Chapter One

An Extended Methodological Introduction

1. A Starting Point

Victor Furnish gave voice in an excellent summary article in 1989 to a central challenge within Pauline interpretation: the extraordinary diversity of Paul’s writings, especially at first glance. Even allowing for the assignment of some or even most of the canonical letters written in Paul’s name to other authors — a move that will be scrutinized carefully in due course — the remaining letters are still characterized by a remarkable variation in argumentation, structure, and expression. Just Romans and 1 Corinthians, whose authenticity is usually uncontested, when placed side by side, seem to come not infrequently from overtly different places in conceptual terms. Meanwhile, adding only 2 Corinthians and Galatians to the comparison diversifies the overall situation further, creating a fundamental methodological challenge. How are interpreters to supply a unified account of various aspects of Paul himself as his texts strain in multiple directions?

An important response to this phenomenon was articulated by J. C. Beker (1984 [1980]) in a classic study that is now less read than it ought to be. It is Beker’s initial methodological proposal that concerns us here. He observed that any reconstruction of Paul’s thinking must navigate an interplay within his texts between “contingency” and “coherence.” This proposal seemed to capture much that Pauline scholars had been trying to do, justly passing into standard scholarly parlance, at least for his generation.

“Contingency” denotes the occasionality of Paul’s texts. The interpretative posture being identified here is, more expansively, the conviction that Paul’s letters were crafted and dispatched to deal with quite specific circumstances in the communities to which they were addressed, and hence were shaped significantly by Paul’s perceptions of those particular circumstances. All interpretation of Paul, at any level, must therefore take this dimension into account programmatically. Every statement is not merely a word of and from Paul but also, as Beker famously said, “a word on target” (1984 [1980], 12). (This axiom will be nuanced in due course.)

Beker’s hypothesis speaks immediately to the phenomenon of diversity with which we began. It could be that Paul crafted his letters to speak to specific communities so successfully that the very different communities to which he wrote thereby elicited very diverse communications. If Paul, something like a chameleon, changed his structures, arguments, and terms to reflect the very different situations of his auditors, then any collection of his letters would contain an initially bewildering array of colors and patterns. This is not ultimately surprising if we are explaining a rhetorical chameleon responding to diverse environments — or, in perhaps more positive terms, a good missionary contextualizing his message.

Having said this, however, Beker also insisted that Paul’s letters were nevertheless informed significantly by what he called Paul’s “coherence” — a term that can confuse as much as it clarifies, so it needs to be defined carefully. Coherence denotes for Beker certain important convictions on Paul’s part about God’s activity in Christ and its implications. Beker claimed that a constant coherence in basically this sense is discernible within all of Paul’s different letters — a “deep structure” underlying all the occasional argumentation.1 (He argued more specifically that it was “apocalyptic,” suitably defined, but we do not need to explore this particular proposal just yet.)

But Beker did not just name two important dynamics within Pauline explanation. He named them in relation to one another. By identifying the importance of both contingency and coherence, Beker insisted that Paul’s conceptuality could not simply be read out of Paul’s texts automatically or in any straightforward way; his texts are not simply coherent. Their basic texture is irreducibly contingent. Paul’s coherence can be recovered only through the patient interpretation of their occasional expression. Yet any occasional accounts of these texts must nevertheless be informed, he insisted in equal measure, by some account of Paul’s coherence at work within them. Paul’s contingent texts are infused in some way with an overarching coherence. A single chameleon, we might say, underlies the shifting colors and patterns.

Few if any scholars today accept the exact terms in which Beker set up his dialectical interpretative hypothesis; however, all do accept its basic claims and tend to proceed accordingly: with accounts within any explanation of a Pauline text of its contingency, its coherence, and the interplay between these two dynamics. Indeed, historical work on Paul, whether in toto or in part, is effectively impossible in any other terms. Given the brute fact of textual diversity, an utterly coherent account — the view that Paul is simply providing propositions exclusively drawn from his own theological system — is untenable. Clearly, Paul is overtly circumstantial in much that he writes. But the equally obvious facts of textual overlap coupled with, at the least, occasional moments of sustained argument within letters, not to mention terms and claims held in common across different texts, entail that some degree of coherence must be in play as well. Paul must hold certain propositions constant within and between sentences for much of the time, or his words would be simply random — meaningless — which they patently are not. All scholars in effect concede, then, that contingency and coherence are in play in some sense within all Pauline interpretation, and hence the question of their relationship is posed automatically. Beker’s formulation captures the complex ways in which Paul’s letters both overlap with and diverge from one another in terms that are essentially undeniable. And given its importance, we ought now to probe it a little more deeply.

2. A Conundrum

When we press on Beker’s account, we quickly discover that to name the problems of contingency, coherence, and their relationship within Pauline explanation more broadly is by no means to have resolved them. Beker himself tends to jump straightaway from this prescient articulation to a particular solution rather as a magician pulls a rabbit out of a hat (see, e.g., 1984 [1980], 16-19). He does try to justify his thesis in due course. But his frequent restatements and difficulties (see esp. xiv-­xxi)2 indicate that further work on the account of the problem needs to be done before any proposed solution can be successful.

The difficulty begins to emerge when we realize that there are significant variations in the ways key interpreters have understood and correlated the basic dynamics of contingency and coherence. We can consider five important and rather different accounts of their relationship here first.

(i) A more traditional account of Paul tends to read him as a thinker who takes pains to articulate at any given moment only what accords with a carefully constructed prior theological orchestration. The circumstances surrounding each letter therefore catalyze the deployment of some of these resources but do not shape or elicit them in any stronger sense. It is possible simply to extract from the letters the data and arrange it topically without further ado, yielding a description of Paul’s conceptuality that holds together overtly and intelligibly as a theological account. As a result, Paul’s letters are viewed as largely coherent in propositional and even logical terms, merely mediated by their contingency or circumstances.

This is of course a maximal account of Paul’s coherence, although its nature is also being envisaged in certain terms. And most modern interpretation of Paul has been deeply unconvinced by it. But it represents one basic approach to the correlation of contingency and coherence in Paul, arguably lingering in the topical approach of many of the major current accounts of Paul’s theology.3 We can turn now to consider its polar opposite.

(ii) Some interpreters suggest a maximal account of Paul’s contingency that minimizes the role of any coherence — so, for example, the reading offered by Heikki Räisänen.4 Räisänen argues that Paul is a demonstrably ad hoc thinker, manufacturing statements and positions for temporary rhetorical and local advantage. We might say, then, that in each letter he basically tells his auditors whatever their itching ears want to hear, as long as this maintains his own objectives within their situation (and these can be viewed rather diversely, whether in terms of power, money, loyalty, and so on). It is therefore fruitless to search for any overarching coherence, because Paul is a conceptual opportunist. A slightly less cynical variation on this basic approach that also appears at times in Räisänen’s analyses suggests, rather, that he is deeply confused. Paul would deploy coherence if he could, but he cannot hold together the issues that he is debating in a coherent fashion, arguably because his different positions cannot be held together. He tends to paint himself into corners. Different letters — and even some letters within themselves, such as Romans — articulate incommensurate conceptualities, although the overall result is the same: an essentially contingent corpus informed by no basic coherence at all.

Most scholars feel, however, that Paul’s coherence cannot be dissolved completely into momentary rhetorical advantage or confusion, even if at times it can be. So they posit accounts of contingency and coherence that lie somewhere between these two options.

(iii) Beker clearly gestures toward an alternative account of contingency that takes Paul’s coherence more seriously than does Räisänen, without losing sight of the influence of circumstances on the texture of Paul’s texts, as a more traditional account tends to. We have already noted its main suggestions and so will only summarize them here. According to Beker, Paul contextualizes his material within the forms and situations of his auditors so that his letters take on a very different texture for each specific situation. This sensitive hermeneutical behavior on his part does not entail, however, that his coherence has been lost. Rather, it has merely been translated into local idioms, in response to specific local issues — a dynamic account of the presentation of the gospel that Beker rather admires. So a reversal of this process by the Pauline interpreter should be able to recover the coherence that is at work within the letters. We do not reduce Paul’s letters directly to coherence, as in option one, but neither do we abandon coherence altogether, as in option two.

But is this all we need to say? Other scholars have suggested that other key dynamics within this process need to be recognized.

(iv) E. P. Sanders can represent a fourth principal option (at least, in his famous early publications 1977 and 1983).5 Sanders articulates an account of the relationship between contingency and coherence in Paul that lies in effect between Räisänen’s and Beker’s. Like Beker, he is persuaded that coherence can be recovered from Paul’s texts in relation to certain pressing issues that his communities forced him to address. However, like Räisänen, he is not convinced that this coherence is always commensurate. According to Sanders, different issues raised by different situations in different communities can lead Paul to draw on different clusters of coherence as he responds. Within each cluster and in response to certain questions these communications are coherent, but the clusters are not especially coherent when placed side by side. (Although Sanders claims that his account of Paul is coherent in another sense from Beker, namely, that it explains coherently why Paul is often conceptually incoherent: Paul responds consistently to different questions with different systems. But this is of course a very different notion of coherence from Beker’s.)6 In short, Sanders sees Paul’s coherence as possessing areas of what we might call unresolved material. Not everything has been strictly correlated in systematic terms. But there are bunches or clusters of coherent material activated discernibly by different issues.7

Up to this point, our main options have basically assumed that the phenomenon (or not) of conceptual coherence in Paul can be explained in essentially synchronic terms that nevertheless hold firm diachronically: he makes perfect sense consistently at all times; he makes no consistent sense at any time; he translates consistently in every situation; or he draws intelligibly on particular clusters of coherence in relation to particular questions as they arise. But we need now to consider a rather different explanatory angle on our central problem that abandons this assumption.

(v) Some scholars have argued that Paul changes his mind about certain things, which introduces a further explanatory dynamic.8 As certain contingencies arose, such interpreters suggest, Paul was forced to think through new questions that had not yet occurred to him. And the result was, they suggest — perhaps not surprisingly — that he modified his coherence. That is, he did not merely apply a coherent position that was already in place just waiting to be deployed. Nor did he make up something altogether new. His existing coherence was altered by this engagement. There are minimal and maximal suggestions in this regard. Minimalists suggest that Paul introduced minor alterations, and perhaps only to positions that later tradition would view as secondary matters. Perhaps he adjusted his views on local church government. Perhaps he changed his mind on whether church leaders should be paid. More worryingly, however, he might have changed his mind about the future fate of deceased Christians, at which point we are edging toward a more maximal account of his development. Indeed, some go so far as to suggest that each new circumstance bequeathed to Paul a significant new set of terms and arguments that then resonates traceably through subsequent letters, fading, but still detectable, as it becomes less pertinent.

This view clearly introduces a further significant variable into the possible correlation between contingency and coherence in Paul’s texts. Minimal developmental advocates see Paul’s coherence modifying in a nuanced fashion in relation to each contingency; maximalists detect a rather small initial reservoir of coherence that is augmented significantly by each new situation. Hence, not only does Paul’s coherence shift, but it grows, although in each case just how much and in what sense is fiercely debated. It is enough for now, however, simply to note this basic suggestion.

Although it is tempting to place these options on a continuum, and not altogether improper at times to do so, their accounts are not mutually exclusive. They propose (in Beker’s terms) different, at times overlapping, hermeneutics concerning the relationship between contingency and conceptuality. But more worryingly, it is apparent on reflection that these could easily interchange from text to text and situation to situation. At one moment, Paul could draw on innate tradition and prior conceptuality directly and simply and then apply it (option one). At another moment, he could feel a strong rhetorical need to create a certain dynamic. But lacking the traditional resources, he might make something up or deploy a rather random appeal (option two). Alternatively, on a given issue, he might draw on his most important convictions but translate them carefully into local idiom (option three). Then again, he might switch between different and somewhat unresolved traditions and conceptualities as still another issue arises (option four). However, as certain situations arise, it might be that he is prompted to engage instead in some hard thinking, modifying his principal coherence and proceeding to think and to argue accordingly (option five). In short, all these options could plausibly be in play at some point within Paul’s letters and hence need to be drawn upon for accurate explanation by the Pauline interpreter. So articulating clearly the relationship between contingency and coherence in Paul is a deeply challenging task. And it is about to get even more so.

Although Pauline scholars work fairly frequently with the foregoing approaches, they tend to direct far less attention to the nature of Paul’s coherence, which can be conceived in oddly stereotypical and anachronistic terms. We need to appreciate that different options exist here as well, a situation that can be illuminated further if we initiate a brief conversation with George Lindbeck (1984).

Lindbeck famously suggested that theological conceptuality can take different forms. (He is of course concerned with modern theological movements, but his insights are still instructive, suitably modified, for the interpreter of Paul.) He identifies three basic modes. The first is strictly propositional — “cognitivist” — this being the bête noire that many interpreters resist attributing to Paul. The second is more experiential — “experiential-­expressive,” as he puts it. The third, overlooked in Lindbeck’s view for far too long, is traditioned — his designation is “cultural-­linguistic.”

Lindbeck’s views are subtle, and have been paralleled and developed in many ways, but for our purposes here it suffices to appreciate his argument that conceptuality per se does not exist apart from language, and that language does not exist apart from practices, habits, and communities, which themselves possess a history. These claims suggest that Paul’s coherence should not necessarily be conceived of fundamentally as a distinguishable schema arranged propositionally within his own mental universe, which is inserted in some way into his texts as he engages with local circumstances. Indeed, this looks suspiciously like a prejudicial construct animated by the post-­Cartesian predilections of much modern Western philosophy. Nor should it necessarily be understood as a translation of an underlying primordial religious experience of God, which risks reifying anachronistically some of the central claims of modern liberal Protestantism and even of European romanticism. Paul’s conceptuality might have a fundamentally different nature from these explanatory suppositions. It might be embedded in a broader community and within various traditions as a set of habits and practices (understanding these in a certain way) that are adjusting to changing contexts. And language would consequently articulate this conceptuality in a different mode from what we might expect — principally as reformulated narrative and tradition (and here we link hands with Lakoff).

We do not need to pronounce here on the validity of Lindbeck’s well-­known assertions. They suffice to introduce further possibilities into our unfolding consideration of Beker’s approach to Paul. Lindbeck challenges the Pauline interpreter to be open to the problem of divergence — not to mention anachronism — in the way that Paul’s coherence is conceptualized. He suggests that the analysis of Paul’s thinking might often contain an unexamined endorsement of individualism and interiorization — a demonstrably culture-­bound and fallacious account. Further, he opens up the particular possibility that Paul’s conceptuality is very unlike our own, perhaps being more broadly Aristotelian in mode, or at least distinctively ancient rather than modern and so frequently communal, narrative, and traditioned.9 And the task of the Pauline interpreter in relation to Beker’s program has now become even more complex than it was before. Not only must she coordinate contingency and coherence, but she must remain open to the different possibilities concerning the shape of Paul’s coherence within that coordination, not to mention Paul’s location vis-­à-­vis his broader community.

We might be tempted to throw up our hands at this point and say, “Who cares?” — that is, to abandon the quest for the historical Paul’s thinking. But many people care deeply about this situation. For those who attend to Paul’s writings in historical mode, if only to some degree, the situation is deeply disturbing.10 His coherence is — to borrow a Matthean image — a pearl of great price (13:45-46), and we need to find some way of buying it if we can. But in order to do so, we need to appreciate that the challenge is no longer merely complex. This situation and its central methodological conundrum as Beker has formulated it is almost entirely unresolvable. The differences between these fundamentally different options cannot be adjudicated except in a certain, carefully defined way. Conversely, the way that these options are adjudicated at present seems almost invariably to be circular and incoherent, trapping much modern Pauline interpretation in a morass of corrupt conclusions.

3. The Solution in Outline: A Framing Account

In order to assess the plausibility of the different options concerning Paul’s coherence, both in nature and in influence, the interpreter must be able to articulate, at least in rough terms, the circumstances through which any coherence is being expressed. Moreover, the circumstances must be known across at least some, and ideally all, of the letters. Only this allows the interpreter to reach valid judgments about the probity of the different views of Paul’s coherence being offered, although even then it clearly remains a difficult task.

Unless we know the basic circumstances surrounding the composition of a letter, we simply cannot tell whether Paul primarily is deploying his own conceptuality, is making things up in an ad hoc manner to influence those circumstances directly, is contextualizing within those circumstances, or is drawing on different conceptualities for the different questions those circumstances generate. All these widely differing accounts of the text would be entirely possible, because we have no way of filtering out the contribution from the circumstances, not knowing them. And without knowing the circumstances of his other letters as well, we cannot make any progress on these vital issues. Only a comparative analysis within which the contribution of circumstances to the composition of more than one letter can be controlled (at least to some degree) offers any hope of valid judgments on questions of contingency and coherence. (And it is this realization above all that rules out atomistic, piecemeal approaches to Pauline interpretation, working from each letter individually, as a sufficient interpretative act.)11

If a basic account of some or even all the relevant letters’ contingencies is in place, then the interpreter can control for the contribution of contingency across the different situations and reach a judgment about the nature and role of Paul’s coherence. As Paul operates coherently in relation to different circumstances — assuming that he does — the Pauline interpreter should be able to identify that coherence. And this identification can then perhaps be cross-­checked with the opposite situation, namely, his responses to similar contingencies, where his coherence should hold reasonably steady, if he possesses it. The presence of both comparative scenarios will allow the Pauline interpreter to adjudicate between the different options posited in explanation of the dialectic between contingency and coherence in Paul’s texts. And this all suggests that any valid Pauline interpretation in any historical respect must begin with a workable account of the letters’ circumstances in relation to one another. We must tell the story of their interrelated composition. This is the sine qua non of all valid historical interpretation of Paul. Furthermore, this implication needs to be appreciated in all its programmatic force.

Any interpretation of any Pauline text in historical-­critical terms that extends beyond textual superficialities and claims in terms of immediate exigence presupposes that the interpreter knows the story within which that text is embedded. If this story is wrong by way of methodological overreach or contamination, then the analysis is probably significantly flawed as well. How much current Pauline exegesis is mistaken because the historical story of the letters it presupposes is corrupt? The figure might be disturbingly high.

We should go on to note immediately that this story will have a certain distinctive shape. It will be the story of the circumstances in which Paul wrote his letters, so it will have a biographical quality. It will not be a complete biography in the sense of an intellectual biography; we are pursuing this preliminary biography so that we can write that thicker, richer account on its basis. Moreover, it will need to extend past Paul, to a degree, into his relationships with his communities, and into some of their significant relationships with one another in turn, although only the relevant contingent information at these points is required; not every biographical detail that can be gleaned from the texts informs the circumstances that surround the letters’ composition. In short, Pauline interpreters working in historical mode desperately need — to draw at this moment on Derrida — the Pauline story that frames the letters.12 (And there are other good reasons for this project to be minimal,13 although it is emphatically not an “atomistic” account.14) Once this is appreciated, we can turn to consider the nature of the evidence that must underlie its derivation — another vital discussion.

4. The Frame’s Arguments

4.1 Avoiding Vicious Circularity

Put at its simplest, we need to appreciate that certain arguments must be avoided when constructing this frame or it will be compromised irretrievably. And the key constraint here is actually quite simple. We must at all costs, during the construction, avoid introducing judgments concerning interpretative dimensions in Paul that we are going to reach later on, after the frame has been completed — and so are based on it. This seems to be a fairly obvious caution, but it is violated so frequently by Pauline interpreters that it needs to be grasped with complete clarity.

If, for example, we introduce convictions about Paul’s coherence into our initial frame, we will generate a corrupt explanation in several respects. First, we have no basis for this privileging of our conceptual convictions; it is merely arbitrary. Second, this arbitrary privileging contaminates our framing account, distorting it and rendering it effectively useless as a control. Third, if we go on to claim eventually that this frame underwrites our conceptual convictions, then we fail to recognize the vicious circularity both undergirding and invalidating our conclusions.

This concern is no straw man. For example, Udo Schnelle has relatively recently provided a benchmark account of Paul’s thought that both distinctively and impressively recognizes its integration with his biography. Yet Schnelle excludes from the outset six of the thirteen canonical letters bearing Paul’s name from his complex and subtle theological reconstruction, assigning them to a Pauline “school” (2005 [2003], 146-51). This exclusion takes place on the grounds that “in all the deutero-­Paulines, Paul’s emphasis on the doctrine of justification, as found in Galatians and Romans, remarkably subsides. Apocalyptic motifs in Christology likewise decline in importance, and an emphasis on present eschatology prevails. The focus is on matters of church order and problems of ethics that have emerged because of the altered situation in the life of the church (the appearance of false teachers; coming to terms with the failure of the parousia)” (150).

Schnelle gives a faithful account of the thinking of most Pauline analysts here. But what is remarkable is the overt circularity present in the a priori assertion of each of these claims if they are introduced at the most basic level of his analysis, which they are. For this case to hold, Pauline interpreters must know that Galatians and Romans contain a “doctrine of justification” and, moreover, that this is a constant for Paul, necessarily appearing in all his authentic works irrespective of contingency. And they must know the same things in turn concerning “apocalyptic motifs in Christology,” and the same things in reverse in relation to “present eschatology,” “church order,” and “ethics” (i.e., that the authentic Paul did not emphasize these and no circumstances ever unfolded that might have led him to address them). But they can know these things only after the historical Paul has first been reconstructed in toto and the circumstances surrounding the composition of his letters controlled so that this picture of his contingency has emerged. Hence, to build these assumptions into that reconstruction at the outset is to vitiate the entire process. Schnelle’s construction must now be suspect in fundamental terms. The very shape of his account of the Pauline data must be viewed as problematic.

We cannot appeal to Paul’s principal theological convictions when framing, because we do not yet know what they are.15 Similarly, we cannot appeal to his lexical choices, at least strongly, because we do not yet know whether he is a hermeneutical chameleon, adapting his lexicon to local circumstances, as against someone with a very rigid and limited lexicon. (Claims concerning the lexical encyclopedia available culturally to him in general terms will still be relevant.) Moreover, we do not yet know what the circumstances are to which he might be adapting. And so on.16 We can, in short, use only evidence and argumentation that do not beg important later questions.17 Where question-­begging arguments are offered, we must rebuff them.18

One might respond to these strictures with the concern, “What arguments are left?” So some clarifications concerning the nature of the evidence underlying the frame should now be helpful. Indeed, the rest of this book will unfold rather like a classic treasure hunt.

4.2 The Appropriate Roles of Criticism and Doubt

Just as fictional treasure hunters face an overt problem — usually, piecing together subtle clues into a map that leads the heroes to a buried store of ancient riches — so we face the overt problem of the Pauline letters’ frame. And just as the fictional investigators must sift a mass of evidence for the relevant pieces of information, that is, the clues, and decode their cryptic messages, we too must sift a data pool for the appropriate pieces of evidence and fit them together. But we need to give some thought to just what our clues might consist of.

We have already established that certain clues must not be introduced. We cannot introduce claims based on concealed judgments that have not yet been generated by our overarching investigation, injecting vicious circularity into the process. The treasure hunter should not simply decide in advance of her investigation that the treasure is hidden in a particular room or house, or even on a specific island. Moreover, we need to be attentive to the different levels of confidence that we can place in alternative construals of our ancient and fragmentary data. Frequently, this data admits of multiple explanations; we need to proceed with construals that are certain or at least highly probable — as against probabilistic and only corroborative, or merely possible and thus little more than broadly supportive. We will need to be alert to the infiltration of evidence from this last category up through the hierarchy to the first, that is, possible readings masquerading as probable or even certain evidence. And we need to be sensitive to the correct methodological functions of criticism and doubt, since it will be easy to cast doubt in general terms on the small clues that underpin much of the reconstruction of the Pauline frame, a consideration worth exploring in more detail.

Descartes suggested, in a classic argument widely influential in the modern period, that everything is in effect guilty until proved innocent. The result was, rather famously, the reduction of all certain knowledge to the conviction that his mental processes at least guaranteed his existence. In other words, he used radical doubt as a fundamental method. Everything must be doubted until it can be demonstrated indubitably to be true.

Now, it can certainly be granted that it is helpful at times to reorient paradigms and to strip away faulty thinking, and radical doubt seems to achieve these advances. But the Cartesian method has struggled to get anywhere significant and has, moreover, been subjected to ferocious critique, not least from Wittgenstein, who pointed out (characteristically indirectly) that the use of language implies participation in a broader linguistic community, which is in turn difficult to detach from a complex broader reality that cannot be doubted in the first instance without lapsing into utter incoherence. So Descartes’s key initial claims are in fact delusional. Unfortunately, however, the critical method, which played such a significant role in the rise of the modern university, has had a long dalliance with Cartesianism, so the latter tends to live on, haunting the corridors of the academy like a restless shade. It allowed figures like Kant to reject tradition out of hand and to argue from simpler and more certain first principles, although Kant too struggled to develop his principles with the certainty and extension that he really sought. It is not a completely crass oversimplification to suggest, then, that many modern Pauline scholars, shaped in part by the traditions at work in the modern university, seem to assume, at least at times, that the “critical” assessment of evidence simply involves the application of doubt in a generic way, ultimately in the manner of Descartes. It is a posture of comprehensive skepticism. One must be unconvinced until one is convinced of something’s probity on certain grounds. But I would suggest that when practiced in this generic and universal manner, this is an invalid and self-­defeating methodology and a false understanding of criticism.19

We have already gestured toward the method’s invalidity by way of a failure to account for language, but it is worth adding here that this method will lead us in Pauline studies to precisely nothing. Everything the Pauline scholar claims can be doubted on Cartesian grounds. One of the method’s great difficulties in general is reaching in any significant way into what we might call immediate external reality (and certain scholars have gone on to struggle mightily with this), which falls far short of any reconstruction of the past. There have not been many rigorously Cartesian historians. And it follows that any partial application of the method is inconsistent as well. That is, any scholar laying claim at any point in historicist terms to something originally true in relation to Pauline reconstruction cannot turn at another point and make a Cartesian claim, because that would be fundamentally incoherent. But if the programmatic practice of doubt is self-­defeating, then we ought to pry the notion of criticism loose from its dead hand.

Criticism, correctly conceived, is the application of the appropriate feedback from any data under consideration to any hypothesis currently being suggested to account for it. Hypotheses are not ultimately endorsed without the appropriate data (although we should add that the sort of feedback that critics expect will depend, most importantly, on the nature of the object under investigation, i.e., on the actual nature of the data). Criticism assesses this central explanatory relationship. And doubt can be an important part of the process. Appropriate doubt is directed at a hypothesis when it fails to account for the relevant data, at which point it should be modified or abandoned, and this irrespective of its relationship to tradition. Interpreters will have difficulty deriving and initially processing hypotheses without tradition, so tradition remains essential, but it is not unquestioned. The critical process is thus broader than the mere presence of doubt and is by no means reducible to it. Indeed, it begins in exactly the opposite way, somewhat fideistically, with the positing of an appropriate hypothesis in explanation of a given problem in a data set, sensitive to the nature of that data and the methods appropriate to evaluation, with doubt potentially entering the process at a later stage. Appropriate, essentially scientific criticism, we might say then, never takes place in a vacuum, and neither does appropriate doubt; both are always considering the relationship between evidence and its interpretation.20 The following reconstruction therefore welcomes both criticism and doubt at the appropriate points but is uninterested in criticism and doubt per se.21

We will rely on slender snippets of evidence in what follows, because that is all that we have — occasional and fragmentary remains of conversations that took place millennia ago. But we do have evidence, and it will not do to dismiss parts of the following reconstruction with a generic claim that “this is insufficient” or “there is still not enough evidence.” If this is the evidence that we have and it explains the data in the best existing fashion, then the correct scientific conclusion must be to endorse it and not to complain that we need more data that unfortunately does not exist.

With these brief methodological clarifications in place, we can turn to the important practical question that remains: Can this reconstruction actually be achieved? Does the Pauline treasure hunter, in effect, have sufficient clues to find the cache of lost riches? The short answer is yes, and the reason for my confidence lies in the fact that the way forward has already been indicated in all essentials by John Knox, although much yet remains to be done. It is as if a basic map has already been drawn indicating where the treasure lies buried, but the difficult journey there has yet to take place, and one or two unforeseen challenges need to be navigated on the way.

5. Completing Knox

I suggest that John Knox articulated some time ago all the main moves that need to be made in relation to the data to generate a reliable account of the Pauline frame.22 Just why it has taken the world of Pauline scholarship so long to recognize this is an interesting question, but it should not detain us here. Certainly, he has not lacked support altogether.23 It must suffice to say that I will try in what follows to update, occasionally correct, and thereby complete Knox’s project, and this should result in a persuasive account of the Pauline frame.

Knox made two distinguishable and especially significant proposals in this regard. First, he drew a vital distinction between different sources in the data. This has been much misrepresented, however, so we will have to re-­present that suggestion here, and to articulate it further, since it lacks precision with respect to our current concerns. Second, he sketched out an argument that can, through a series of specific steps, derive a frame plausibly from that data without begging any key questions. After much further research and reflection, I would suggest that this argument cannot be improved upon in basic terms. Once again, however, it requires considerable further articulation, along with, at times, slight correction.

We should now pause to consider the two key dimensions in Knox’s work in more detail, since they underpin so much of what follows. We will take up his initial clarifications with respect to the sources in the remainder of this chapter and then move on in the rest of the book to work out his second contribution in detail. That is, we will spend most of our time reconstructing the Pauline frame by way of a series of carefully argued steps through the data as Knox basically suggested them. But before proceeding to that task, we must appreciate clearly his claims about the sources.

5.1 Handling the Sources Appropriately

Knox introduced a strong distinction concerning the sources within the broader pool of available data (see 1987 [1950], esp. 17-28, rhetorically anticipated by 3-16) — a distinction that has been much misunderstood. Further reflection suggests, however, that it is both simple and incontestable.

Knox observed that Paul’s authentic letters are “primary,” meaning that they come from his own hand and hence directly from the circumstances that enfolded them. Their data is close to the events in question; it emerged directly out of the events, at times functioning causally within them. Over against this, the data concerning Paul in the book of Acts, the second principal historical reservoir for his life, is something of an unknown quantity. We do not know who wrote Acts, when, where, or — perhaps most importantly — why. We do not know — and certainly not at first glance — what the relationship was between the author and Paul. So this data is of “secondary” status over against the primary epistolary data. It is an unknown quantity that might come from much later on and at some remove from Paul himself. It follows that we ought to begin our reconstruction of the Pauline frame with the epistolary data and with this alone. Acts’ secondary data cannot be allowed to intrude into the processing of the epistolary material or it will in effect contaminate it.

Once the epistolary data has been processed successfully and some sort of explanatory frame derived, then the data from Acts can be introduced into an exquisitely controlled situation. We will be able finally to derive just the judgment that many have been seeking for so long — a sense of the accuracy of the book’s account of Paul, which can be derived by comparing its data with the Pauline material. Any bias in the Acts account will be overtly detectable and so could be corrected by the modern Pauline interpreter. Indeed, having noted the overt advantage of this situation, it is frustrating to observe how frequently Pauline interpreters flout this principle and muddle epistolary and Acts data together from the outset. In my experience, a great deal of the confusion generally present in Pauline biography derives from the failure to prescind from this practice. Once Knox’s key distinction has been abandoned, there is no way back — no coherent way to correct a situation that has descended into methodological chaos. When the advantages of preserving this distinction and following this rule are so great, and the disadvantages of violating it so debilitating, it is puzzling why more scholars do not seem to hold to it. But the reason may lie in some pervasive misunderstandings that we should now consider.

(i) Neither Knox nor I would claim that Paul’s data is perfect or completely unbiased. Paul is clearly imparting his own rhetorical bias to many of his biographical claims. But there is Pauline data pertinent to framing concerns that might have been somewhat inadvertent to Paul himself (where his biography is not an overt issue but biographical details arise in passing), and this can be mined for information with less risk of deceptive bias. Admittedly, Paul is at times deeply invested in narrating his activities appropriately — that is, where competing and frequently unflattering accounts of his behavior are being offered by others. However, we must appreciate that Knox’s claim concerning a chasm in initial quality between the data in the letters and that found in Acts is an essentially comparative judgment. The Acts data is initially opaque, irrespective of what we make of Paul. It could be spun out of thin air, for all we know. So the distinction between the two data sets remains valid in spite of Paul’s obvious imperfections.

(ii) Unfortunately, many advocates of the importance of Knox’s work in Pauline studies have launched vigorous and essentially independent attacks on the veracity of Acts.24 This tactic might have derived, at least in part, from the conviction that Pauline interpreters need to have their faith in Acts’ story of Paul shaken in order to appreciate Knox’s approach (and we touch on this to some extent in our fourth point here). But I suspect that this has done more harm than good. In the first instance, it is unnecessary. No prior judgment needs to be reached about Acts in order to appreciate Knox’s distinction concerning the respective initial value of the two sets of sources in historical terms. It stands on its own merits. Second, this strategy actually obscures the control that Knox’s approach provides for the reception of Acts. Acts should be evaluated later in the light of the epistolary data, because this is the clearest and most objective point from which to judge that book’s veracity. Initial attacks consequently obscure and even short-­circuit this important practice. And third, this attack creates a rhetorically plausible if ultimately invalid way of undermining Knox’s entire approach.

If Knox’s position is held to be linked tightly to the independent discrediting of Acts, then any demonstration of Acts’ veracity can be held conversely to overthrow Knox’s approach in its entirety. This is of course a bad argument, because it overlooks what he actually suggests and does not refute his central claim; it largely misses the point. But it is rhetorically effective if Knox’s advocates do in fact undertake sweeping indictments of Acts that can be countered.

However, we have just seen that such indictments are unnecessary and even distorting. From this point on, then, I suggest that the application of Knox’s proposals to Pauline framing be detached from independent critiques of Acts. We must let Knox’s own method show us how to evaluate Acts later on, for better or worse. An interesting possibility now presents itself.

(iii) Knox affirmed the importance of original source material over later and less-­determined data. So in effect, he privileged Paul’s eyewitness information as the correct starting point for the Pauline interpreter. It follows that if Acts contains eyewitness information, then that information must be elevated to evidence of the first rank. And at first glance, Acts does lay claim to eyewitness data (see Lk 1:2-3; Acts 16:10-17; 20:5–28:31). We will have to evaluate this carefully in due course, open to the possibility that data from Acts will join the epistolary evidence in the most basic layer of material undergirding the construction of the Pauline frame. We cannot begin with this evidence (because it is not overtly firsthand), but we can certainly affirm it if that judgment ultimately proves appropriate, and our investigation in accord with Knox’s strictures will provide the most controlled way of reaching such a judgment. Moreover, simply in general terms, it may be possible to affirm the veracity of a great deal of the data in Acts in the light of Knox’s methodology. The result of a strict application of Knox’s approach could be a stunning validation of Acts.25 But we will have to wait and see whether this proves to be the case.

(iv) Even as we must prescind from preliminary critiques of Acts and remain open to its function at times as a primary source, so too must we be alert to its inappropriate use and reject this. The field of Pauline studies is not aided by the frequent presence of what is effectively an Acts fundamentalism. This generally holds as accurate the individual episodes within Acts and the sequence in which they are placed by the narrative, along with the temporal intervals that the narrative periodically supplies. In short, everything that Acts says is true, and even in the order in which Acts says it.26 A dogged defense of the Acts frame for Paul, both episodically and sequentially, is by no means limited to scholars who explicitly endorse a historicist account of inspiration in terms of inerrancy, as we might at first expect. It is widespread outside these traditions, so we need to ask what else might be going on. I suggest that something as simple and as powerful as unexamined psychological and explanatory dependence may be in play. We must learn to recognize this, and to resist it.

Most Pauline interpreters have not been raised to think about Paul as John Knox recommends. The Paul of the church and of popular discourse is either the Paul of Acts or one inextricably and uncritically infused with Acts data. This is simply in place, like an ancient but familiar operating system installed years ago in the minds of interpreters so that they can run other programs and undertake other tasks. This story therefore almost certainly frames most interpreters’ more detailed and expansive work on Paul, and there is a painful explanatory price to be paid if it is to be rigorously reexamined — an onerous wiping clean of the relevant hardware and installation of new and unfamiliar software. Hence, it is entirely understandable that alterations in fundamental explanatory structures would be resisted, and they certainly are in the matter of Pauline biography.27

Precisely because of this necessary struggle against the psychological availability of framing explanations rooted in Acts data, we need to make a completely fresh start.28 We must attempt a comprehensive analysis of Paul’s life informed only by epistolary data.

With these caveats in place concerning Knox’s first crucial contribution to this debate — the distinction between primary, initially epistolary data and secondary data drawn largely from Acts — we should consider more closely some of the specific implications of addressing initially just the epistolary data, a consideration that Knox himself largely overlooked. That is, we must ask exactly how the epistles ought to be approached once we have realized that we need to begin with their data exclusively.

As we saw earlier, we are not yet interested in a comprehensive biography of the apostle. This would attempt too much. We need only to frame his letters with a minimal reconstruction, on the basis of which we can attempt richer descriptive accounts of his life and thought later on. In order to move forward, then, and to generate an initial frame for Paul’s letters from the data present in letters, we will need to answer five specific questions and little more: questions about the letters’ authenticity, integrity, sequence, intervals, and reference. That is, in order to derive a frame for Paul’s letters from Paul’s letters, we must reach judgments about which letters he actually wrote (authenticity) and whether different original letters have been editorially pieced together into their current canonical form (integrity). And at the same time, we need to order them relative to one another (sequence) and, if possible, ascertain the length of the gaps between them in absolute temporal terms (intervals). Moreover, we need as far as possible to anchor any developing sequence to other events taking place at the time (reference), thereby ultimately generating dates and an absolute chronology overall. If we drop one of these questions from our explanatory quest, then we lose the ability to construct a frame (although some might argue that we do not, strictly speaking, need to solve the question of reference; we might be able to frame the letters independently of external events — an issue we will discuss further shortly). If we answer them all, however, then we do generate a useful framing account. Consequently, these five questions will be our particular focus through the analysis of Paul’s epistolary data that follows. Their mutual necessity is the principal reason why the following is not just a “Pauline chronology,” an exercise that generally does not attend simultaneously to questions of authenticity. A Pauline frame is both less and more than a Pauline chronology. It does not address all the questions of chronology raised by Paul’s life but only those necessary to frame the letters. At the same time, in framing the letters it has to address more than merely chronological questions, and in a particular methodological fashion not always discernible in chronological treatments of Paul. And at this point we must introduce a significant corrective to Knox (thereby drawing a further distinction between reconstructions of the Pauline frame and most chronologies).

In his most famous work on Paul, Knox basically sketched out the key contours of Paul’s life and thought; he was concerned with the big picture. So he did not reflect in depth on some of the methodological issues that we have just considered. As a result of this oversight, he went on to make a mistake when dealing with what I have identified here as the issue of authenticity, which he addressed only vestigially in any case, although his thinking here ultimately developed in an interesting way.29

If we exclude question-­begging argumentation from our initial analysis of the epistolary data and maintain a similar vigilance against illegitimate Cartesian doubt, then when considering the question of authenticity we cannot at the outset simply exclude as an obvious matter any letters bearing Paul’s name. We must make a case for exclusion with respect to each putative Pauline letter; epistolary data is in effect innocent until proved guilty. Consequently, all thirteen letters in the canon bearing Paul’s name must be carefully evaluated for their authenticity during framing.30

With this important if surprising correction in place concerning the issue of epistolary authenticity, our consideration of Knox’s famous source distinction is complete. We can now turn to the second set of key insights in Knox’s work that assists us when framing Paul: his actual case for a frame. In my view, Knox is again basically right, although he made some important mistakes as he developed his case, which will necessitate some revisions as we move forward.

5.2 The Reconstruction’s Key Steps in Knox

Knox did not just identify the key methodological distinction within the sources at the outset of rigorous Pauline reconstruction. He also gestured toward the way in which, in a series of steps, the all-­important epistolary data could be seen to yield a coherent and largely complete frame of Paul’s apostolic career, and hence of the period during which he composed his letters. Knox argued his case in five steps (1987 [1950], 32-42).

First, he established a sequence for Paul’s three longest letters — Romans and 1 and 2 Corinthians — by noting the way in which they all speak of an important collection of money destined for Jerusalem. This activity orders them as follows, covering a period of roughly one calendar year: 1 Corinthians — 2 Corinthians — Romans.

Second, he argued for the insertion of Galatians into the middle of this sequence, largely by suggesting (partly in dependence on others) that Galatians 2:10 is speaking of the collection as well. This critical move linked Galatians’ uniquely extensive and detailed prehistory, especially concerning Jerusalem, to the sequence already in place. And Knox now knew that Paul undertook a visit to Jerusalem just before the main letter sequence, his second visit; and another some fourteen years before that, his first. Given that Romans anticipates a further visit to Jerusalem to deliver the collection, Knox now hypothesized that three visits to Jerusalem occurred during Paul’s apostolic career. Moreover, Paul’s call must have occurred some three years before the first visit. Hence, after these simple moves, a fairly extensive sequence and frame for Paul’s apostolic work is in place.

Third, although Knox did not explore the point in any detail, he did note in passing that a reference might be detectable within this data in relation to Paul’s escape from an ethnarch appointed by King Aretas to govern the city of Damascus (1987 [1950], 67), which the apostle speaks of somewhat oddly in 2 Corinthians 11:32-33. Almost certainly, this immediately predated his first visit to Jerusalem as an apostle (Gal 1:18-20). If this event could be referenced, then the sequence already developed in relative terms could be anchored to broader events. And Knox basically got the dating of this event right, that is, around 37 ce, although for rather ironic reasons.31 Paul’s call had to occur, then, around 34 ce, his first visit to Jerusalem around 37, his second around 51, the letters 1 Co­rinthians, 2 Corinthians, and Galatians had to have been composed in 51-53, and Romans, anticipating a third visit to Jerusalem, in 54. The entire frame was now anchored in relation to external events through a key reference, although for Knox this was drawn from Acts (see his 1987 [1950], 67-68).

Fourth, Knox introduced the letters still unaccounted for into this framework. As we have already intimated, he fell victim at this point to vicious circularity, not to mention uncritical doubt, so he did not consider correctly all the extant epistolary evidence bearing Paul’s name. Nevertheless, those letters he deemed authentic were inserted where he thought most appropriate into the existing framework, which is the right procedure.

Fifth, and finally, Knox introduced the evidence of Acts into his reconstruction, although under the control of the material he had already derived from epistolary data. Where he detected a clash, he modified or dropped the information from Acts. Consequently, although the content of the various visits by Paul to Jerusalem recounted by Acts matched up rather nicely with the visits apparent in the letters, Knox reduced the five visits found in Acts to the three attested by the letters, and again, I think he fundamentally got the correlations right (1987 [1950], 35, 44). Indeed, I think that Knox got most things right in this series of steps and the resulting frame, so we will essentially trace it out again in what follows. However, it will need to be adjusted at several key points, which can be usefully anticipated here.

5.3 Beyond Knox

The elaboration and correction of Knox’s case will take place in six analytic steps spread through five chapters. (Steps two and three are best treated in a single discussion.)32

(i) The backbone of Knox’s reconstruction is the stringing together of 1 Corinthians, 2 Corinthians, and Romans in relation to the collection of money for the community in Jerusalem. But this step has become rather more complex recently than it was in Knox’s day — although it was complex enough then. This collection has been narrated in many different ways by modern scholars. And if 2 Corinthians is partitioned in certain ways, as many scholars advocate, then Knox’s broader argument is problematized if not invalidated. There are more letters in play than just canonical 2 Corinthians; these begin to spread out temporally, and some fail to reference the collection. The partitioning of 2 Corinthians will in any case raise the whole question of the “interim events” that unfolded between the composition of 1 Corinthians and certain pieces (at least) of what is now 2 Corinthians. Complex reconstructions of these events will tend to undermine Knox’s claims concerning a straightforward sequence.33 And we encounter the issue of partition in relation to Romans as well, since scholars have long nursed various suspicions about chapter 16. So we must revisit Knox’s articulation of a backbone for the broader framing project and see whether it is still viable in the light of this complex research on Romans and 2 Corinthians.

I will, however — to tip my hand — end up arguing for a great simplification with respect to much modern explanation of this material, suggesting that the simplest account of the collection is the most plausible, and that complex reconstructions of the interim events rest too much on possible — and not enough (if ever) on certain or even probable — readings of the data, while similar considerations apply to the partition of both Romans and 2 Corinthians. So we will end up affirming Knox’s basic claim at this point, if not simplifying it still further.

But we will introduce a relatively new interpretative dimension that is often overlooked when considering the location of letters — what Charlotte Hartwig and Gerd Theissen (2004) have dubbed Nebenadressat, or “addressees alongside.” Paul’s letters were written for and spoken to a primary audience, which is named explicitly, but we learn from the letters in this initial sequence that their composition and performance were not limited to that principal audience. Other, secondary audiences in effect overheard these letters, thereby constituting audiences alongside the original addressees, as Paul was well aware. A new attentiveness to this dimension in Paul’s letters provides important corroborative evidence concerning their composition. So here too we will go beyond Knox in a supplementary fashion.

(ii) The introduction of Galatians into the epistolary backbone oriented by the collection of money for Jerusalem is perhaps the most important step in the entire argument. It locks Galatians, which supplies so much information about critical events in Paul’s earlier ministry, to the large letters already positioned next to one another. I do not disagree with Knox that this is what ought to happen or that his judgment here is right. However, I am unsatisfied with the arguments he adduces in its support — essentially, the solitary and rather fragile claim that Galatians 2:10 refers indubitably to the collection. So I will try to strengthen this critical move, in two principal ways.

First, I will increase the data in play, and hence the leverage on the move, by introducing as much data as I can into the backbone prior to considering Galatians. Some years ago, I thought that this meant inserting both 1 Thessalonians and Philippians into the sequence. I have recently been persuaded that 1 Thessalonians belongs elsewhere and only Philippians belongs here; nevertheless, this positioning alone is important. When the astonishing appropriateness of the location of Philippians just prior to Romans — during a trial in Corinth — is grasped, the introduction of Galatians into the sequence becomes more compelling. In addition, I will argue for a slightly new interpretative angle on material many scholars excise from Philippians as an interpolation — 3:2–4:3 — material that, understood correctly, contributes significantly to framing considerations.

Following this new insertion, we will reconsider Knox’s suggestion that Galatians, and 2:10 in particular, references the collection. I will argue that the presence of the collection in Galatians is rather more subtle than Knox supposed, yet still detectable, and hence Knox’s original claim can be deployed as an important secondary, corroborative argument in support of this location for Galatians.

(iii) Knox noted that a datable event could possibly be detected in Paul’s epistolary data, something of which many scholars seem to be unaware.34 He gestured specifically toward what we will call “the Aretas datum” (1987 [1950], 67). However, he did not press forward to a precise analysis, a point where the present project will go further than his proposals. I will argue that the event can be dated specifically, between the fall of 36 and the spring of 37 ce, and that this priceless chronological insight can anchor the rest of the biographical material.35 The other potential date Knox noted derives from Acts and so will not be addressed here, and his mistake in relation to authenticity led him to overlook the only other possible datable external reference found in a Pauline letter, which will be noted here momentarily. In these discussions, then, we will be rather more consistent and precise than Knox was in the important matter of absolute dating, although pressing forward along lines that he has already identified.

(iv) Following these three opening and very important explanatory steps, Knox turned to consider where Paul’s other letters might fit into his developing frame. We have had cause, however, to question his inconsistent limitation of the evidence under consideration. So where Knox considered only a few letters, we must broaden the discussion to a consideration of all the canonical letters bearing Paul’s name that have yet to feature within his schema (although one has already been located, namely, Philippians). This single step in Knox will expand into three as we consider the relevant groups of remaining epistolary material — the Thessalonian correspondence; the interrelated cluster of Philemon, Colossians, and Ephesians; and 1 and 2 Timothy and Titus. Certain distinctive considerations will have to be brought to bear in relation to each cluster of letters, since questions of authenticity arise with special force during these steps in the overall argument.

We turn initially to consider the Thessalonian correspondence, and some surprises again await us. We encounter here for the first time the way in which a rigorous application of Knox’s methodology, coupled with a rejection of invalid, circular argumentation, opens up important new angles on epistolary questions long considered closed.

First Thessalonians is the first letter that does not seem to fit into the epistolary sequence already enumerated. It seems earlier. But the frame of reference now in play opens up the fascinating possibility that it could be much earlier, that is, anytime within the 40s ce, although at this point we cannot say exactly when. But its similarity to 1 Corinthians invites further explanation. So we will consider as well whether Paul kept and circulated copies of his letters that had been sent to other destinations, ultimately opening an important avenue of explanation concerning the creation of the original Pauline letter collection, along with new insights into the letters’ original receptions. (By now we will have added two new methodological considerations to Knox’s case in terms of Paul’s ancient epistolary reception: the phenomena of Nebenadressat and of copies of letters circulating in places other than their named destinations.)

After the consideration of 1 Thessalonians, it seems most obvious to turn to 2 Thessalonians. And here we learn that many of the reasons usually adduced for this letter’s pseudonymity are rooted in false methodology and, often, an oddly fragile biography (that is, in something approaching an Acts fundamentalism). We will find that a judgment of authenticity seems surprisingly solid, and this opens up further fascinating possibilities for the frame.

Second Thessalonians seems to contain strong references to “the Gaian crisis,” which unfolded from 39 through early 41 ce. Suitably interpreted, this offers a second critical reference or date, one that fits nicely with our first — the Aretas datum — which stemmed from late 36 to early 37 and was associated with Damascus and Jerusalem. The links between 2 Thessalonians and 1 Thessalonians now draw both letters into this early and dramatic phase in the mission of the church, although the present allusion is not as delimited as the Aretas datum. First and Second Thessalonians — probably, for certain fragile reasons, in this order — turn out to be our earliest extant sources for Paul and may, moreover, be dramatically earlier than most scholars have previously suspected, being written during the early 40s ce. They have always been fascinating and important texts, but if these judgments hold good, then their value for the Pauline interpreter increases still more. To possess sources from this phase within the expansion of the early church, perhaps just over a decade after the death and reported resurrection of its founder, would be nothing short of extraordinary.

(v) We must now try to position Philemon, Colossians, and Ephesians in relation to this developing frame, as well as to make the appropriate judgments about their authenticity.

Few scholars in fact question the authenticity of Philemon, perhaps because it seems rather innocuous. But it cannot be positioned with any confidence in relation to the broader Pauline frame without a consideration of at least Colossians. And of course we have a new perspective for adjudicating the authenticity of Colossians by now within our overall discussion. Its authenticity can be addressed in the terms that were introduced to adjudicate 2 Thessalonians, including appeals to the modern, statistically assisted analysis of style, that is, to “nontraditional authorship ascription.” Much previous analysis seems worryingly out of date in this regard. And both Colossians and Ephesians obviously raise the further issue of dependence, since these two texts are so closely intertwined. But our analysis of 1 Thessalonians in relation to 1 Corinthians will already have opened up a new angle on this phenomenon. We noted there the possibility of letter copies being kept in the Pauline circle and appearing in nondesignated churches, to which we should now add considerations of orality and performance. (The seminal methodological figure concerning orality is Walter Ong.)36 And in view of these considerations, it seems that nothing decisive stands against a preliminary affirmation of the authenticity of Colossians. Certainly, there is insufficient data at this point to reject its authenticity, so it will remain for the time being within the Pauline group.

The numerous connections between Philemon and Colossians now connect both letters with the nearby Lycus valley. (The recent Stanford project on ancient travel supervised by Walter Scheidel — ORBIS — becomes exceptionally helpful at this juncture.)37 Moreover, they seem to hover just before the principal epistolary sequence that has already been identified, so sometime around Paul’s second visit to Jerusalem, when so many vital issues concerning the shape of the early Christian mission were negotiated — something they may now shed light on.

Ephesians fares similarly, although it seems most likely to have been composed first, insofar as this distinction remains relevant. Derivation from Colossians would cause difficulties for authenticity, but determining the causality of this relationship is in my view impossible on stylistic grounds alone, especially when considerations of Paul’s incarceration, orality, and resulting compositional dynamics are taken into account. The composition of texts in prison is a common phenomenon within the Christian tradition, but it is not well understood by modern scholars, largely owing to its peculiar context. A brief consideration of some of the dynamics surrounding the production of prison literature is therefore instructive.

If Ephesians is authentic, however, it seems necessary to affirm its identity as the letter to the Laodiceans mentioned in Colossians 4:16, now often thought to be lost. (So we find that it has been hiding in plain sight.) This takes the collection of Pauline letters judged authentic and sequenced thus far to ten.

(vi) It remains only to consider 1 and 2 Timothy and Titus. The most appropriate letter to evaluate first is Titus, since it seems to fit most obviously into our developing frame.

The correct method for evaluating authenticity is quite well established by this point in our project. We consider a letter’s account of its immediate circumstances or exigence. We then ask whether and where it fits into our developing critical biography. After this, we address issues of style and dependence, the former suitably informed by modern statistical analyses and the latter by suitably nuanced explanations concerning the circulation and composition of Paul’s letters. Then finally, we probe for possible mistakes by a putative pseudepigrapher, for example, blatant anachronisms.

Titus struggles more with respect to these considerations than the other disputed letters have, leading ultimately to a judgment of pseudepigraphy, and similar dynamics then play out with 1 Timothy. This letter arguably contains explicit anachronisms, along with other difficulties for authenticity. However, the situation with respect to 2 Timothy is more finely balanced.

Arguably nothing decisive is apparent that might lead to an immediate judgment of pseudepigraphy. But there is a steady accumulation of doubts leading ineluctably in that direction. So these three letters do ultimately end up being grouped together as “the Pastorals,” and are excluded from further involvement within the construction of the preliminary frame — although it needs to be emphasized that all preliminary framing judgments are provisional. This leaves the number of canonical Pauline letters judged authentic at this preliminary stage, on internal grounds, at ten, with the sequence established by the end of chapter 5, at which point we turn to consider quickly a useful corroborative contention.

The shape of the original Pauline letter collection is arguably apparent in the data surrounding Marcion, who seems to have inherited an edition of Paul’s letters composed of just these ten — an important correlation — that was apparently addressed, rather elegantly, to seven churches (see also Rev. 2:1–3:22). That is, Marcion seems to have worked with a collection of Paul’s letters that contained Galatians, 1 and 2 Corinthians, Romans, 1 and 2 Thessalonians, Laodiceans (Ephesians), Colossians and Philemon, and Philippians. But he seems not to have known of
1 and 2 Timothy or Titus, suggesting that those letters might have been composed, at least in part, against his program. This attestation can be challenged; hence, our discussion also has to address briefly certain considerations in the evidence of Clement, Ignatius, Polycarp, Irenaeus, and Eusebius. Nevertheless, it will be suggested that the simplest explanation of this complex situation — which has just been noted — is also the most persuasive. And this evidence concerning transmission does seem to be a useful confirmation of our judgments reached on internal grounds. That the two lines of investigation match up exactly can increase our confidence in the entire disposition.

In sum, after these six steps, it seems that we will be able to offer an epistolary frame for Paul in two senses — that all the letters initially deemed authentic will now be nicely framed in terms of their circumstances, perhaps to a surprising level of detail, and that the evidence underlying this frame will be entirely epistolary and hence primary. Our task will be complete. But one final issue needs to be briefly considered before this initial methodological discussion draws to a close — namely, the ongoing status of this reconstruction once it has been successfully derived.

By the end of the argument, we will have generated an account that frames Paul’s authentic letters. This will have been derived in the correct way, using the appropriate arguments and evidence. We will have used initially only epistolary material, and will have processed it using arguments sensitive to problems of vicious circularity, different levels of construal, and the appropriate deployments of criticism and doubt. We will have resisted inappropriate atomization and tacit psychological dependence on Acts. The result should be an account that frames Paul’s letters by rigorously answering the key questions of authenticity, integrity, sequence, intervals, and reference. This can then serve as the basis for richer explanations of other aspects of Paul, which can now proceed in a way that avoids begging the question and similar problems. But there is one final issue hanging, concerning the relationship between those later, richer explanations and this frame that has been derived in first position, as a preliminary explanation, and hence in a deliberately restricted and controlled fashion. Does this frame need to be held constant during those later investigations, come hell or high water, so to speak, or can it be judiciously modified? In short — is it, to a degree, also provisional?

I think the answer to this question is yes. The frame is provisional, provided its possible later modification is undertaken appropriately.

It seems to me that the careful examination of texts later on might lead to comprehensive accounts that encourage us to reverse some of our initial judgments. So, for example, a student of Colossians might be able to show in due course how all the complex theological concerns of that letter just do not fit plausibly with those apparent now in the other nine letters presently judged authentic and that it ought to be excluded from the Pauline canon. Or a student of 2 Timothy might succeed in running this analysis in the opposite direction, arguing that that letter ought to be included — and the grounds for the initial exclusion of 2 Timothy are not that strong. Both analyses would accept the way that our preliminary frame has established the burden of proof. And it only seems fair to concede that these arguments, and others like them, are possible. Our framing account is provisional. However, it is vital for us to appreciate that any later arguments of this nature, seeking to reverse the frame’s judgments at particular points, will carry certain specific explanatory burdens.

Any such suggested modifications will have to show how the original framing decision that turns out to be incorrect can be corrected in a way that does not undermine the rest of the frame that is currently being used to make that case. If it fails to do this, it will risk falling back into vicious circularity. Put a little more colloquially, it will be sawing off the branch that it is sitting on. Hence, alternatively — if the branch is indeed being sawed off — an entirely different frame will have to be demonstrably derived that still supports its judgments, and this looks like a difficult task. The risk lying in wait here is the collapse of all the data into an unframed situation, at which point we are back to the central conundrum with which this chapter began — the inability to adjudicate between any of the main interpretative options for Paul at the most basic levels of contingency and coherence, this being the very distinction that makes a complaint about the authenticity of Colossians possible.

Consequently, we should affirm that the framing account being offered here is provisional. But any future modifications will have to be undertaken very carefully indeed if we are not to return unwittingly to a fundamentally distorted and confused situation. A preliminary frame might turn out to be rather better than a modified frame that is really no frame at all.

With this caution concerning the frame’s provisional nature duly noted, we can now turn to the fascinating but difficult task of reconstructing it in detail. And we will begin where Knox did: with the reconstruction of a critical one-­year sequence that can serve as the frame’s backbone. This is argued in relation to 1 Corinthians, 2 Corinthians, and Romans, and their apparent concern, at least at times, with a collection for the saints in Jerusalem.

1. His use of the language of structuralism, presumably in distant dependence on the work of Noam Chomsky, is somewhat unfortunate, since it mobilizes false analogies. Some structuralists (notably, Propp and Greimas) claimed that various folk stories contained underlying formulae that could be discerned across individual instances, allowing the derivation of actantial maps of texts that identified standard actors such as heroes and helpers. Beker might have been appealing to this basic notion with his claim concerning coherence. But this claim is problematic. There is arguably no “underlying” structure or “deeper” entity in stories and texts. Stories do not move through spaces that have shallow and deep locations, as a submarine might move across the ocean closer to or farther away from the water’s surface. Observable similarities between texts are possibly better explained in terms of a shared tradition and set of practices, equally recognizable to authors and readers or auditors. This proposal does, however, open up important explanatory opportunities for Paul’s texts, provided that they are pursued correctly. Overlapping narratives might become apparent as readers sensitive to these possibilities analyze and compare texts. But these are not “deep structures.” (The method is pursued famously and rather more responsibly by Hays 2002 [1983]; B. Longenecker 2002 is a further useful overview and assessment of the method with respect to Pauline analysis.)

2. Beker suggests at one point that “the coherent center of Paul’s thought is constituted by a symbolic apocalyptic field, originating out of a profound experience, which comes to expression in an apocalyptic grammar” (xx).

3. See, e.g., the topical organization of Dunn (1998) and Schreiner (2001). The famous earlier treatments of Whiteley (1964) and Ridderbos (1975 [1966]) are also indicative. Schnelle (2005 [2003]), with its initial biographical discussion, is a partial exception discussed in more detail shortly, although ultimately it still treats Paul’s thought topically (387-597). More fully exceptional is J. Becker (1993 [1989]), which is biographical throughout. As we turn to note Bultmann’s vastly influential work, we step from specialized treatments of Paul per se to treatments of Paul’s thought within the broader ambit of NT theology as a whole. But the same judgment tends to be appropriate there. Bultmann himself treats Paul’s thought entirely under two topical headings — “Man Prior to the Revelation of Faith” (1951-55, 1:190-269) and “Man under Faith” (1:270-352). Recent major contributions to NT theology are reviewed helpfully by Rowe (2006); see also Matera (2005). (Rowe’s monumental review treats Stuhlmacher 1991-99; Gnilka 1994; Strecker 2000 [1996]; Vouga 2001; Hahn 2001-5 [2002]; and Wilckens 2002-5; as well as Caird 1994; Esler 2005; I. Marshall 2004, 209-488; and Thielman 2005; noting in closing the more expansive Childs 1992.) To these figures we should now add, at the least, Swartley (2006, 189-253); and Matera (2007, 99-258). Topical treatments are evident throughout.

4. Most famously 1987 [1983]; but see also 1988. His corresponding account of NT theology can be found in 2000 [1990].

5. And more so in 1983 than his more famous 1977 (which treats Paul directly in 431-523). To reiterate, I am referring here to what we might dub the “early” or “classic” Sanders; for the views of the “later” Sanders, which are introduced momentarily, see his 2008.

6. Sanders has recently clarified that he distinguishes conceptual coherence in Paul from systematizing. The former is, in his definition, a certain sort of “clumping” or “clustering” of notions together; the latter is, in his definition, a more hierarchical and rigid taxonomical ordering of ideas or notions (2008, 325-30). He is happy to use coherence (in his definition) to describe Paul’s thought, but not systematizing. Here, however, I am using Beker’s definition of coherence, which is probably more like the “systematizing” that Sanders rejects, although Beker too would doubtless reject the suggestion that Paul systematized in quite the way that Sanders describes this in 2008, i.e., essentially taxonomically. However, Beker thinks that Paul’s underlying conceptuality is a “core” of convictions about the Christ event that is translated in different contexts, and this is in fact a reasonable account of what many contemporary theologians would recognize as “systematic theology,” where the eschewal of overly rigid systematizing and of taxonomies is widespread. So, e.g., Karl Barth in his Church Dogmatics clearly attempts to provide a rigorous account of appropriate Christian thinking, i.e., of systematic theology; indeed, his conceptualizing could hardly be more systematic. But he firmly repudiates any suggestion that his rich account is a “system,” along with the very possibility that God’s activity in Christ can be systematized — the sort of conceptual orchestration that both Sanders and Beker also reject. And this realization concerning the actual nature of most systematic Christian thinking limits the force of its repudiation in relation to Paul. It seems that if the descriptor “systematic” can be understood in more flexible and appropriate terms, then it can be applied to Paul’s thinking, at least as a possibility. (Whether it is the best account of his underlying thought must wait for the framing of his letters.)

7. Lakoff (2002) provides an interesting point of comparison when he analyzes the “bi-­conceptual” nature of most current American voters; two fundamental and quite different conceptual models inform their political activity. The situation has a straightforward explanation in underlying bodily terms, namely, the simultaneous coexistence of different worldviews within the brain. Once both have been introduced, political decision making is then in part a matter merely of activation, a view of mental and bodily activity that reinforces Sanders’s intuitive sense of the same operating in Paul in soteriological and ethical terms.

8. A classic account of this is also regrettably superficial: Buck and Taylor (1969). Buck’s approach lay at the center of a cluster of developmental analysts that included Riddle (1940) and Hurd (1983 [1965]). R. Longenecker, my Doktorvater, also emphasized development in Paul’s thinking strongly, although he did not publish any sustained treatment of it. However, the approach is noted briefly in his 1998. More rigorous is Tatum (2006); he has influenced his Doktorvater, Sanders (see now 2008).

9. The modern recovery of his discourse is associated especially with the work of Alasdair MacIntyre (see his 1988; 1990; 2007 [1981]).

10. It is the more so if his writings, read in these basic terms, are viewed as authoritative, if not infallible.

11. The careful address of individual letters in their own terms is an important interpretative move in certain situations, and will be undertaken here esp. in chs. 4, 5, and 6; it is usually referred to later as the text’s “immediate implied exigence” — the account that a letter supplies of its own immediate exigence. But my claim here is that broader judgments about the nature of Paul’s thinking will be impossible if analysis stays at this level. Indeed, deeper analysis even of an individual letter will ultimately be frustrated as well, since judgments about the contributions of Paul’s theology will be impossible, as will accurate assessments of any implications flowing from adjacent letters and situations in Paul’s life.

12. Framing is explained helpfully by B. Johnson (1980 [1978]), as noted earlier.

13. If what I have argued up to this point is broadly correct, then interpreters will need to agree on this framing account if they are to move forward into richer explanations of Paul’s life and thought. This frame will need to be the basis for some consensus. The simpler, then, not to mention the clearer, the better. Interpreters tend to find such framing discussions complex and forbidding. The analysis of the details of Paul’s life, and of things like the dates when he arrived in certain cities, can lead interpreters down various rabbit holes (and into entire warrens). We need at all costs, then, to enhance this account’s clarity and accessibility, and this can presumably be assisted by minimizing what we need to agree on.

14. This framing account is still an integrated explanation of a significant swath of Paul’s life and hence in part an explicit rejection of methodological atomism. We cannot assume that the broader explanation of Paul will emerge if we work independently on its smaller constituent units — that localized interpretation, done well, will simply lead to the crafting of small explanatory blocks that can fit together in only one way, so that, by doing this repeatedly, we can gradually build up a convincing overarching edifice. To be sure, there is a partial truth in this approach. But the Pauline data is linguistic and thus unstable. In effect, the building blocks change shape depending on the broader explanations of the field that are in play. Linguistic data will shift depending on the rules, practices, and connotations thought to be operative. Localized interpretation that is insensitive to this will consequently risk endorsing a questionable overarching explanation in the name of a naïve presuppositionless objectivity, invalidly of course. It is better, then, to appreciate that Pauline explanation is seeking to grasp an integrated field of evidence of great complexity that shifts, to a degree, depending on the explanations being brought to bear at any given moment. Hence, explanations must be attempted at a suitably controlled and broad level — in this case, in an account that frames all the letters together. This seems to be the correct explanatory unit for our discussion.

15. But there should be a palpable sense of relief when discussions of things like “Paul and the law” are liberated from their inextricable tangle with various biographical claims. Multiple biographies of Paul and the law are possible, and adjudicating between them is largely impossible when all the data and all the explanatory levels are on the table at the same time. Paul’s view of the law should be assisted if not controlled by a biographical frame that is already in place, and on grounds that are independent of the adjudication of the issue.

16. This approach will also allow a useful control over the influence of presuppositions derived from tradition. Such presuppositions are not in themselves necessarily bad, but they can function unhelpfully.

17. Resisting such arguments psychologically is a different matter. Kahneman (2011) documents numerous ways in which such methodological vigilance can falter, and we will call upon his exposés frequently in what follows.

18. These constraints sharply limit the value for this effort of the vast majority of the secondary literature addressing matters of Pauline biography and the circumstances surrounding the composition of the letters, as will become apparent in much that follows.

19. A lively introduction to the roles of Cartesian and post-­Cartesian “critical” thinking within the modern university can be found in Readings (1996, esp. 44-69). Deeper critiques specifically of the Cartesian subject are C. Taylor (1989); MacIntyre (1990); and, in an important earlier treatment that dialogues especially informatively with Bultmann, Thielicke (1974 [1968]). More accessible overviews of the entire development of this discourse are supplied by Gellner (1988) and Poole (1991). I cover some of this territory in ch. 9 of my 2009 (295-309). Also significant is Hauerwas (2002), the issues becoming apparent as he contrasts the thinking of William James and Reinhold Niebuhr, two doyens of modernity, with Karl Barth, the scourge of modernist thinking within theology.

20. Polanyi (1964 [1958]) has been particularly seminal for my understanding here (see also his more accessible 1966). Many of the critical theological consequences are grasped and articulated peerlessly by T. F. Torrance (see esp. his 1969a; 1969b; 1971; 1976).

21. How might we recognize inappropriate doubt masquerading as valid criticism? Such doubt generally does not attend to the actual data and its explanation, falsifying it directly. It begs the question. Or, more commonly, it suggests a comparative situation but fails to supply the comparison; a given argument might be pronounced insufficient to convince, but what exactly establishes argumentative sufficiency is not stated (and usually cannot be). Of course, such judgments are meaningless without an overt standard or measure of sufficiency. And that measure is the data itself in relation to the broader object under investigation and the current explanation in play! Do these actually match up, or is a problem discernible in their relationship(s)? If the latter, the appropriate critical process should elicit doubt, along with the modification or abandonment of the hypothesis. Modification or the clear provision of an explanatory alternative is a signal that the appropriate critical method and doubt are operative. Without these elements, a doubting critic runs the danger of merely posturing.

22. His key work is 1987 [1950], esp. pp. 31-52, with the epistolary evidence treated on pp. 31-41. It was anticipated by 1936 and 1939, and intriguing subsequent reflections followed in 1983 and 1990.

23. Some have already been mentioned, notably, Hurd (1983 [1965]); Buck and Taylor (1969); Tatum (2006). Sanders (1977, 432) provides passing support. But see also Hare (1987 [1950], ix-­xxii), in the introduction to Knox (1987 [1950]); Donfried’s entry concerning Pauline chronology in the ABD (1992); and, perhaps most significantly, Luedemann (esp. 1984 [1980]; but see also 1989a; 1989b [1983]). Jewett (1979) is also an important figure, who is close to Knox in certain respects but charts his own course in others.

24. See, to a degree, Knox himself (1987 [1950], 3-16); but also (i.a.) Hurd (1983 [1965], 13-42); and Luedemann (1984 [1980], 1-19; 1989a).

25. It is often overlooked that Knox himself believed that his reconstruction provided “a strong affirmation of the historical value of Acts” (1987 [1950], 10).

26. We cannot really make this claim without appealing to doctrines of inspiration and inerrancy, and I will appeal to neither here. But even for those that do, I suggest that the situation does not need to be this rigid. A cursory comparison of the Gospels suggests that variation in the sequence of stories has widespread canonical legitimation. The stakes are consequently lower in relation to the sequence of the data in Acts as against the veracity of the individual episodes — what we can call sequential as against episodic reliability. So the possibility raised by Knox’s approach that the arrangement of Acts is informed by literary and theological considerations as against strict chronological order ought to occasion no alarm. And we can safely leave judgments about the veracity of individual episodes in Acts until later on, when we can begin to compare stories about Paul from Acts with the gleanings about the same events gathered from the letters.

27. Kahneman (2011, esp. 7-9, 129-45 [chs. 12-13], 425-27 [app. A, originally published in 1974]) documents the advantages that explanations of data characterized by “ease” and “availability” have over explanations not so characterized.

28. When a negotiation has been initially primed with a false and deeply distorting anchor, further negotiations must stop entirely and the terms of the debate be psychologically reset. Kahneman’s somewhat flamboyant advice in this regard: “If you think the other side has made an outrageous proposal, you should not come back with an equally outrageous counteroffer, creating a gap that will be difficult to bridge in further negotiations. Instead you should make a scene, storm out or threaten to do so, and make it clear — to yourself as well as to the other side — that you will not continue the negotiation with that number on the table” (2011, 126). The present study is in a certain respect, by concentrating almost entirely on epistolary material, storming out of the room if Acts data is introduced prematurely into framing.

29. See esp. his 1990.

30. A consideration of just these letters in what follows will be “sufficient trouble for the day.” Moreover, if we cross the threshold from authentic to pseudonymous letters within the canonical material, it is improbable that further authentic epistolary material will be found outside the canon. If a canonical letter is demonstrably shown to be pseudonymous, then the other ancient letters bearing Paul’s name outside the canon will likely prove to be pseudonymous as well in due course, namely, 3 Corinthians, Laodiceans, the Letter to the Alexandrians, and Paul’s contributions in the Letters of Paul and Seneca, since these generally seem like less persuasive and skillful forgeries than any canonical counterparts would be (although none have yet been identified). A brief introduction to these texts and the issues can be found in Pervo (2010, 96-116), although this is not a methodologically reliable book. Similar considerations apply to Ehrman (2013, 425-32 [3 Corinthians], 439-45 [Laodiceans], 521-22 [Letters of Paul and Seneca]). The question of Hebrews arises here as well in an interesting way. Because it does not bear Paul’s name explicitly, we will not consider its Pauline authorship. This question must be faced in due course, but precisely because it does not bear that explicit mark, it must, strictly speaking, make its way (or not) into the Pauline corpus from the outside in rather than vice versa. This investigation too is best undertaken elsewhere.

31. In fact, he was not convinced that any precise time spans for this event can be determined, other than that it must have occurred before Aretas’s death in 40 ce. He generated his chronological schema by calculating the intervals and the basic frame and then tying it to an event described in Acts: Paul’s trial before Festus. He dated this to 55 ce (again, basically correctly in my view), citing Tacitus, Annals 13.14-15; Josephus, Jewish Antiquities 20.8-9; and Eusebius’s Chronicon. In my view, his framework is basically out from this point by one year, and then more imprecise later on.

32. Knox’s final move, which presupposes the detailed analysis of a great deal of data from Acts, is best undertaken in a separate study.

33. Strictly speaking, partitioning 2 Cor does not necessarily raise a complex set of interim events, i.e., a complex reconstruction between 1 Cor and what is left of 2 Cor in its traditional position, and vice versa. We will assess the precise methodological relationship between partition theories of 2 Cor and reconstructions of the interim events in due course. Furthermore, the complication of this sequence in various ways, due to partition and/or complex interim events, does not necessarily entail the abandonment of a framing project altogether. But that project would have to be significantly reconceived, as it is, e.g., in Tatum’s (2006) important study. Tatum is not persuaded that, among other things, the issue of reference can be addressed plausibly. He (among others) is also unconvinced by Knox’s first step here, namely, the strict and relatively simple sequencing through one calendar year of 1 Cor, 2 Cor, and Rom, in that order.

34. The denial of this claim is documented in more detail in the final section of ch. 3, where I attempt to refute it.

35. The case is argued in detail in my 2002 and summarized here in ch. 3.

36. See esp. his 2000 [1967]; 2002; 2012 [1982].

37. See ORBIS (2013); also Scheidel, Meeks, and Weiland (2012); Scheidel (2013).