward cases.

Fourth, as the 1941 debate has progressed, it has leaned increasingly on primary sources. Now that both sides of the debate have cited extensive historical scholarship, several published primary sources, and archival documents, how can readers determine which account is superior? Close examination of qualitative research in practice, juxtaposed with prevailing methodological guidance and templates, places the substantive debate in methodological perspective and highlights areas for improvement on both sides. As researchers track causal arguments through the bramble of primary sources, their analyses will remain unpersuasive, and debates irresolvable, without clearer guidelines for evidence selection and inference. Existing methods literature in political science and history, however, has not fully developed the tools to correct the problem. This article elaborates a set of practices for social science research with archival and published primary evidence, both on FDR's foreign policy and, more broadly, in security studies.

In the first section, I discuss the role of causal and descriptive inference in political science, the contributions and limitations of current qualitative methods techniques to strengthen inferences in documentary research, and the distinctions between history and political science with respect to inference from textual sources. Second, I critique the selection of primary sources in the articles and chapters that form the 1941 debate. Third, I discuss several concrete proposals to replicate and extend this research and advance the debate. Fourth, to place the political science corpus on Roosevelt's foreign policy in cross-disciplinary perspective, I consider the far more extensive documentary research employed in two landmark historical monographs, and I argue that future increases in the availability of documents offer little prospect for resolving scholars' empirical debates—whether over historical cases such as World War II or more recent controversies such as the 2003 Iraq War—without clearer standards of evidence and inference. Fifth, to help develop these standards, I suggest eight guidelines for improving inference, transparency and replicability, and analytical leverage with primary documents for hypothesis testing. Sixth, to illustrate these recommendations and help break the logjam over FDR's foreign policy, I conduct a targeted replication analysis of a subset of documents that both sides of the debate identify as critical evidence. I conclude by reviewing the lessons of the 1941 debate for security studies scholarship, emphasizing the value and promise of further archival research in international relations, and summarizing my suggestions for how to conduct this work more effectively.

Approaches to Inference

Inference is central to political science research methods. As Gary King, Robert Keohane, and Sidney Verba explain, no causal relationship can be proven definitively in the social world, merely supported through a series of inferential tests; proceeding indirectly, scholars “[use] the facts we know to learn about the facts we do not know,” both for description and for explanation.12 Thus, researchers remain skeptical, not only subjecting our arguments to hard tests but also continually engaging rival explanations and alternative causal factors.13 Quantitative and qualitative political scientists address this problem in different ways, and historians offer an alternative template. The arguments we make about cases depend, however, on the evidence we can draw from our sources: there are multiple stages of inference in any case study, and some of these have clearer methodological guidance than others.

Scholarship on causal inference is largely quantitative and employs an understanding of causation based on the laboratory experiment, comparing effects across groups of cases. In an experimental setting, researchers are able to randomly assign subjects (or cases) to a treatment or a control group, making it relatively straightforward to interpret a correlation between treatment and outcome as a sign of causal impact. When working with existing observational data, though, it is usually problematic to employ statistical models that assume random assignment. Even if we control for variables that might have an independent effect, other factors (known as “confounders”) could influence the relationship between our hypothesized causal variable and the outcome of interest, producing a correlation that would erroneously suggest causation. Recent work has developed several tools for causal inference to correct for these problems, by focusing on comparing treatment effects across data points that are as similar as possible with respect to many contextual factors.14 Qualitative researchers have likewise developed tools for more effective case selection for cross-case causal inference.15

The predominant approach to causation in qualitative work, however, is based on process tracing to investigate causal mechanisms (and claims about the necessity or sufficiency of causal factors) in individual cases rather than “average effects” across a population.16 Process tracing may face the steep challenge of a “continuity criterion”: uninterrupted observation of a causal mechanism and explanation of “all the intervening steps in a case,” the “intervening dominoes” as well as those at the beginning and end of a row, rather than a looser series of congruence tests or causal process observations within a case.17 Even if the continuity standard is relaxed, using “causal process observations” at all still suggests that the evidence offers straightforward access to mechanisms, particularly when a “smoking gun” is uncovered, even if many other data are unavailable.18 Conversely, scholars in the Bayesian tradition have argued that observations in process tracing are usually indirect and qualified, involving the accumulation of inferential “clues” with varying degrees of confidence or probability.19

Thus, relatively formalized inferential procedures in case study work, including counterfactual analysis as well as Bayesian updating, take on a heavy burden of evidence. Counterfactual work requires enough case data to suggest a plausible minor change and investigate its hypothetical consequences, and Bayesian work mandates assertions based on what the best available evidence shows relative to alternative possibilities.20 How do we know if we have sufficiently powerful sources, enough of them, and what the preponderance indicates? Part of the solution involves acquiring stronger sources, and this informs qualitative scholars' turn to archival research, but we may be unduly confident about what we can learn simply from combining primary sources with process tracing.

Guidebooks on process tracing persuasively depict an inferential endeavor in which empirical tests and observations provide leverage on theoretical claims about causation in a case study.21 But because each test is ultimately an assertion or judgment about what the evidence collectively shows, we need to examine a prior stage of inference, regarding source selection. Persuasive explanation depends not just on the quality, quantity, accuracy, or transparency of sources, but on their selection and their structural relationships to other potential sources of evidence relevant to inference bearing on the theoretical propositions. George and Bennett's row of dominoes is a useful metaphor for evidence on causal mechanisms, but it implies a monolithic concreteness of each observation that a limited sample from a fragmentary documentary record is hard-pressed to support. Even extensive exploration of primary and archival sources affords only limited windows into the policy process. Thus, not only are we likely to fall short of the continuity criterion for process tracing, but also any given causal process observation is going to be incomplete, contestable, and inferential. Nor is deception and bias the central problem, either from a document's author or its scholarly analyst.22 Rather, the issue is descriptive inference from a sample of sources to a probative piece of evidence for or against a causal argument.23 Each causal process observation, each falling domino, is a composite of descriptive inferences that ultimately enables causal inference at the level of the overall case. To make claims about the “uniqueness and certitude” of the fit between theoretical prediction and documentary evidence,24 we might visualize instead a row of partially completed jigsaw puzzles and turn our attention to the structure of our sources.

Several scholars have raised concerns about source selection. As Lustick argues, rather than “the difficulty of finding necessary information,” “[t]he more daunting question is how to choose sources of data without” stacking the deck in favor of our arguments.25 Andrew Moravcsik observes a field in “self-imposed crisis” because qualitative researchers lack “explicit methodological rules…. The selection, citation, and presentation of sources remain undisciplined and opaque”; some works might even “cherry-pick” sources to offer a misleading “patina of primary-source content.”26 Jack Snyder notes that because “[s]cholarly debates rarely hinge on smoking-gun evidence,” persuasive research requires “a well-explicated theory, a systematic discussion of methodology, and a textual narrative that explains how diverse evidence combines to support an argument.”27

Addressing these concerns, scholars are developing and vigorously debating transparency initiatives, including innovations in expanded citations and online supplementary materials to facilitate replication.28 Hyperlinked sources offer a check against misstating key passages; additionally, clarifying the relationship between a quotation or paraphrase and the original helps readers assess whether the quote accurately reflects the text surrounding it, or whether it was taken out of context.29 Between the document and the claim, however, lie further inferences involving the source's relationship to the theory, the source in the context of available sources, and the available materials relative to what the historical actors originally produced. Much of this evidentiary reasoning can be articulated in citations or appendices, or in the body of a manuscript, but we still need to know what to say—that is, how to specify the logic of source selection and inference. This article elucidates strategies for inference from source documents to individual observations, so that process tracing can connect the dots.

In developing these practices, to what extent should political scientists emulate historians in our selection and analysis of primary documents? After all, archival research is the lifeblood of history as a professional discipline. Moreover, history is a fundamentally inferential enterprise, and classic texts on historical methods emphasize the same points about uncertainty, skepticism, and extrapolation from evidence that political science does.30 For instance, historians are excellent on the selection biases imposed by document survivability and intentional collection into archives, and thus on inference from the body of surviving documents to the original unobserved whole.31

Historians tend to assume, however, that to make those inferences, scholars should tackle comprehensively the records that have survived. As a result, some historians suggest that archival work, for all its painstaking nature and serendipity, is ultimately fairly “simple” and “straightforward”32—figure out where the collections are, use the finding aids, and engage all the relevant sources. This perspective applies particularly to diplomatic historians relying primarily on national governments' central records. In contrast, social and cultural historians, and scholars of marginalized populations or authoritarian regimes, might argue that sources require more creative unearthing, and that existing collections must be read critically and against the grain.33 The urge for comprehensiveness informs the recent wave of multinational, multiarchival research in diplomatic history, based on the understanding that U.S. foreign policy requires “de-centering” the United States and examining it “in the world” through other eyes.34 Completeness is an impossible standard, and even if it were possible to explore all relevant archives and read all available records, this would not resolve every analytical dispute. Making claims about what the preponderance of the evidence supports, however, suggests a burden of consulting the bulk of relevant historical material, at least within some delimited scope.

For political scientists, who are interested in targeted rather than comprehensive searches, a second level of inference needs to be thought through, in the language of sampling. Here, historians may offer less help. Certainly, there are calls for more explicit attention to methodology in historical research: Marc Bloch, for instance, argues that “[e]very historical book worthy of the name ought to include a chapter, or if one prefers, a series of paragraphs inserted at turning points in the development, which might almost be entitled: ‘How can I know what I am about to say?’“35 This appears to be a minority position, however. John Lewis Gaddis explains that the “functional theory of historical verification … is largely implicit”; the community of historians, borrowing from Supreme Court Justice Potter Stewart, will know good scholarship when they see it, so researchers “get on with doing history as best we can, leaving it to our readers to determine which of our interpretations come closest to the truth.”36 More bluntly, as Samuel Eliot Morison famously argued, “It matters little what ‘method’ the young historian follows, if he acquires the necessary tools of research, a sense of balance, and an overriding urge to get at the truth. Courses on historical methodology are not worth the time that they take up…. Historical methodology, as I see it, is a product of common sense applied to circumstances.”37

So how should political scientists proceed with the available sources, if we aspire to qualitative research with a degree of rigor? A more accurate distinction between political scientists and historians of international relations is not that the former pursue causal explanation and the latter emphasize interpretation or understanding, but rather that political scientists test and refine relatively parsimonious theoretical models with historical data, whereas historians, albeit equally interested in explanation, tolerate greater complexity and even contradiction in their accounts.38 The methodological implications of this theoretical divide have yet to be spelled out.

In particular, supporting or infirming a relatively parsimonious hypothesis, while tolerating a lower degree of fit between evidence and argument (i.e., more unexplained noise or variation), is a more limited aim than illuminating a historical episode. If historians consider a broad interplay of factors and even reciprocal causation, and if, though guided by questions, they welcome induction at least as much as deduction, then their pursuit of comprehensive exploration of the documentary record is understandable. For political scientists, reliance on a more limited evidentiary base is justifiable, but that base must be strategically selected, and its boundaries and trade-offs clarified and defended to support inferences and claims. If “what I am not looking at and why” is not an issue on which political scientists can expect much help from historians, then we need to develop our own standards.

Sources and Scholarship on U.S. Entry into War in 1941

The 1941 debate among international relations scholars is fascinating from the standpoint of documentary sources and descriptive and causal inference. Political scientists employ primary and archival documents as evidence for competing explanations of a specific historical case with major theoretical implications. Trachtenberg and Schuessler argue that before the Japanese attack on Pearl Harbor, FDR was more interventionist than the majority of the American public, and that his foreign policy toward Japan and his naval policy in the Atlantic involving Britain and Germany aimed to provoke a conflict that would convince the U.S. public to support entering World War II both in the Pacific and in Europe.39 If this thesis holds, it would suggest that democratic leaders, at least when facing major external threats, are relatively unconstrained by public opinion because they can circumvent and manipulate it. More broadly, this finding would undermine the idea that domestic regime type significantly affects foreign policy behavior and the prospects for war and victory with particular kinds of adversaries. Conversely, Reiter argues that Roosevelt sought to deter rather than provoke Japan, to minimize rather than exacerbate naval confrontations in the Atlantic, and to persuade rather than deceive the American public (who, moreover, Reiter argues, already strongly supported more vigorous pressure on Japan and Germany despite the risk of war).40 If Reiter's argument finds superior empirical support, it would suggest that the “marketplace of ideas” does function well in democracies; that leaders are closely constrained by public opinion; and that countries with different domestic political systems do have different prospects for war initiation and victory. No single case study will resolve paradigmatic disputes among international relations scholars such as the question of democratic exceptionalism, but the 1941 case is a core theoretical battleground, and a methodologically stimulating one.

The theoretical and empirical “architecture” of the “historical problem,” as Trachtenberg calls it, has been laid out with exemplary clarity.41 Reiter articulates six points of debate (three of which involved naval activity in the Atlantic) with exceptional focus in a single paragraph.42 In a similar spirit, I identify three main areas of debate for the 1941 case: policy toward Japan, naval policy in the Atlantic, and the decisionmaking process in the Roosevelt administration. Linkages between these three are the crux of the “back door to war” thesis that FDR covertly provoked Japanese escalation to bring an otherwise reluctant United States into the war in Europe.

In Asia, where Japan had occupied Manchuria in 1931 and invaded eastern China in 1937, the United States supported nationalist Chinese resistance and worried that Japan might move further into Southeast Asia, particularly against Dutch Indonesia and its oil reserves. In 1941, the United States maintained diplomatic negotiations with Tokyo on these issues, before and after instituting an embargo on oil shipments to Japan. Two debates follow. First, was the embargo, particularly in FDR's view, designed to provoke or to coerce Japan, or to serve some other objective? Second, were U.S. negotiating efforts with Tokyo undertaken in good faith, or designed to provoke through unacceptable demands or to stall for time while the United States prepared for an inevitable war?

In what historians call the “Battle of the Atlantic,” the United States ramped up support for a beleaguered Britain, which had suffered the “Blitz” of Nazi bombardment from September 1940 through May 1941, even as the Roosevelt administration publicly pledged to keep the United States out of the war.43 Roosevelt proposed the Lend-Lease program in December 1940, and Congress approved it in March 1941. As the U.S. Navy began escorting supply convoys to Britain, naval encounters multiplied between U.S. vessels and German submarines, including an exchange of fire between the USS Greer and a German U-boat in September, and U-boat sinkings of the USS Kearny and Reuben James in October. To coordinate this complex unofficial alliance, FDR and British Prime Minister Winston Churchill met aboard the USS Augusta at Argentia off the coast of Newfoundland, Canada, in August; afterward, the two leaders announced the Atlantic Charter, a broad statement of shared values, war aims, and a liberal vision of postwar order.44 Again, two core questions follow. First, what exactly did FDR say to Churchill at Argentia about U.S. intentions regarding war entry and naval policy in the Atlantic? Second, was FDR's Atlantic naval policy, particularly in practice, one of avoidance or provocation, of downplaying or brandishing incidents between U.S. and German vessels?

In Washington, the foreign policy process brought together FDR's domestic policy concerns, including the New Deal and continued recovery from the Great Depression, popular and congressional support and opposition (Roosevelt won an unprecedented third presidential term in November 1940), and bureaucratic cross-pressures, as well as a circle of advisers with their own personalities and rivalries. All of this, and not simply the chessboard of international strategy, may have affected policy decisions in the Pacific and Atlantic theaters. Two questions are particularly salient. First, to what extent and in what way did public opinion considerations enter into FDR's decisionmaking with respect to the Atlantic and Japan? Second, which actors (societal, governmental, or foreign) had a relatively clear view of FDR's objectives and policies? Conversely, if there was deception by the Roosevelt administration, who was deceived? These, and similar questions, are core targets for new process-tracing research with primary sources.

The architecture of the documentary record, however, is opaque. In addressing the empirical questions outlined above, Reiter, Schuessler, and Trachtenberg have fought to a standstill in their use of secondary sources (particularly, works by historians). For instance, on naval policy, Reiter cites Gerhard Weinberg's A World at Arms to argue that FDR ordered the navy to avoid incidents with Germany in the Atlantic.45 Schuessler counters with a different page from Weinberg, arguing that Weinberg (and Reiter) misinterpreted the relevant primary source; Schuessler, however, cites Trachtenberg for this point.46 Moreover, Weinberg's own source is a chapter in a German-language edited volume of proceedings from an academic conference.47 Trachtenberg indicates that he read this chapter, by Jürgen Rohwer, “closely,” interested in “what Rohwer's evidence actually showed,” but he does not indicate what Rohwer's sources were.48 Reiter cites Rohwer on Roosevelt's removal of a passage in WPL-51, a U.S. naval strategy document, that (in Reiter's words) “would have permitted British vessels to join American convoys and enjoy their protection.”49 Also unclear is each scholar's use of this German-language source (I do not read German). Rohwer's original chapter contains ten pages of maps of naval activity in the Atlantic and the Americas; the remaining thirteen pages employ numerous references and quotations (in German) from U.S. primary documents, but without notes or source locations.50 The primary sources in question are neither specified nor consulted firsthand, at least according to the three scholars' citations.

Thankfully, the debate has progressed, and now incorporates primary sources, including archival documents, more extensively. Doing so produced another impasse, however: both sides present powerful quotations from several published and manuscript sources that contradict one another. The authors address various documents from different repositories without clarifying what else was examined or why they cited these files and not others. This approach to evidence readily establishes the plausibility of their competing arguments, but makes inference—and, for readers, adjudication of the debate—more difficult. Trachtenberg's chapter in The Craft of International History, though citing a range of documentary sources, is primarily an illustrative methodological demonstration of “textual analysis” of scholarly sources in historiographical debates.51 Schuessler's article offers extensive engagement in international relations theory, and then relies almost exclusively on secondary sources (and primary sources quoted within scholarly monographs). Reiter analyzes several American and British primary sources, including archival documents from the FDR Presidential Library. Trachtenberg's H-Diplo paper musters an even greater range of documentary sources. Schuessler's Deceit on the Road to War updates his article and responds to Reiter's and Trachtenberg's articles; most significantly from an evidentiary standpoint, it cites two additional primary sources. (An online appendix to this article compares each scholar's U.S. primary sources.52)

In sum, the 1941 debate has progressed from arguing about historians to arguing about historical documents, and as it evolved, scholars have produced a mass of footnotes to primary sources. What should readers make of this evolving body of research, and who presents the superior account? We could replicate directly as much as possible, checking quotations for error, misinterpretation, or worse. Even if all citations are accurate, however, the authors still disagree about fundamental empirical points regarding FDR's foreign policy, and all have quality sources. Stepping back to consider problems of inference, what types and locations of sources are being addressed and why these instead of others? Given the empirical questions under debate, are we looking for evidence in the most appropriate places?

Toward Replication and Extension

Deductively, where should scholars look to answer the core empirical questions in the 1941 debate? If we want to know what was said at Argentia, and we are citing the correspondence of some of the conference participants, what about the others? At least thirty individuals were present; the four Americans whose diaries Reiter, Schuessler, and Trachtenberg cite were not there.53 Charles Beard, but not the political scientists, cites the memoirs of Elliott Roosevelt (FDR's son), who actually did participate.54 Political scientists debate what Roosevelt said to Churchill, but their citations (and their sources' citations) boil down to what Churchill reported, after returning to Britain, about FDR's declarations.55 This evidence seems indirect at best. Similarly, to learn about U.S. naval policy in the Atlantic, how much information can we glean from State Department sources? And to debate a deception thesis, which sources would have participated in the conspiracy and which would have been its targets? Which lower-level officials and allied leaders would have been privy to Roosevelt's mind? What can foreign (e.g., British) documents, even archival ones, reveal about contested points of Roosevelt's foreign policy?

Inductively, the sources cited in existing works suggest some potential discrepancies or omissions.56 The partial use of the State Department's Foreign Relations of the United States series is particularly revealing about inferential problems in the 1941 debate. For some topics, FRUS might not be a natural source. For instance, on military policy such as naval procedures and deployments in the Atlantic, the State Department—even post facto, through its Office of the Historian—might not be authoritative. Additionally, public opinion is not a traditional priority of diplomats and foreign ministries; even if FDR read polls closely, FRUS might not select documents on that topic for publication. For topics such as bilateral relationships between the United States and allies such as Britain or adversaries such as Japan, however, FRUS is an essential resource. No doubt this is why one of the few primary sources now cited by Trachtenberg, Reiter, and Schuessler is the two-volume FRUS collection Japan 1931–1941, published in 1943 (notably, during wartime).57 This seems important for evidence on Roosevelt's policy toward that country, particularly on the goals of the oil embargo in the summer of 1941 and the objectives in preliminary peace talks and ultimatums that fall.

To what extent, however, does the corpus of documents in FRUS Japan 1931–1941 offer a comprehensive or representative evidentiary base for assessing that policy? There is an entire FRUS Far East volume (volume 4) for 1941 that Trachtenberg cited in The Craft of International History, but to which none of the subsequent works refer.58 That volume has 729 pages on Japan, and was released in 1956. (The Japan 1931–1941 volume, as noted previously, appeared in 1943; some sensitive documents from 1941 might have been harder to release while the war was ongoing and FDR was still president.) Similarly, if Trachtenberg, quoting Paul Schroeder, suggests that “the war came about over China,”59 then scholars should examine FDR's China policy in volumes 4 (Far East) and 5 (also Far East, with a China emphasis). None of the works on 1941 by Reiter, Schuessler, and Trachtenberg cite volume 5, which contains 922 pages of documents, including 450 on the Japan-China undeclared war and 396 on Japanese southward expansion.60 Moreover, the preface to the Japan 1931–1941 volume notes that the collection explicitly omits documents concerning U.S. relations with third parties bearing on Japan, to focus on the strictly bilateral agenda.61 This suggests that China-Japan-U.S. triangulation might not be best served by this volume.

Moreover, because the back-door-to-war thesis emphasizes connections between the European and Asian theaters, scholars should explore overall foreign policy strategy in volume 1 of FRUS for 1941 (which Trachtenberg's H-Diplo paper alone cites) and regional volumes on Europe (volume 2, uncited by anyone) and the British Commonwealth (volume 3, also uncited).62 The Soviet Union, France, and the Netherlands (in addition to Britain) had significant Asian and European interests. We might at least look for “Japan” or “public opinion” in those volumes' indices. Conversely, the volume 4 index on the Far East suggests that many documents concern FDR-Churchill conversations and relations with other countries discussing Japan, with extensive British requests for coordination.63 If FDR expressed to Churchill the idea of conflict with Japan as a “back door” to war with Germany, there is no particular reason to expect this comment to appear in a FRUS volume on Japan, as opposed to the Far East, Britain, or overall foreign policy.

Further deductive puzzles arise from the promising but preliminary consultation of the FDR Presidential Library archives. If political scientists want to explore FDR Library sources, which specific collections are the most valuable? Combined, Reiter, Schuessler, and Trachtenberg have cited documents from boxes 4, 55, 62, and 80 of the FDR Library's President's Secretary's File (PSF) series, which is available, in its entirety, online.64 That is, 4 boxes out of 174 in the series (the PSF contains up to 260,000 pages).65 Why these 4? Are the others irrelevant? Hardly. Robert Dallek, for instance, cites material from about 24 boxes—even if these had been chosen at random (which they surely were not), that sample would still comprise more than 13 percent of the total. The series' contents list suggests additional relevant material. Box 43 has folders on Japan; many of these documents (according to notes on images of individual pages) were published in FRUS (some in Japan 1931–1941 and some in 1941 volume 4 on the Far East) or in the Pearl Harbor hearings; a few appeared in both, and some in neither.66 If nothing else, this scattered distribution of primary material for publication tells us how partial a picture of U.S.-Japan relations we obtain by examining only one printed source. Box 22 contains bound dispatches from Ambassador Joseph Grew in Japan; these are not listed as published in FRUS or elsewhere.67 For Roosevelt's Atlantic policy, we might look at the numerous boxes on the Navy, the War Department, and the administrators of each; for the Argentia summit and relations with Britain, several British leaders have their own subject files.68

Moreover, the PSF is one of eleven series from FDR's papers as president available at the FDR Library, and hardly the largest. The nine smallest series—including the PSF—amount to 343 linear feet.69 The two largest collections, the Official File (OF) and the President's Personal File, contain 1,174 and 608 linear feet, respectively, totaling about 3.5 million pages.70 Neither series has been digitized, but both are clearly important to historians doing serious archival work on Roosevelt's foreign policy.71 The FDR Library maintains lengthy online finding aids and container lists for each.72 As one example of a potential search site based on container titles, OF 18 (Department of the Navy) boxes 17–18 include summer 1941 material for and against firing Navy Secretary Frank Knox.73 To assess FDR's naval policy, especially if scholars have already cited communications involving Knox, this might be important.

If political scientists have turned to primary sources, and particularly archives, to answer important historical debates on which major theoretical disputes depend, why is it that the sources cited are generally few, scattered, and easily electronically retrievable? Pending answers to questions such as these, readers may have a hard time taking sides in the 1941 debate over FDR's foreign policy.

What Would Historians Do?

In contrast to the scholarship discussed above and to common practice in political science, historians of FDR's foreign policy generally aim at exhaustive exploration of the best available sources.74 From the standpoint of documentary support for historical and theoretical arguments, it is particularly instructive to compare political scientists' sources with those employed in two historical monographs, decades apart: Charles Beard's provocative volume, President Roosevelt and the Coming of the War, 1941, and Robert Dallek's Bancroft Prize–winning book, Franklin Delano Roosevelt and American Foreign Policy, 1932–1945. A large number of high-quality sources does not automatically produce superior analysis—Beard's book, which presented a revisionist argument that the Roosevelt administration stealthily maneuvered the United States toward war before the Pearl Harbor attack, has long been roundly criticized for its framing and interpretation of evidence. Therefore, this is not an exhortation for political scientists simply to increase our citations or time in the archives, particularly because, as I have argued, we generally have different research objectives than historians. The asymmetry of sources between works of political science and history on the same events is sobering, however, and requires reflection. Even excellent process-tracing research is generally not emulating historians' practices: its use of sources is far more selective. Partial exploration of the historical record is defensible, even laudable for certain analytical purposes, but only if such selection is both transparent and strategic.

The works under discussion by Schuessler, Trachtenberg, and Reiter distance themselves from Beard's book as if it were a polemic devoid of serious evidence.75 Schuessler calls the Beard volume part of a “first wave of revisionist scholarship” that advanced a “plausible” thesis about Roosevelt pushing an unwilling and unaware United States into war.76 Trachtenberg omits any mention of Beard or his work.77 Reiter, however, links Beard to Trachtenberg and Schuessler through a “see also” reference (yet Reiter at one point cites Beard as a source confirming a point Reiter makes).78 Similarly, James McAllister, introducing the H-Diplo roundtable on Reiter's article, notes that “[p]utting the words ‘backdoor’ and ‘Roosevelt’ in the same sentence understandably conjures up unfortunate associations with revisionist historians like Charles Beard.”79

So what, really, was the problem with Beard's book? In mid-century broadsides, historian Samuel Eliot Morison critiqued not merely Beard's “innuendo” and errors, but also his selection of documents, and thereby his “frame of reference” method of analysis.80 Morison claimed, “Without misstating many facts or garbling quotations … Beard, by ingenious arrangement and selection, ruthless rejection of attendant circumstances, and a liberal use of innuendo, compiled a powerful brief for the thesis that Franklin D. Roosevelt was the aggressor against Germany and Japan.”81 Morison recognized that partial selection is inevitable, but argued that the goals should be an accurate representation of what happened, a “sense of balance” despite the impossibility of purely objective analysis, and eternal “skepticism” about the “preconceptions” that affect scholars' “selection and arrangement of facts.”82 If Morison is right that the central failing of Beard's explanation of FDR's foreign policy lies not in misrepresenting documents, but in cherry-picking them to undermine accurate inference, then despite marshaling ever more documents to support a thesis, political scientists will make little progress in inference until we improve and clarify our own selection procedures.

Beard, writing in the 1940s, actually drew on the best repository of primary documents available at the time: those obtained from various corners of the executive branch and published as “exhibits” by the 1945–46 congressional committee of inquiry into the Pearl Harbor attack. (All forty volumes are now accessible online.83) Trachtenberg's H-Diplo article alone in the 1941 debate cites this source.84 Nor was this the extent of Beard's primary-source exploration. Among published sources, he cited the Public Papers of FDR, the Congressional Record, various newspapers, the captured German documents translated and published as Fuehrer Conferences in 1947; the State Department's volume Peace and War 1931–41; the two-volume FRUS collection Japan 1931–1941; and British House of Commons Debates. Moreover, Beard explored the unpublished papers gathered by the Pearl Harbor committee (housed at the National Archives), including full transcripts of proceedings, mimeographs of Secretary of War Henry Stimson's testimony and diary, and various letters of Chief of Naval Operations Adm. Harold Stark.85 According to Beard, because FDR's 1941 press conferences had not yet been published and he could not gain access at Hyde Park (the nascent FDR Library), he obtained access to a newspaper's stenographic notes, including off-the-record statements by Secretary of State Cordell Hull.86 And when Beard found published references to documents that he did not possess, he explained that he had tried to obtain the originals: reading about a press release for which he could find no newspaper account, he requested the source from the State Department, but was told that no such document existed.87

In taking these steps, Beard repeatedly offered his rationales for inference and interpretation by contrasting two sets of documents, or two versions of the same document or event.88 Thus, Beard argued that “[s]tudents of history should be on guard against basing conclusions solely on the voluminous printed record of the Congressional Committee on Pearl Harbor.”89 Similarly, Beard offered a clear logic of inference and interpretation by contrasting the initial official narrative and the subsequently released version of the same events or documents.90 Scholars can criticize Beard's use or abuse of these materials, and we can engage more recently released documents to which he did not have access, but we ought at a minimum to engage the declassified sources with which he grappled so extensively.

Let us fast-forward to Dallek's monograph, which Trachtenberg, Schuessler, and Reiter all cite repeatedly and enthusiastically. Trachtenberg calls it “the most important book on Roosevelt's foreign policy” and cites it thirteen times in chapter 4 of The Craft of International History.91 Schuessler calls Dallek's book “the most comprehensive account of Roosevelt's foreign policy” and cites it (or quotations reproduced therein) more than a dozen times.92 In a letter in International Security, Reiter cites Dallek in a quarter of the footnotes; in his Security Studies article, Reiter cites Dallek ten times out of 132 footnotes.93 The period of major debate—1941—largely falls within chapter 11 of Dallek's book. Within that one chapter, Dallek cites more primary sources than do the works under discussion by Reiter, Schuessler, and Trachtenberg put together—not that quantity alone should persuade. From the President's Secretary's File series of the FDR Library alone, in chapter 11 alone, Dallek cites documents from twenty-three separate subject entries, some of which span multiple boxes.94 For instance, the PSF has material on public opinion polls, enabling Dallek to comment on unpublished polls commissioned by the government, as well as on FDR's and his aides' reactions to those polls. This has tremendous inferential value for the 1941 debate, but none of the political science authors have cited these documents, even though they are available online.

Because of historians' dedication to comprehensive exploration of the body of available records, and their sensitivity to its limitations, they understandably anticipate potential revelations from future waves of declassified documents. In 1948, Beard looked forward to the eventual opening of FDR's personal papers, the public release of captured Japanese documents, and the opening of other U.S. and foreign archives, without all of which many of the core questions about Roosevelt's foreign policy “must remain for the present a matter of conjecture.”95 In a similar spirit, Schuessler and Reiter seem to agree that future declassification will help resolve their dispute over the role of presidential deception in the 2003 Iraq War.96 For historians, but even more acutely for political scientists, additional tranches of potential evidence provide diminishing returns to understanding, and even complete access to the archives would not settle all debates. Six decades after the end of World War II, Trachtenberg observed that historians still “[find] it hard to nail down, in any conclusive way, exactly what FDR's policy was”; this point “implies that the absence of hard, ‘smoking gun’–type evidence does not necessarily mean that much, and that in interpreting Roosevelt's policy one has to place greater emphasis on indirect reasoning than one would perhaps like.”97 Hence Schuessler's frank admission of offering a “circumstantial” case, noting that “[a]dmittedly, some uncertainty about Roosevelt's intentions is irreducible.”98 If we cannot resolve the 1941 debate now, even with the massive and high-level documentation currently available, what are our prospects for assessing the 2003 debate at an equal distance into the future? Because political scientists are generally not emulating historians' drive for comprehensiveness, we should temper our optimism regarding the proliferation of sources, and we should turn our attention to sharpening our research strategies. For “circumstantial” or “indirect,” let us read “inferential,” and strive to make it so.

Guidelines for Dealing with Documents

How then can political scientists most effectively approach the sources we have? The existing approach in political science is to read some primary sources and cite them as evidence. The historians' approach is to read as much as possible and then cite evidence. In between, political scientists interested in establishing a strong evidentiary foundation for causal inference should try to specify what to read and cite, and what not to, as a strategic selection, and to explain their rationale for these decisions as part of a research design. To support this endeavor, I suggest eight guidelines for improved inference in research with primary documents, including in the archives. I also derive further proposals for new, targeted research on FDR's foreign policy. The first three points concern transparency and replicability, clarifying the inference for readers to follow and assess.

First, scholars should specify which archival series, boxes, and folders we examined during our research (not just those eventually cited), not merely which archives or vast series or record groups (too broad for most types of inference) or which documents (too narrow for inference). Likewise with published documents—which volumes of FRUS were consulted for which issues, for instance. The goal is not to impress with quantity or to lead future scholars into blind alleys of irrelevant material. In fact it is the opposite: identifying unhelpful boxes saves future scholars effort, helps them target sources not yet consulted, and makes their work more compelling if they uncover damning evidence that we overlooked or omitted. Clarifying the relationship between our argument and search strategy might also yield important null findings where a subject surprisingly went unmentioned in the records of a particular office or decisionmaker. Veteran scholars who have explored hundreds of containers of material might understandably question the practicality of a comprehensively detailed list. It seems reasonable, however, to ask scholars to outline with some precision where they chose to search for evidence, and what they found. Intriguingly titled but ultimately unhelpful archival boxes could usefully be summarized in a brief line about the hunch that led the scholar to those files and why that material is not cited in the narrative, thus clarifying the research process and augmenting its persuasiveness.

Second, citations should specify the location (series, box, folder, hard copy or online) of the document, not simply what it was—the same diplomatic cable matters differently for inference depending on its larger documentary context.99 Several copies of a document may be available in FRUS and other publications, or in the archival records of different offices or individual decisionmakers; and where a document resides, not merely what it says, can be significant. This point goes double for documents consulted online, including on the website of the archive that maintains the original, or in a printed or third-party collection (e.g., the National Security Archive). No citation should omit explicit mention that a document was read electronically (in which case scholars should provide a hyperlink), which would misleadingly imply that archival research had been undertaken on the ground. (This guideline is not simply an issue of deception or transparency, but again speaks to inference, establishing the relevant population of documents from which a scholar sampled.) Such citation detail is common practice, but it is also frequently violated in political science, so it bears emphasis.100

Third, either for an overall case study, or for particular empirical points, scholars should articulate the inferential rationale for selecting the sources examined and cited as opposed to others. If we look at only one folder in a box, box in a series, series in an archive, section of a volume, or volume in a series, why? For instance, citing documents solely by their box or volume number, if they were uncovered through a selective but unspecified process such as electronic keyword search or an index entry, misleads by implying that the author reviewed the rest of the material, of which the quoted sample is representative. Similarly, even if we read a document in the original but had searched for it directly after following another scholar's citation, we need to clarify this route and establish what documentary context we have also read. Search terms may also induce selection bias: looking for an individual will enhance his or her prominence, and a search for particular crises will find them, whereas a search for peace overtures would uncover those instead. Scholars are responsible for ensuring that cited material faithfully represents the context from which it was selected, which implies the need to review and report surrounding material within a particular scope.101 Because political scientists are generally employing partial rather than exhaustive documentary investigations to test hypotheses, that scope may be fairly narrow, but it needs to be delimited and justified.

Another three suggestions help establish such inferential rationales based on research design. To analyze a historical case through primary documents, which bodies of text will be explored and why? We are not testing a hypothesis with a case in toto, or a particular document, but with a sample of documents. Therefore, we should frame implications of hypotheses that point toward some sources or collections over others. In turn, it should be possible to assess the inferential basis and prima facie persuasiveness of a particular study even before looking at the content and authority of the quotations, by looking at from whence these came. Which sets of documents should provide evidence on competing propositions about, for example, Roosevelt's preferences regarding Japan in 1941?

The fourth suggestion, then, is to search where other scholars have already worked, the files established as bearing the best evidence on the contested points. If we already cite Dallek or Wilson on Argentia, then follow the lead of their footnotes. Similarly, we might also rely on guidance from archivists, taking expert directions that likely cued other researchers as well. Either way, we should explain this strategy, with citation, for transparency. This approach might be particularly effective in the early stages of a large research project, and it is essential in replication analysis. It is also important for revisionist arguments, because countering a prominent scholarly work or predominant consensus in the field based on texts already admitted into evidence is a powerful rebuttal. Debates on the drivers of European integration illustrate this approach, proceeding not primarily by the constant interjection of new material, but rather by critiquing competitors' use of a mutually agreed-upon, manageably sized corpus of documents.102 Conversely, omitting these documents might look like avoiding evidence against one's argument. To defeat a “deception” argument, what do we make of the Pearl Harbor Hearings files, on which Beard relies so heavily?

Fifth, scholars could try the reverse approach: explicitly seek out new evidence, previously uncited. Explore on a hunch, in series or repositories where predecessors have not—but with a clear rationale for what one hopes to find and why. We might target overlooked material—Eleanor Roosevelt's papers, perhaps, for private references to her husband's objectives. Or we might focus on newly released material, assessing whether findings based on these documents contradict those employing previously available, perhaps less highly classified ones. Prominent examples of this strategy (which, like the previous approach, seems useful for revisionism) include works by Fred Greenstein, John Lewis Gaddis, and Caitlin Talmadge.103 This approach also facilitates triangulation across document repositories and types of evidence (including multi-archival research), which is increasingly common in international and diplomatic history and is a touchstone of mixed-method work in political science.104 Done transparently and strategically, such work should contribute both to cumulativity of knowledge and to inference, akin to digging in separate squares of an archaeologist's grid. If the evidence in one set of papers supports one theory, whereas a second trove of documents undermines the argument, we are in a position to assess which repository should be treated as more authoritative, or to discuss the selection effects that produced this variation. Without a clear logic of source selection, however, triangulation can simply proliferate rather than resolving inferential problems, by expanding the population of sources from which a scholar is sampling.105

Sixth, rather than new materials or unexplored old ones, we might map out the structure of existing documentation and search for evidence more systematically based on our knowledge of decisionmaking processes, historical context, and relevant hypotheses. For instance, on Argentia we might consult only the files of individuals actually present, or we might focus on military documents rather than State Department sources to determine Atlantic naval procedure. For cabinet officials' roles in White House policy, their papers at the relevant presidential library might be the best repository, while the workings of departments (including cabinet members' interactions with subordinates) are better tracked in those organizations' files at the National Archives. Secondary sources and newspapers could also help us account for other factors in decisionmakers' situational awareness, for instance by stated positions in documents before and after salient events such as FDR's reelection or Hitler's invasion of the Soviet Union.106 As with Frank Schimmelfennig's suggestion of “efficient” process tracing by analyzing selected parts of a causal chain, this strategy would target portions of the documentary record estimated to have the greatest theoretical relevance or empirical payoff.107 Archival finding aids and indices to published volumes of documents are invaluable for targeting, transparently and efficiently, particular actors, subjects, and time periods within a case.108 Deduction and design may, in many cases, replace a great deal of less focused textual research.

Two final suggestions aim to bolster inference based on individual documents. We should pursue these most aggressively for documents that support our argument, and those not already contextualized through the research design points above. The temptation is to explain away documents that point against us, and this can often be readily done given the mass of available documentation. Seventh, then, because a given document is usually insufficient to trace a causal process, clarify for which part of a causal argument it is serving as evidence: one text should not be used simultaneously to measure an independent variable, a dependent variable, and a causal mechanism. For instance, a source might indicate the degree of aggressiveness in U.S. naval policy toward Germany (dependent variable), FDR's objectives regarding Germany (independent variable 1), the public's desires regarding Germany (independent variable 2), or the influence of public opinion on FDR's decisionmaking (mechanism), but it cannot realistically bear the weight of process tracing to connect all of these elements. Political scientists already warn against excessive credence in apparent archival “smoking guns” without corroboration and verification of the author's intent.109 As Trachtenberg notes, though, historians tend to assume, usually rightly, a text's authenticity, and move on to additional inferential hurdles (which, I would add, political scientists have yet to tackle systematically).110

Eighth, when a document seems to present a major piece of evidence supporting a favored hypothesis, a few supplementary searches or robustness checks might be conducted to shore up not the accuracy of the text in itself, but the inference we can draw. Consider a memorandum from a cabinet official to the president, found in a published volume or archival folder organized by subject. We might look at contemporaneous memoranda from the same author to other recipients; we might review a time series of the author's communications to the recipient; we might gather the other side of that conversation; or we might look for inputs to the memorandum, including drafts, notes, recently received information, and other materials available at the time of composition. We need to establish the context, the population of documents from which our document could have been selected. For any seemingly explosive document, there may have been a note shortly thereafter counteracting the previous verdict. Again, the goal is not the comprehensive exploration often pursued by historians, but a thorough exploration within narrowly defined limits, while communicating our research strategy to the reader. Ultimately, causal inference is stronger from a narrow but firm foundation than from an expansive platform plagued by holes, cracks, and unmapped boundaries.

A Test Run: Ickes and the Interior

To illustrate the guidelines laid out above in practice and demonstrate their value, I conduct a replication and extension analysis that targets a key body of evidence in the 1941 debate. Reiter, Schuessler, and Trachtenberg all cite documents from box 55 of the FDR Library PSF series (and from few other locations in this archive), suggesting an area of signal importance for evaluating their competing arguments. The box (digitized, online) contains two folders on the Justice Department and five with chronologically organized correspondence of Harold Ickes, secretary of the interior; the material cited by Reiter, Schuessler, and Trachtenberg comes from the 1941 folder, which consists of 196 page images.111 Reiter cites Ickes' letters to argue that “[t]he historical record does not support the deception narrative” and that Roosevelt and his advisers “sometimes argued that sanctions might deter Japan.”112 Conversely, Schuessler calls one letter (the only archival reference in the FDR chapter in Deceit on the Road to War) “the key piece of back door evidence linked to Ickes”; Schuessler argues that Ickes supported an oil embargo “even if it did provoke Japan because war in the Pacific would provide a back door into the war in Europe,” and he identifies Ickes as one of three “Anti-Japanese hardliners” in Roosevelt's administration.113 Between these positions, Trachtenberg critiqued Reiter's interpretation of the letters, and Reiter downgraded his claim that FDR and his advisers intended the sanctions to deter Japan, stating instead that “Ickes and Roosevelt believed that the sanctions might not provoke a Japanese attack”; Reiter recognized that Ickes “indirectly and vaguely talks about how the embargo might provide an avenue for entry into war, but the scenario Ickes envisions is unclear.”114

Ickes' opaqueness raises larger questions about what the interior secretary believed, whether he had a clear picture of U.S. policy and could speak for Roosevelt, and how much these documents ultimately can tell us. Ickes' letters to FDR now stand in for a large part of the overall debate: Was the administration trying to deter or to provoke Japan? Notably, the three scholars are arguing about interpreting specific texts, rather than considering what inferences to draw from the context in which these documents were embedded. It seems important, for instance, that Ickes wrote the “back door” letter the day after Hitler invaded the Soviet Union, and that a week later, Ickes threatened to resign over criticism (and perceived slights) he had received from FDR in the interim.115 Inferences about Ickes' and FDR's policy positions will be stronger if we examine the cited letters in the context of their chain of communications.

There are reasons to doubt, however, whether Ickes' correspondence is really the best source of evidence for any aspect of the 1941 debate, and whether political scientists should look elsewhere instead. It is not clear what role Ickes played in the foreign policy process, how close a confidant he was of Roosevelt's, and what other scholars have made of Ickes and his archival papers. After all, we might expect the Interior Department to be out of the foreign policy loop. Moreover, historians of FDR's foreign policy have, perhaps tellingly, kept Ickes in the background. Dallek cites Ickes' PSF folders only once in chapter 11, addressing a document written by someone else that Ickes forwarded to FDR; Dallek uses Ickes' diary, but apparently set aside Ickes' archival papers after review.116 Neither Ickes nor the Interior Department appears in Beard's index, whereas foreign policy principals such as Cordell Hull, Frank Knox, and Henry Stimson, and departments such as War, State, and Navy, are well represented.117 These points could be reversed and used explicitly as research strategies. If historians have overlooked Ickes' papers, then perhaps political scientists are exploring new areas. And if the Interior Department received the same policy guidance as the Department of the Navy, then this might be strong evidence against a compartmentalized conspiracy to maneuver the United States into war despite public opinion. A close reading of this folder suggests three main observations that undermine the deception argument and call into question the evidentiary value of Ickes' papers for analyses of FDR's foreign policy.

First, the Ickes-FDR correspondence in the week following the June 23 backdoor letter indicates that Ickes' ideas fell outside the mainstream and that he himself stood outside the foreign policy process. FDR replied tersely, asking whether Ickes' call for an oil embargo on Japan “would continue to be your judgment if this were to tip the delicate scales and cause Japan to decide either to attack Russia or to attack the Dutch East Indies.”118 Ickes responded on June 25 with speculation, self-aggrandizement, bureaucratic sniping, and flattery. Moreover, he avoided the essence of FDR's question—the impact of U.S. policy on Japanese decisionmaking—through redoubled emphasis on oil as a consumable national resource rather than an instrument of international strategy (either for deterrence or provocation). Ickes expected Germany to defeat the Soviet Union, and Japan not to expand until after that; in any event, Japan “will attempt the Dutch East Indies if she chooses, whether or not we are supplying her with oil.”119 After admitting that “[i]t is presumptuous of me to offer you advice in an international matter,” and accusing the State Department of making “mistake after mistake … particularly with respect to Japan,” Ickes declared: “Foreign wars cannot be fought without oil and gasoline, and we are furnishing Japan.” This letter likely crossed in the mail a same-day note from FDR chastising Ickes for overstepping from oil policy into oil export policy, which as a matter of foreign affairs was outside his authority.120 Whatever position Ickes was expressing, he was out of favor with the president at the time.

In a five-page response, Ickes offered to resign immediately as petroleum coordinator for national defense (though not as interior secretary). Ickes complained about “the lack of a friendly tone in letters” from FDR, adding that “it is clear that my services as Petroleum Coordinator do not meet with your approval,” and he accused the State Department of “hostility that is all too apparent” and “pressure, both proper and improper.”121 Ickes' resignation threat prompted FDR to reflect on foreign policy, and mollify his interior secretary, in a two-page letter. “You are doing a grand job as Petroleum Coordinator,” FDR wrote, musing about expanding Ickes' portfolio into coal, water, and other resources. Some of this may be puffery rather than performance evaluation: FDR's comments on foreign policy follow an “it will interest you to know” clause, suggesting that Ickes was not involved in these conversations. The Japanese were arguing internally, Roosevelt explained, over “which way they are going to jump,” against Russia or toward Indonesia, or “whether they will sit on the fence and be more friendly with us. No one knows what the decision will be but, as you know, it is terribly important for the control of the Atlantic for us to help to keep peace in the Pacific.”122 These remarks undercut the back-door argument by hoping for peace in the Pacific and estimating that it might be possible. These are the final comments on Japan in this folder before the July 26 U.S. freeze of Japanese assets, so if Ickes was pushing for escalation with Japan, he apparently did not persuade FDR.

Second, although Ickes consistently displayed a conspiratorial mind-set, he focused not on manipulating foreign adversaries or domestic opinion but rather on uncovering bureaucratic plots against his own position. Alternately paranoid and petulant, Ickes framed perceived threats to himself as assaults on the president's authority, and he employed flattery and emotional manipulation to put FDR on his side in bureaucratic squabbles. When the Navy Department planned to park 5,000 cars indefinitely on the polo field on the National Mall, which as federal land fell under the purview of the Interior Department, Ickes complained of exclusion from the decision, predicted that “[t]his demand of the Navy will doubtless be followed by others,” and warned that “[i]t is fortunate that the parks are nearer to the Navy building than the White House or he might choose to use the grounds there.”123 Ickes expressed alarm that the State Department was supposedly planning to create a Caribbean Division incorporating not just foreign countries but also Puerto Rico and the Virgin Islands, which as U.S. territories fell under Ickes' Interior Department. Based on “several months” of “reports from various sources,” Ickes argued that “the evidence is clear” that the State Department was “planning a surprise raid” while having “done all that it could to prevent any information of its doings from reaching me.”124 He protested moves by the Office of Production Management (OPM) and Vice Chairman of the National Power Policy Committee Leland Olds as “attempting to usurp” his own power as NPPC Chairman and darkly wondered whether Olds and his compatriots were “shortcutting you, as they have been shortcutting this Department for many months?”125 Ickes complained about Commerce Secretary Jesse Jones and OPM planning magnesium plants in Nevada without Interior Department oversight, explaining that he had “suspected” their collusion “[f]or some time” and warning of “a scandal at the end of this Administration” and of having “troubles enough now without wanting to lock horns with the most powerful man in the Administration barring only yourself.”126 Ickes sought FDR's support in transferring the Forest Service from the Agriculture Department to Interior, anticipating resistance from Agriculture Undersecretary Paul Appleby, who “knows how to give lip service while sabotaging effectively,” and the Forest Service, whose leaders “pretend to obey the orders of the Chief Executive … you have had experience with insubordination in this quarter in the past.”127 And when the White House reprimanded Ickes over alleged press leaks from the Interior Department regarding oil for Japan, Ickes hotly denied the accusation, which, he told FDR, “does not make particularly pleasant reading”; he also complained of “getting pretty fed up with incorrect rumors,” and argued that “I should have the name of your informant.”128

Third, Ickes' comments on foreign policy are out of step with FDR's thinking, out of touch with world affairs, and short on strategic calculation. His reports and positions involved a mix of comments from academic correspondents and long-held personal beliefs. In May 1941, Ickes forwarded a letter from a history professor proposing that FDR call for a world peace conference as an overture to Hitler, which, even if it failed, might make subsequent war entry more popular.129 Peace overtures to Germany were far from FDR's intentions in the summer of 1941, though. Roosevelt instead considered calling a “congress or parliament of democratic countries,” but Ickes replied that “now that Russia is with us, even if not of us, this could not be done without great embarrassment and, therefore, it should not be done at all.”130 Roosevelt remained committed to democracy as a foundation of international order, launching with Churchill the Atlantic Charter on August 14. On oil, Ickes focused persistently on resource management (his bureaucratic responsibility) rather than coercion and foreign policy. On November 24, Ickes underscored to FDR that “my consistent attitude from the day that Japan invaded China was that we should not ship a single barrel of oil to Japan.”131 A week later, Ickes expressed concern about Mexican nationalization of U.S. companies' oil assets, conspiratorially suggested that “[t]he time may come when our Government will wish that it owned these properties itself in order to exploit them in the national interest”; in addition, he anticipated U.S. commandeering of oil throughout Latin America to supply depleted reserves and to compete with Germany, which might seize much of the world's oil after beating the Soviet Union.132 These suggestions contradicted FDR's Good Neighbor Policy in Latin America and would likely have upended regional support for the war effort. Moreover, FDR dismissed some of Ickes' fantasies, such as the State Department's mythical Caribbean Division, and tersely underscored the significance of agencies other than the Interior Department as sources of foreign policy advice on Latin American oil.133

Most important, on May 24 Ickes presented the only commentary in this folder on naval operations and Atlantic security. Rhetorically, he showed the same lack of strategic thinking as in his June 25 letter on Japan, by not anticipating that U.S. actions would influence the behavior of an adversary. Ickes argued that “the Germans will not create an incident for us until Hitler is ready to strike and then he will strike, incident or no incident. I know that we cannot cold-bloodedly go to war with Germany, but isn't there something that we could do to clarify the issue, such as declaring a general emergency or announcing that all of our Pacific Fleet has been ordered to go into the Atlantic …?”134 Although it is not clear what he meant by “clarify the issue,” Ickes certainly did not anticipate that withdrawing U.S. naval forces from the Pacific would either embolden the Japanese or make them more cooperative; nor did he think that a more aggressive posture toward Germany would either deter or provoke a German response.

These points allow us to revisit the back-door letter more skeptically. Ickes renewed his call for an oil embargo on Japan, linked this to public opinion (“as popular a move … as you could make”), and connected the public in turn to war entry in Europe: the embargo might “make it, not only possible but easy, to get into this war in an effective way.” And yet, this is hardly conclusive evidence for the back-door thesis. First, Ickes suggested that this is a good moment to do something he had long advocated: cutting off oil to Japan. Second, he suggested that an embargo would not affect Japanese behavior, because “Japan is so preoccupied with what is happening in Russia … that she won't venture a hostile move against the Dutch East Indies.”135

The bottom line is that one of the few sets of documents mustered by both sides of the 1941 debate, when examined in detail and in context, breaks rather sharply against the deception thesis, while also revealing its own limitations as a source of evidence. Harold Ickes' letters do not disprove the deception thesis, but they do indicate that such a causal process did not run through the Interior Department, where Ickes was hardly a foreign policy mastermind. Overall, the documents do indicate a conspiratorial slant—not of the Roosevelt administration's foreign policy decisionmaking, but rather of Ickes' imagination. Future work should strategically target alternative sources that more effectively track key policymakers' engagement with public opinion, naval operations, and other empirical issues listed above. One targeted study cannot resolve the entire 1941 dispute, but it demonstrates the value of selecting source material with inference in mind.


Process tracing through archival research offers a potent method of foreign policy analysis. Primary documents can empower researchers to assess how decisionmakers weighed risks, options, and uncertainty; to follow the subtle currents of leader personality and ideology; and to contextualize and critique the lessons of the past for enduring and contemporary security challenges. What guidance do scholars have for conducting this research most effectively, and what standards should readers use to evaluate whether such work is persuasive? Qualitative research methods literature has firmly established the scientific role of case studies and enumerated best practices for process tracing. Separately, we have advice on conducting archival work and gathering and interpreting other primary and secondary sources. This article seeks to bridge these two themes, and to address emerging debates on research transparency. I argue that political scientists need to explain how our source selection yields the kinds of observations necessary for core descriptive and causal inferences at the case level. Exploring the terrain of documents is akin to conducting other forms of social science research, from ethnographic fieldwork to survey experiments, in that we need a clear idea of where we are going (and where we have been) and why. Political scientists generally do not treat documentary material the way most historians would—as comprehensively as possible, to illuminate and understand particular episodes in context—but rather to gain leverage with which to explain case studies, test hypotheses, and assess and improve theories. Differing objectives imply different methods, but this requires political scientists to develop a new template for effective analysis. Ultimately, we need to be more reflective, transparent, and strategic about our selective use of available documentation.

To this end, I offered eight suggestions for improving qualitative research with documentary primary sources. Scholars should cite their sources in detail for replication, but we should also clarify the main directions of sources explored but not cited. More broadly, we need to articulate a research strategy, why we headed one direction rather than another in search of evidence. Strategies might include following where others have worked, breaking new ground elsewhere, or following our knowledge of the policy process to target particular sources. When we have relevant sources in hand, we should clarify their specific relationship to our overall claims about the causal process, and provide some robustness checks by probing the communicative context of key documents.

To demonstrate how document-driven case studies work in practice in security studies, and critique the limitations of our existing techniques and norms, I reviewed the exemplary security studies debate over Franklin Roosevelt's foreign policy in the months before Pearl Harbor. Recent scholarship by Marc Trachtenberg, John Schuessler, and Dan Reiter has brought competing theories to bear on a significant historical case. Did FDR escalate foreign conflicts and manipulate public opinion to bring the United States into World War II, or not? The deception question matters, both for theoretical arguments between realists and liberals in the field of international relations, and for contemporary concerns about democratic representation and the conduct of U.S. foreign policy. The FDR debate increasingly employs primary documents, and the conversation makes compelling reading. A shared opaqueness about research strategy and source selection, however, leads to scholarly stalemate: each side cites incisive but offsetting quotations, with little discussion of why these sources instead of others are core evidence. I identified several readily available and seemingly relevant sources not cited in this debate, including some cited by historians of the Roosevelt era. This array of potential evidence not only indicates how partial and informal our selection processes have been in political science, but also points out areas for targeted future research on Atlantic naval operations, policy toward Japan, and the role of public opinion in FDR's decisionmaking. In that spirit, applying some of my research guidelines, I examined a set of documents that all three scholars have cited (correspondence involving FDR's secretary of the interior, Harold Ickes, online at the FDR Presidential Library), and I argued that this small replication study undercuts the deception thesis and underscores the importance of source selection for effective research design. In the context of Ickes' contemporaneous correspondence, his explosive quotations about producing an “incident” with Germany while pushing for sanctions on Japan appear as innuendos among innuendos, from a bureaucratic infighter rather than a foreign policy strategist. This analysis is hardly definitive, but it suggests how rebuttals and other future work might proceed: not by simply digging up yet more quotations, but by specifying which sources should matter most for the particular disputes, and prioritizing these over others for examination and for inference.

Archival research is inductive, incremental, and inspirational.136 Scholars undertaking it are likely to encounter files misplaced and mislabeled, and to doggedly pursue hunches into piles of dross, but also to discover illuminating remarks in a long-overlooked box of miscellany. Yet, serendipity is not a strategy, and mere citation of primary sources is no talisman of excellence in qualitative research. Nor is the accumulation of archival documents a panacea, particularly as declassified material proliferates online. Qualitative research projects with documentary sources need design, not just depth. We cannot rely on other scholars to check our work against misinterpretation of the evidence, but we should facilitate and encourage more replication analyses. We cannot expect future declassification to set free the truth, but we should specify how new releases might put our claims to the test. And even at the expense of some methodological scaffolding alongside the main narrative of a case study, scholars need to show our work, dead ends and all. Ultimately, this will make security studies findings more cumulative, debates more acute, and research more persuasive.


For comments during revisions of this article, the author thanks Matthew Green, Laura Mayhall, Afshon Ostovar, John Schuessler, Wallace Thies, Andrew Yeo, the anonymous reviewers, and particularly Brendan Green and Emily Meierding. He would also like to thank graduate students in his courses on qualitative methods and U.S. foreign policy at the Naval Postgraduate School and the Catholic University of America for spirited discussions of arguments and evidence in the debate over President Franklin Roosevelt's decisionmaking about whether to take the United States to war. The views in this article are those of the author and do not represent the position of the U.S. Navy or any other government entity.



Recent examples include Stacie E. Goddard, “The Rhetoric of Appeasement: Hitler's Legitimation and British Foreign Policy, 1938–39,” Security Studies, Vol. 24, No. 1 (2015), pp. 95–130, doi:10.1080/09636412.2015.1001216; Galen Jackson, “The Showdown That Wasn't: U.S.-Israeli Relations and American Domestic Politics, 1973–75,” International Security, Vol. 39, No. 4 (Spring 2015), pp. 130–169, doi:10.1162/ISEC_a_00201; Jonathan N. Brown, “Immovable Positions: Public Acknowledgment and Bargaining in Military Basing Negotiations,” Security Studies, Vol. 23, No. 2 (2014), pp. 258–292, doi:10.1080/09636412.2014.905351; and Keren Yarhi-Milo, “In the Eye of the Beholder: How Leaders and Intelligence Communities Assess the Intentions of Adversaries,” International Security, Vol. 38, No. 1 (Summer 2013), pp. 7–51, doi:10.1162/ISEC_a_00128.


Deborah Welch Larson, “Sources and Methods in Cold War History: The Need for a New Theory-Based Archival Approach,” in Colin Elman and Miriam Fendius Elman, eds., Bridges and Boundaries: Historians, Political Scientists, and the Study of International Relations (Cambridge, Mass.: MIT Press, 2001), pp. 328–336, 339–341, 350, at 339.


Alexander L. George and Andrew Bennett, Case Studies and Theory Development in the Social Sciences (Cambridge, Mass.: MIT Press, 2005), pp. 103–104.


Larson, “Sources and Methods in Cold War History,” pp. 342–343.


Marc Trachtenberg, The Craft of International History: A Guide to Method (Princeton, N.J.: Princeton University Press, 2006), especially chap. 5; and Scott A. Frisch et al., eds., Doing Archival Research in Political Science (Amherst, N.Y.: Cambria, 2012). See also Diana Kapiszewski, Lauren MacLean, and Benjamin L. Read, Field Research in Political Science: Practices and Principles (Cambridge: Cambridge University Press, 2015), chap. 5.


Trachtenberg, The Craft of International History, p. 59.


Cameron G. Thies, “A Pragmatic Guide to Qualitative Historical Analysis in the Study of International Relations,” International Studies Perspectives, Vol. 3, No. 4 (November 2002), especially pp. 353–359, doi:10.1111/1528-3577.t01-1-00099; and Ian S. Lustick, “History, Historiography, and Political Science: Multiple Historical Records and the Problem of Selection Bias,” American Political Science Review, Vol. 90, No. 3 (September 1996), especially pp. 608, 613–615, doi:10.2307/2082612.


Trachtenberg, The Craft of International History, chap. 4. Trachtenberg explains that chapter 4 “is not a finished product,” but “leave[s] a lot of the scaffolding up … to show what goes into an historical interpretation.” See ibid., p. viii. See also John M. Schuessler, “The Deception Dividend: FDR's Undeclared War,” International Security, Vol. 34, No. 4 (Spring 2010), pp. 133–165, doi:10.1162/isec.2010.34.4.133; John M. Schuessler, Deceit on the Road to War: Presidents, Politics, and American Democracy (Ithaca, N.Y.: Cornell University Press, 2015), especially chap. 2; and Marc Trachtenberg, “Dan Reiter and America's Road to War in 1941,” in H-Diplo/ISSF Roundtable 5-4, “Democracy, Deception, and Entry into War,” H-Net, May 17, 2013, http://h-net.org/~diplo/ISSF/PDF/ISSF-Roundtable-5-4.pdf. See also John M. Schuessler, “Correspondence: FDR, U.S. Entry into World War II, and Selection Effects Theory,” International Security, Vol. 35, No. 2 (Fall 2010), pp. 181–185, doi:10.1162/ISEC_c_00017; and John Schuessler, “Democracy, Deception, and IR Theory,” in H-Diplo/ISSF Roundtable 5-4.


Dan Reiter, “Democracy, Deception, and Entry into War,” Security Studies, Vol. 21, No. 4 (2012), pp. 594–623, doi:10.1080/09636412.2012.734229. See also Dan Reiter, “Correspondence: FDR, U.S. Entry into World War II, and Selection Effects Theory,” International Security, Vol. 35, No. 2 (Fall 2010), pp. 176–181, doi:10.1162/ISEC_c_00017; and Dan Reiter, “Response to Trachtenberg, Schuessler, and Kaiser,” in H-Diplo/ISSF Roundtable 5-4. Ronald R. Krebs's portrayal of FDR facing meaningful domestic opposition to internationalism and striving to persuade the public (unsuccessfully prior to Pearl Harbor though brilliantly thereafter) also fits this perspective. See Krebs, “Tell Me a Story: FDR, Narrative, and the Making of the Second World War,” Security Studies, Vol. 24, No. 1 (2015), especially pp. 151–155, doi:10.1080/09636412.2015.1001215.


Many authors claim that consolidated liberal democracies face domestic constraints on foreign policy decisionmaking, especially through public opinion. These constraints may be beneficial by keeping democracies from fighting one another and from involvement in unnecessary or unwinnable wars; by helping mobilize to win the wars that must be fought; by making governments' threats more credible and bargaining more effective (because leaders face “audience costs” imposed by the public's ability to punish poor decisions by removing leaders from office); and by enhancing policy deliberation and learning through a marketplace of ideas. In contrast, several authors maintain that democracies behave like other states, especially in situations of acute threat, and that leaders rarely pay audience costs in practice and can circumvent or restructure the “marketplace of ideas.” See, for example, Marc Trachtenberg, “Audience Costs: An Historical Analysis,” Security Studies, Vol. 21, No. 1 (2012), pp. 3–42, doi:10.1080/09636412.2012.650590; Jack Snyder and Erica D. Borghard, “The Cost of Empty Threats: A Penny, Not a Pound,” American Political Science Review, Vol. 105, No. 3 (August 2011), pp. 437–456, doi:10.1017/S000305541100027X; John J. Mearsheimer, Why Leaders Lie: The Truth about Lying in International Politics (Oxford: Oxford University Press, 2011); Chaim Kaufmann, “Threat Inflation and the Failure of the Marketplace of Ideas: The Selling of the Iraq War,” International Security, Vol. 29, No. 1 (Summer 2004), pp. 5–48, doi:10.1162/0162288041762940; Dan Reiter and Allan C. Stam, Democracies at War (Princeton, N.J.: Princeton University Press, 2002); Michael C. Desch, “Democracy and Victory: Why Regime Type Hardly Matters,” International Security, Vol. 27, No. 2 (Fall 2002), pp. 5–47, doi:10.1162/016228802760987815; and Kenneth A. Schultz, Democracy and Coercive Diplomacy (Cambridge: Cambridge University Press, 2001).


“Roosevelt's actions … are not easy to explain”; studying them “gives one the feeling of peering into a kaleidoscope.” See Robert Dallek, Franklin Delano Roosevelt and American Foreign Policy, 1932–1945, rev. ed. (Oxford: Oxford University Press, 1995), p. vii. But see also James MacGregor Burns, Roosevelt: The Soldier of Freedom (New York: Harcourt Brace Jovanovich, 1970), pp. 607–608. Similarly, see Robert E. Sherwood, Roosevelt and Hopkins: An Intimate History (New York: Harper and Brothers, 1948), p. 9; and Hamilton Basso, “The Roosevelt Legend,” Life, November 3, 1947, pp. 126–147, especially p. 127. Both works draw on Roosevelt's advisers' early reflections.


Gary King, Robert O. Keohane, and Sidney Verba, Designing Social Inquiry: Scientific Inference in Qualitative Research (Princeton, N.J.: Princeton University Press, 1994), pp. 46, 79. See also W. Phillips Shively, The Craft of Political Research, 8th ed. (Boston: Pearson, 2011), pp. 75–78.


See King, Keohane, and Verba, Designing Social Inquiry, pp. 32–33, 200–206; Shively, The Craft of Political Research, pp. 78, 86, 90; and Stephen Van Evera, Guide to Methods for Students of Political Science (Ithaca, N.Y.: Cornell University Press, 1997), pp. 32–39.


On data matching, natural experiments, regression discontinuities, instrumental variables, and stratification, see Jasjeet S. Sekhon, “The Neyman-Rubin Model of Causal Inference and Estimation via Matching Methods,” in Janet M. Box-Steffensmeier, Henry E. Brady, and David Collier, eds., The Oxford Handbook of Political Methodology (Oxford: Oxford University Press, 2008), pp. 271–294; John Gerring, Social Science Methodology: A Unified Framework, 2nd ed. (Cambridge: Cambridge University Press, 2012), chap. 11; and Thad Dunning, Natural Experiments in the Social Sciences: A Design-Based Approach (Cambridge: Cambridge University Press, 2012), pp. 25, 63–68, 87–90, 105–107.


On concept formation to clarify scope conditions, qualitative comparative analysis and configuration, and paired comparisons, see Gary Goertz, Social Science Concepts: A User's Guide (Princeton, N.J.: Princeton University Press, 2005); Charles C. Ragin, Fuzzy-Set Social Science (Chicago: University of Chicago Press, 2000); and Sidney Tarrow, “The Strategy of Paired Comparison: Toward a Theory of Practice,” Comparative Political Studies, Vol. 43, No. 2 (February 2010), pp. 230–259, doi:10.1177/0010414009350044.


James Mahoney and Gary Goertz, “A Tale of Two Cultures: Contrasting Quantitative and Qualitative Research,” Political Analysis, Vol. 14, No. 3 (Summer 2006), especially pp. 230–232, doi:10.1093/pan/mpj017.


David Waldner, “What Makes Process Tracing Good?” in Andrew Bennett and Jeffrey T. Checkel, eds., Process Tracing: From Metaphor to Analytic Tool (Cambridge: Cambridge University Press, 2013), pp. 128–129 (emphasis in the original); and George and Bennett, Case Studies and Theory Development in the Social Sciences, pp. 206–207.


See David Collier, Henry E. Brady, and Jason Seawright, “Sources of Leverage in Causal Inference,” in Brady and Collier, eds., Rethinking Social Inquiry: Diverse Tools, Shared Standards, 2nd ed. (Lanham, Md.: Rowman and Littlefield, 2010), especially pp. 184–185, 195–196.


Tasha Fairfield and Andrew E. Charman, “Explicit Bayesian Analysis for Process Tracing: Guidelines, Opportunities, and Caveats,” Political Analysis, Vol. 25, No. 3 (July 2017), pp. 363–380, especially p. 365, doi:10.1017/pan.2017.14. On “clues,” see Macartan Humphreys and Alan M. Jacobs, “Mixing Methods: A Bayesian Approach,” American Political Science Review, Vol. 109, No. 4 (November 2015), p. 656, doi:10.1017/S0003055415000453.


On evidence, see Richard Ned Lebow, Forbidden Fruit: Counterfactuals and International Relations (Princeton, N.J.: Princeton University Press, 2010); and Jack S. Levy, “Counterfactuals and Case Studies,” in Box-Steffensmeier, Brady, and Collier, The Oxford Handbook of Political Methodology, especially pp. 636–638. On surprise, diversity, and adequacy, see Andrew Bennett, “Process Tracing: A Bayesian Perspective,” in Box-Steffensmeier, Brady, and Collier, The Oxford Handbook of Political Methodology, especially pp. 704, 710, 713. See also Fairfield and Charman, “Explicit Bayesian Analysis for Process Tracing,” pp. 366–368. Notably, both counterfactual and Bayesian approaches also involve a hypothetical alternative case, somewhat substituting for cross-case comparison.


See Andrew Bennett and Jeffrey T. Checkel, “Process Tracing: From Philosophical Roots to Best Practices,” in Bennet and Checkel, Process Tracing, pp. 4–5, 11; George and Bennett, Case Studies and Theory Development in the Social Sciences, pp. 109–110, 137; and Derek Beach and Rasmus Brun Pedersen, Process-Tracing Methods: Foundations and Guidelines (Ann Arbor: University of Michigan Press, 2013), pp. 68–69.


Contrast Bennett and Checkel, “Process Tracing,” pp. 24–25; and Trachtenberg, The Craft of International History, pp. 146–147.


Gerring, Social Science Methodology, p. 332, aptly notes that causal process observations (CPOs) are “sometimes more appropriately labeled as descriptive rather than causal.” Dunning, Natural Experiments in the Social Sciences, pp. 208–230, explains how CPOs yield pieces of evidence on the presence or magnitude of particular variables or the operation of specific mechanisms. Beach and Pedersen, Process-Tracing Methods, p. 73, similarly distinguish between evidence and observations. Collier, Brady, and Seawright, “Sources of Leverage in Causal Inference,” p. 185, usefully note that CPOs could be called “causal-process information“ (emphasis in original).


Van Evera, Guide to Methods for Students of Political Science, p. 34; and Humphreys and Jacobs, “Mixing Methods,” pp. 657–659.


Lustick, “History, Historiography, and Political Science,” p. 608.


Andrew Moravcsik, “Active Citation: A Precondition for Replicable Qualitative Research,” PS: Political Science & Politics, Vol. 43, No. 1 (January 2010), pp. 29–30, 34, doi:10.1017/S1049096509990783.


Jack Snyder, “Active Citation: In Search of Smoking Guns or Meaningful Context?” Security Studies, Vol. 23, No. 4 (2014), pp. 708–709, doi:10.1080/09636412.2014.970409. See also Elizabeth N. Saunders, “Transparency without Tears: A Pragmatic Approach to Transparent Security Studies Research,” Security Studies, Vol. 23, No. 4 (2014), p. 694, doi:10.1080/09636412.2014.970405.


See work on active citation, annotation for transparent inquiry, and online appendices and data repositories, for example in Andrew Moravcsik, “Trust, but Verify: The Transparency Revolution and Qualitative International Relations,” Security Studies, Vol. 23, No. 4 (2014), pp. 663–688, doi:10.1080/09636412.2014.970846; Jeffrey C. Isaac, “For a More Public Political Science,” Perspectives on Politics, Vol. 13, No. 2 (June 2015), pp. 269–283, doi:10.1017/S1537592715000031; Tim Büthe and Alan M. Jacobs, eds., Qualitative & Multi-Method Research, Vol. 13, No. 1 (Spring 2015); James A. Morrison and Joanne Yao, eds., International History and Politics Newsletter, Vol. 1, No. 2 (Winter 2016); and Sebastian Karcher, “Qualitative Data Repository Teams with Hypothesis to Develop Annotation for Transparent Inquiry” (Syracuse, N.Y.: Qualitative Data Repository, Center for Qualitative and Multi-Method Inquiry, Maxwell School of Citizenship and Public Affairs at Syracuse University, May 2, 2017), https://qdr.syr.edu/qdr-blog/qualitative-data-repository-teamshypothesis-develop-annotation-transparent-inquiry-ati. For the evolving political science conversation about transparency and its trade-offs, see, for example, http://www.dartstatement.org; and https://www.qualtd.net.


For discussion of how much context to present in active citations, see Diana Kapiszewski and Dessislava Kirilova, “Transparency in Qualitative Security Studies Research: Standards, Benefits, and Challenges,” Security Studies, Vol. 23, No. 4 (2014), p. 702, doi:10.1080/09636412.2014.970408; and Saunders, “Transparency without Tears,” pp. 692–696.


See R.G. Collingwood, The Idea of History (Oxford: Oxford University Press, 1946), especially pp. 250–255, 261–263, 268, 282; Edward Hallett Carr, What Is History? (New York: Alfred A. Knopf, 1962), pp. 13–20, 24, 136–138; and John D. Milligan, “The Treatment of an Historical Source,” History and Theory, Vol. 18, No. 2 (May 1979), p. 194, doi:10.2307/2504755.


See, for example, Marc Bloch, The Historian's Craft: Reflections on the Nature and Uses of History and the Techniques and Methods of Those Who Write It, Peter Putnam, trans. (New York: Alfred A. Knopf, 1953), pp. 54–55, 71–78, 166; and Laura Mayhall, “Creating the ‘Suffragette Spirit’: British Feminism and the Historical Imagination,” in Antoinette Burton, ed., Archive Stories: Fact, Fiction, and the Writing of History (Durham, N.C.: Duke University Press, 2005), pp. 232–250.


Trachtenberg, The Craft of International History, pp. 234, 243; and Marc Trachtenberg, “Transparency in Practice: Using Written Sources,” Qualitative & Multi-Method Research, Vol. 13, No. 1 (Spring 2015), p. 15.


See, for example, Philip Marfleet, “Forgotten by History: Refugees, Historians, and Museums in Britain,” in Amy K. Levin, ed., Global Mobilities: Refugees, Exiles, and Immigrants in Museums and Archives (New York: Routledge, 2017), pp. 73–84, especially 73–74; Anjali Arondekar, “Without a Trace: Sexuality and the Colonial Archive,” Journal of the History of Sexuality, Vol. 14, Nos. 1–2 (January/April 2θ05), pp. 10–27, especially pp. 11–15; Carolyn Steedman, Dust: The Archive and Cultural History (New Brunswick, N.J.: Rutgers University Press, 2002); and Burton, Archive Stories.


See, for example, Lien-Hang Nguyen, “Revolutionary Circuits: Toward Internationalizing America in the World,” Diplomatic History, Vol. 39, No. 3 (June 2015), pp. 411–422, doi:10.1093/dh/dhv026; and Tony Smith, “New Bottles for New Wine: A Pericentric Framework for the Study of the Cold War,” Diplomatic History, Vol. 24, No. 4 (Fall 2000), pp. 567–591, doi:10.1111/0145-2096.00237.


Bloch, The Historian's Craft, p. 71.


John Lewis Gaddis, “In Defense of Particular Generalization: Rewriting Cold War History, Rethinking International Relations Theory,” in Elman and Elman, Bridges and Boundaries, pp. 307–311, 322. See also Trachtenberg, “Transparency in Practice,” pp. 14–16.


Samuel Eliot Morison, “Faith of a Historian,” American Historical Review, Vol. 56, No. 2 (January 1951), pp. 263–264. For a similar view, see Dallek, Franklin Delano Roosevelt and American Foreign Policy, 1932–1945, p. vii. See also Trachtenberg, “Transparency in Practice,” pp. 15–16, on scholarly “judgment.”


See, for example, Robert Jervis, “International History and International Politics: Why Are They Studied Differently?” in Elman and Elman, Bridges and Boundaries, pp. 389–393; Paul Schroeder, “International History: Why Historians Do It Differently than Political Scientists,” in Elman and Elman, Bridges and Boundaries, pp. 405–409; and Gaddis, “In Defense of Particular Generalization,” pp. 312–314, 319–320.


See Trachtenberg, The Craft of International History, chap. 4; Trachtenberg, “Dan Reiter and America's Road to War in 1941”; Schuessler, “The Deception Dividend”; and Schuessler, Deceit on the Road to War, chap. 2.


See Reiter, “Democracy, Deception, and Entry into War.”


Trachtenberg, The Craft of International History, pp. 58–60, 73, 79, 95, 137, at pp. 59, 95.


Reiter, “Democracy, Deception, and Entry into War,” p. 597.


See, for example, Samuel Eliot Morison, History of United States Naval Operations in World War II, Vol. 1: The Battle of the Atlantic, September 1939–May 1943 (Boston: Little, Brown, 1947).


In September 1940, Britain traded basing rights for used destroyers. Argentia became a U.S. naval operating base in July 1941. On Argentia, see Theodore A. Wilson, The First Summit: Roosevelt and Churchill at Placentia Bay 1941 (Boston: Houghton Mifflin, 1969).


Reiter, “Correspondence,” p. 178, citing Gerhard L. Weinberg, A World at Arms: A Global History of World War II, 2nd ed. (Cambridge: Cambridge University Press, 2005), p. xix.


Schuessler, “Correspondence,” pp. 183–184, citing Gerhard Weinberg, A World at Arms: A Global History of World War II, 1st ed. (Cambridge: Cambridge University Press, 1994), p. 240, and Trachtenberg, The Craft of International History, pp. 84–87. Note that Schuessler cites the first edition of Weinberg's book, whereas Reiter cites the second edition.


Weinberg, A World at Arms, 1st ed., pp. 83 n. 139, 240 n. 227, bibliography on p. 957, and bibliographic entry on p. 997.


Trachtenberg, The Craft of International History, p. 86.


Reiter, “Democracy, Deception, and Entry into War,” p. 609 n. 57.


Jürgen Rohwer, “Die USA und die Schlacht im Atlantik” [The U.S.A. and the Battle of the Atlantic], in Rohwer and Eberhard Jäckel, eds., Kriegswende Dezember 1941 [The war's turning point: December 1941] (Koblenz, Germany: Bernard and Graefe, 1984), pp. 81–103.


Ibid., p. 79.


The online appendix is available at doi:10.7910/DVN/GMYGZT/.


Wilson, The First Summit, pp. xv–xvi. See online appendix.


Beard, President Roosevelt and the Coming of the War, 1941, pp. 477–478 n. 10, citing Elliott Roosevelt, As He Saw It (New York: Duell, Sloan and Pearce, 1964), chap. 2.


See Schuessler, “Correspondence,” p. 183; and Trachtenberg, The Craft of International History, p. 82. Reiter critiques the reliance on Churchill's account, but two pages later also uses British documents for evidence on FDR's Atlantic policy. See Reiter, “Democracy, Deception, and Entry into War,” pp. 612–613.


See online appendix.


Papers Relating to the Foreign Relations of the United States (FRUS), Japan: 1931–1941, 2 vols. (Washington, D.C.: Government Printing Office [GPO], 1943). All FRUS volumes cited in this article were accessed online via the University of Wisconsin–Madison Libraries digital collections: http://digicoll.library.wisc.edu/cgi-bin/FRUS/FRUS-idx?type=browse&scope=FRUS.FRUS1.


Trachtenberg, The Craft of International History, pp. 121 n. 151, 128 n. 167; and FRUS, Diplomatic Papers, 1941, Vol. 4: The Far East (Washington, D.C.: GPO, 1956).


Trachtenberg, The Craft of International History, p. 90 n. 27; and Paul W. Schroeder, The Axis Alliance and Japanese-American Relations, 1941 (Ithaca, N.Y.: Cornell University Press, 1958), p. 200. See also Schuessler, “The Deception Dividend,” pp. 152, 160.


See FRUS, Diplomatic Papers, 1941, Vol. 5: The Far East (Washington, D.C.: GPO, 1956).


“Preface,” FRUS, Japan: 1931–1941, Vol. 1, p. iii.


FRUS, Diplomatic Papers, 1941, Vol. 1: General; the Soviet Union (Washington, D.C., GPO, 1959); FRUS, Diplomatic Papers, 1941, Vol. 2: Europe (Washington, D.C.: GPO, 1959); and FRUS, Diplomatic Papers, 1941, Vol. 3: The British Commonwealth; the Near East and Africa (Washington, D.C.: GPO, 1959).


FRUS, Diplomatic Papers, 1941, Vol. 4, index.


President's Secretary's File, Franklin Delano Roosevelt Library, Hyde Park, New York (henceforth PSF FDRL), http://www.fdrlibrary.marist.edu/archives/collections/franklin/?p=collections/findingaid&id=50. See online appendix to this article. Additionally, Trachtenberg hyperlinks an image of a page from a document in box 31, PSF FDRL, with FDR's handwritten marginalia to supplement a secondary source that Trachtenberg cites. See Trachtenberg, The Craft of International History, p. 133 n. 185.


An overview of the FDRL collections estimates that each linear foot has 2,000 pages and that the PSF is 130 linear feet long. See “Collections in the FDR Library Archives,” FDRL, http://www.fdrlibrary.marist.edu/archives/collections/list.html. Measurements given in finding aids for the collection are lower. See PSF finding aid (PDF version), http://www.fdrlibrary.marist.edu/archives/pdfs/findingaids/findingaid_roos_psf.pdf; and PSF finding aid (online version), http://www.fdrlibrary.marist.edu/archives/collections/franklin/index.php?p=collections/findingaid&id=502.


See folder “Japan, October–December 1941,” box 43, PSF FDRL, http://www.fdrlibrary.marist.edu/_resources/images/psf/psfa0409.pdf (220 pages; compare notes on pp. 4, 9, 11, and 23); and folder “Japan, January–September 1941,” box 43, PSF FDRL, http://www.fdrlibrary.marist.edu/_resources/images/psf/psfa0408.pdf.


Some documents may appear in multiple places, however. See folder “Dispatches—Japan, February 14, 1940–March 13, 1941,” box 22, PSF FDRL, http://www.fdrlibrary.marist.edu/_resources/images/psf/psfa0222.pdf (70 pages); and “Dispatches—Japan, March 13, 1941-November 17, 1941,” box 22, PSF FDRL, http://www.fdrlibrary.marist.edu/_resources/images/psf/psfa0223.pdf (58 pages).


For the folder list, see PSF finding aid (online version).


“Collections in the FDR Library Archives.” This list omits the Grace Tully Collection. See PSF finding aid (online version); and Grace Tully Collection finding aid, FDRL, http://www.fdrlibrary.marist.edu/archives/collections/franklin/?p=collections/findmgaid&id=50.


Linear feet listed in “Collections in the FDR Library Archives.” For page estimates, see President's Personal File 1933–1945 finding aid, FDRL, http://www.fdrlibrary.marist.edu/archives/pdfs/findingaids/ppf_description.pdf; and Papers and President Official File finding aid, FDRL, p. 1, http://www.fdrlibrary.marist.edu/archives/pdfs/findingaids/findingaid_of_description.pdf.


See notes to Dallek, Franklin Delano Roosevelt and American Foreign Policy, 1932–1945, chap. 11. See also the online appendix to this article.


See Collection Description and five finding aids, FDRL, http://www.fdrlibrary.marist.edu/archives/collections/list.html (456 pages).


Numerical Index OF 1–OF 2510, OF FDRL, p. 13, http://www.fdrlibrary.marist.edu/archives/pdfs/findingaids/official_num_a.pdf.


The historiography of FDR's foreign policy and U.S. war entry is massive. For overviews, see Michael A. Barnhart, “The Origins of the Second World War in Asia and the Pacific: Synthesis Impossible?” in Michael J. Hogan, ed., Paths to Power: The Historiography of American Foreign Relations to 1941 (Cambridge: Cambridge University Press, 2000), pp. 268–296; and Justus D. Doenecke, “The United States and the European War, 1939–1941: A Historiographical Review,” in Hogan, Paths to Power, pp. 224–267.


See Charles A. Beard, President Roosevelt and the Coming of the War, 1941: A Study in Appearances and Realities (New Haven, Conn.: Yale University Press, 1948).


Schuessler, “The Deception Dividend,” p. 145 n. 56.


Trachtenberg, The Craft of International History, chap. 4; and Trachtenberg, “Dan Reiter and America's Road to War in 1941.”


Reiter, “Democracy, Deception, and Entry into War,” p. 596 n. 5, 612 n. 69


James McAllister, “Introduction,” in H-Diplo/ISSF Roundtable 5-4, p. 2.


Samuel Eliot Morison, “Did Roosevelt Start the War? History through a Beard,” Atlantic, August 1948, pp. 92–93, 96.


Morison, “Faith of a Historian,” pp. 267–268.


Ibid., pp. 263, 268–270.


Joint Committee on the Investigation of the Pearl Harbor Attack, Pearl Harbor Attack, Hearing before the Joint Committee on the Investigation of the Pearl Harbor Attack, Congress of the United States, 79th Cong., 1945/46, http://www.ibiblio.org/pha/congress/. Volumes 12–20 reprint the 183 “exhibits.”


Trachtenberg, “Dan Reiter and America's Road to War in 1941.” Trachtenberg, The Craft of International History, p. 86 n. 20, cites the hearings for a participant's postwar testimony, but not the documentary exhibits.


Beard, President Roosevelt and the Coming of the War, 1941, pp. 374 n. 44, 519 n. 6, 418 n. 22, 420 n. 24, 509 n. 29, 511 n. 33, 523–527, inter alia.


Ibid., pp. 194 n. 26, 560 n. 80.


Ibid., pp. 487–488.


Ibid., pp. 536 n. 35, 544–545.


Ibid., pp. 420 n. 24 (emphasis in the original), 453 n. 5.


Ibid., p. 573, inter alia.


Trachtenberg, The Craft of International History, p. 81.


Schuessler, “The Deception Dividend,” p. 149.


Reiter, “Correspondence”; and Reiter, “Democracy, Deception, and Entry into War.”


Dallek, Franklin Delano Roosevelt and American Foreign Policy, 1932–1945, chap. 11.


Beard, President Roosevelt and the Coming of the War, 1941, pp. 407 (including n. 2), 483 (including n. 1), 504 n. 23, 575 n. 5.


Schuessler, “The Deception Dividend,” p. 165; and Reiter, “Response to Trachtenberg, Schuessler, and Kaiser,” p. 63. See also Schuessler, Deceit on the Road to War, chap. 4.


Trachtenberg, “Dan Reiter and America's Road to War in 1941,” pp. 9 n. 18, 123–124.


Schuessler, “The Deception Dividend,” p. 145. Reiter, “Correspondence,” p. 179, criticizes Schuessler's “circumstantial” line.


Particular archives' specific format recommendations can also be accommodated, for example by citing a physical call location rather than a series box number.


See Trachtenberg, “Transparency in Practice,” p. 14.


See James McAllister, “Who Lost Vietnam? Soldiers, Civilians, and U.S. Military Strategy,” International Security, Vol. 35, No. 3 (Winter 2010/11), p. 122, doi:10.1162/ISEC_a_00024.


See Andrew Moravcsik, The Choice for Europe: Social Purpose and State Power from Messina to Maastricht (Ithaca, N.Y.: Cornell University Press, 1998), especially pp. 78–85, 173–178, 244–245; Sebastian Rosato, Europe United: Power Politics and the Making of the European Community (Ithaca, N.Y.: Cornell University Press, 2011); Robert H. Lieshout, Mathieu L.L. Segers, and Johanna M. van der Vleuten, “De Gaulle, Moravcsik, and The Choice for Europe: Soft Sources, Weak Evidence,” Journal of Cold War Studies, Vol. 6, No. 4 (Fall 2004), pp. 89–139, doi:10.1162/1520397042350900; Andrew Moravcsik, “Did Power Politics Cause European Integration? Realist Theory Meets Qualitative Methods,” Security Studies, Vol. 22, No. 4 (2013), especially pp. 781–782, 786–787, doi:10.1080/09636412.2013.844511; and Sebastian Rosato, “Theory and Evidence in Europe United: A Response to My Critics,” Security Studies, Vol. 22, No. 4 (2013), especially pp. 810–816, doi:10.1080/09636412.2013.844513.


Fred I. Greenstein, “Eisenhower as an Activist President: A Look at New Evidence,” Political Science Quarterly, Vol. 94, No. 4 (Winter 1979/80), especially pp. 576, 587–93, doi:10.2307/2149627; John Lewis Gaddis, We Now Know: Rethinking Cold War History (Oxford: Oxford University Press, 1997); and Caitlin Talmadge, “The Puzzle of Personalist Performance: Iraqi Battlefield Effectiveness in the Iran-Iraq War,” Security Studies, Vol. 22, No. 2 (2013), pp. 180–221, doi:10.1080/09636412.2013.786911.


On triangulation, see Gerring, Social Science Methodology, pp. 362–366, 382–386; and Elizabeth Levy Paluck, “The Promising Integration of Qualitative Methods and Field Experiments,” Annals of the American Academy of Political and Social Science, March 2010, especially pp. 62–63, doi:10.1177/0002716209351510.


On the limits of triangulation, see Sharon Crasnow, “The Role of Case Study Research in Political Science: Evidence for Causal Claims,” Philosophy of Science, Vol. 79, No. 5 (December 2012), p. 664, doi:10.1086/667869; Evan S. Lieberman, “Nested Analysis as a Mixed-Method Strategy for Comparative Research,” American Political Science Review, Vol. 99, No. 3 (August 2005), p. 435, doi:10.1017/S0003055405051762; and Gerring, Social Science Methodology, p. 386.


See Larson, “Sources and Methods in Cold War History,” pp. 345–347.


Frank Schimmelfennig, “Efficient Process Tracing: Analyzing the Causal Mechanisms of European Integration,” in Bennett and Checkel, Process Tracing, especially pp. 101–108. See Saunders, “Transparency without Tears,” on pragmatism.


Conversely, scholars might explore archival boxes or printed documents unmentioned by such tools; see fifth suggestion, above.


See, for example, George and Bennett, Case Studies and Theory Development in the Social Sciences, pp. 97–100, 107–108; Larson, “Sources and Methods in Cold War History,” pp. 344–345; Snyder, “Active Citation,” p. 708; and Milligan, “The Treatment of an Historical Source,” especially pp. 182–187, 192–193.


Trachtenberg, The Craft of International History, pp. 146–147.


Folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL, http://www.fdrlibrary.marist.edu/_resources/images/psf/b-psfc000035.pdf.


Reiter, “Democracy, Deception, and Entry into War,” p. 615. Reiter also cites the British ambassador's report of a conversation with FDR. See Trachtenberg, “Dan Reiter and America's Road to War in 1941,” pp. 32–33; and Reiter, “Response to Trachtenberg, Schuessler, and Kaiser.”


Schuessler, Deceit on the Road to War, p. 142 n. 122.


Trachtenberg, “Dan Reiter and America's Road to War in 1941,” pp. 32–33; and Reiter, “Response to Trachtenberg, Schuessler, and Kaiser,” pp. 54–55.


Trachtenberg noted the two men's communications quoted in Ickes' diary, arguing that FDR “intended to exercise control” over oil policy. See Trachtenberg, The Craft of International History, p. 98.


Dallek, Franklin Delano Roosevelt and American Foreign Policy, 1932–1945, pp. 599 n. 8, 604 n. 51, inter alia.


See Beard, President Roosevelt and the Coming of the War, 1941.


FDR-Ickes, June 23, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, June 25, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes could hardly have written this after reading FDR's criticism, and FDR would likely have addressed the oil embargo had he received Ickes' letter first. Folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL, does not contain FDR's reply, but it does contain Ickes' diary, which reprints the reply. See Ickes, Secret Diary, Vol. 3, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL, pp. 559, 548–568.


Ickes-FDR, June 30, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


FDR-Ickes, July 1, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, February 13, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, June 2, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, July 21, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, August 5, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, September 20, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, November 24, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, May 12, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL. See also Ickes-FDR, July 29, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, July 3, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, November 24, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, December 1, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


FDR-Ickes, May 24, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL; and FDR-Ickes, December 3, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, May 24, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


Ickes-FDR, June 23, 1941, folder “Interior-Ickes, Harold L., 1941,” box 55, PSF FDRL.


See, for example, Arlette Farge, The Allure of the Archives, trans. Thomas Scott-Railton (New Haven, Conn.: Yale University Press, 2013).