William Lane Craig, Herodotus, and Myth Formation (1999)
Richard Carrier
This essay addresses one specific argument made by William Lane Craig, to the effect that “tests” from Herodotus demonstrate that myths or legends (such as resurrection appearances or an empty tomb) cannot grow within a single generation. A great deal more could be said about myth formation than is covered here, but the aim right now is simpler: to show how Craig misrepresents one source and thus creates an empty argument out of whole cloth. I will preface the essay with a general point about the Gospels as history.
The Gospels as History
The context of the argument in question is a defense of the historicity of the resurrection, in two parts [1]. The first part argues that “Paul’s information makes it certain that on separate occasions various individuals and groups saw Jesus alive from the dead” about which Craig says, “This conclusion is virtually indisputable.” I disagree, but I have argued that case elsewhere [2]. In sum, the existence of appearances is not “indisputable,” since the influence of one man (Paul) on all the recorded traditions is embarrassingly great in comparison with almost any other “virtually indisputable” historical event. The possibility of invention by him is not easily refuted. Nevertheless, I do believe appearances of some sort occurred at least to one person, maybe others, but we have no access to accurate accounts of those events, and thus can say little with certainty about them. They could easily have been of a spiritual, hence subjective nature, and this is not good evidence (for us) of a real rise from the dead. Finally, what has survived has undergone transformations of a legendary sort, affected by dogmatic disputes within a growing church, by the need of leaders to assert authority, by credulity and piety among believers, through at least one if not two generations of oral history, including a major catastrophic event (the destruction of the original church, or at least the creed’s city of origin, by war in 70 A.D.), just to name the most prominent factors. But even if there were actual appearances, non-miraculous explanations remain for them which cannot be confidently falsified by the existing evidence. Given the lack of any modern support for the occurrence of miracles, we are left with no rationale for proposing miracles in antiquity when perfectly reasonable natural explanations are available [3].
But it is the second part of Craig’s argument that interests us here. It goes like this, “in order for these stories to be in the main legendary, a very considerable length of time must be available for the evolution and development of the traditions until the historical elements have been supplanted by unhistorical” [emphasis added]. To begin with, this argument does not lead to the conclusion Craig wants, for few if any real historians today think that the stories are “in the main” legendary, or that all the historical elements have been entirely replaced. Even if they did, refuting such a view does not demonstrate the historicity of any miraculous features of the accounts–since those are not “in the main” the contents of the Gospels, and they may not have supplanted so much as simply added to the historical core of the story.
The obstacle is not the lack of historical elements, but our inability to determine which elements are historical. We can surmise, for example that a man named Jesus preached a reform of Judaism in the time of Pontius Pilate the prefect of Judaea, which involved particular apocalyptic and ethical ideas and the gathering of disciples, that he travelled to certain places, was opposed by Jewish authorities and crucified by the Roman authorities on at least the pretext of leading a rebellion, etc. But we cannot be sure how much of even these details are historical, or to what degree–were there several such men, who were conflated and thought to be the same man? How much of the story is a revision or exaggeration created by those who shaped the church after him, such as Peter and Paul? Was Jesus really starting a rebellion, but when killed his followers changed the story to perfect a less direct means to the same end? How much is rumor created by believers and passed on? How much is dramatic invention? How much is purely the propaganda of Jesus himself or his advocates? Or rhetoric aimed at saving souls or securing authority among vying factions? Before the Gospels can be honestly discussed, one must be able to at least attempt an answer to these questions, and others like them, or else admit that they cannot be answered beyond the limits of human speculation, though the latter entails that the real history is all but lost in the details.
Yet Craig never even acknowledges such questions, or dismisses them outright. He makes it seem as if it is either all history or all legend. No actual historian thinks that way about any text [4]. And yet if Craig acknowledges that at least some of the Gospel stories are legendary or otherwise distortions or inventions, then he must explain why the miraculous features are not to be included among them. His approach allows him to hide this problem by focussing on Paul, but Paul does not tell us anything about the Gospel stories apart from a concise statement of the creed itself, which is simplistic and vague and thus open to numerous interpretations (to which it was no doubt subject even in Paul’s day). That Paul “believed” there was an empty tomb or bodily appearances is never stated by him. It must be “interpreted” from the text. That alone places any such conclusions of shaky, subjective ground. Had we never had the Gospels we would not have any reason to suppose such details were in Paul’s mind when he wrote what he did. Thus, the Gospels must be the starting point for any such speculation–meaning that Craig must face the fact that either the Gospels contain some legends or they do not. The latter is absurd. The former sweeps away any claim to certainty about the appearances or the empty tomb, at least in respect to the details (e.g. who saw what when, what exactly was seen, etc.) if not the basic facts themselves.
Misrepresenting Sherwin-White
Craig’s main defense of the “it cannot be myth” argument is based on a misrepresented reading of A.N. Sherwin-White [5]. Craig’s argument is reproduced here in full to make sure I am not misrepresenting it myself:
Professor Sherwin-White is not a theologian; he is an eminent historian of Roman and Greek times, roughly contemporaneous with the NT. According to Professor Sherwin-White, the sources for Roman history are usually biased and removed at least one or two generations or even centuries from the events they record. Yet, he says, historians reconstruct with confidence what really happened. He chastises NT critics for not realizing what invaluable sources they have in the gospels. The writings of Herodotus furnish a test case for the rate of legendary accumulation, and the tests show that even two generations is too short a time span to allow legendary tendencies to wipe out the hard core of historical facts. When Professor Sherwin-White turns to the gospels, he states for these to be legends, the rate of legendary accumulation would have to be ‘unbelievable’; more generations are needed.
To someone unfamiliar with the text he is citing, it certainly seems as if Sherwin-White believes that no part of the Gospel stories is legendary, that for legends to appear in them is “unbelievable,” that “tests” (plural) have been performed on the text of Herodotus, and that these tests “show,” convincingly enough to use the argument at all, that legends require many generations to develop. We will see that Sherwin-White never argues any of these points, and thus Craig is not representing his source fairly or correctly. I find this to be either a dishonest or an incompetent use of a source that Craig should not be proud of.
The only saving qualification is that Craig says the tests prove that it is impossible for legends to wipe out some undefined hard core of historical facts. But even this entails a more subtle misrepresentation of Sherwin-White. This brings us back to the fault line in Craig’s argument noted before: that the appearances and empty tomb stories are the amplified and distorted record of some core historical facts does not entail that the hard core of fact was wiped out. The “hard core” of fact might be that the appearances were spiritual visions, and that there was a tomb but it’s emptiness was of a spiritual character–e.g., that which was Jesus is risen, the body being irrelevant. This is not even counting the possible natural causes for physical appearances or empty tombs which could then have been lost or overlain with legend. If there are any legends clouding the facts anywhere else in the Gospels, then the same problem could just as well exist here, and so Craig’s entire argument fails to establish the historicity of the stories as told. In other words, it follows that Craig must be implying that Sherwin-White agrees that the physical appearances and empty tomb are the very “hard core” of historical fact that cannot be legendary. Otherwise, there is no reason to cite him in this context–for if Sherwin-White’s actual argument does not entail this, then, whatever Sherwin-White demonstrated, it is irrelevant to the argument Craig is building. If that is the case, then Craig should not be referring to him at all, or at least he should honestly lay out the features of Sherwin-White’s argument that weaken or undermine Craig’s conclusions, and not present him as only being in agreement.
What Sherwin-White Actually Wrote
Consider what Sherwin-White says about Acts. “For Acts the confirmation of historicity is overwhelming. Yet Acts is, in simple terms and judged externally, no less of a propaganda narrative than the Gospels, liable to similar distortions” [6]. In other words, there are genuine historical details in Acts, but it is clouded with “propaganda” and “distortions,” just as the Gospels are. This does not support Craig. It supports the belief that Acts and the Gospels contain untruths. Sherwin-White does not tell us which details are to be regarded as historical, but clearly he believes that legends and falsehoods can and do exist in these documents, and thus he does not believe that a few generations are “not enough” for such elements to appear. Why does Craig represent him as thinking otherwise? When we actually read all of Sherwin-White’s argument we discover that he merely objects to the notion that “the historical content is…hopelessly lost.” Thus, he is arguing against an extreme minority of scholars who reject the Gospels entirely. He is not arguing for the historicity of the empty tomb or the physical appearances of Jesus, and his arguments cannot be extended to include them [7].
Sherwin-White then turns to Herodotus for an analogy. Regarding accounts of the Persian Wars:
[They] are retold by Herodotus from forty to seventy years later, after they had been remodeled by at least one generation of oral transmission. The parallel with the authors of the Gospels is by no means so far-fetched as it might seem. Both regard their material with enthusiasm rather than detached criticism. Both are the first to produce a written narrative of great events which they regard as a mighty saga, national or ecclesiastical and esoterical as the case may be. For both their story is the vehicle of a moral or a religious idea which shapes the narrative….Yet the material of Herodotus….has not been transformed out of all recognition under the influence of moral and patriotic fervour [8]
Note the details: Sherwin-White admits that even in Herodotus there has been “remodeling” and a lack of “detached criticism” (i.e. “a receptiveness to falsehood”) as well as a motivating bias to shape the stories toward an agenda, and that all we can say is that the result has not been mythologized out of all recognition. This is a substantially weaker conclusion than Craig represents it to be, as I have discussed already. When I examine below the actual stories in question, we will see plenty of legendary material being believed, or created, by Herodotus or his audience, thus demonstrating–if we accept the analogy, as both Sherwin-White and Craig apparently do–that the same thing certainly could have happened in the Gospels. I can only assume that Craig did not examine Herodotus himself, yet this exhibits that basic historical incompetence which characterizes Christian apologists, undermining confidence in their conclusions.
Indeed, even Sherwin-White goes too far, driven by his own agenda (to refute the extremists). He, like Craig, fails to note the crucial differences between the Gospel authors and Herodotus (not the least of which being the anonymity of the former). Yet in any argument from analogy, the differences are as crucial as the similarities and thus not to be left out. The Gospel authors are driven by a religious faith and an evangelical agenda lacking in Herodotus–though he has a moral to tell, for him any facts would do, and thus he was not bound to any particular events, nor did his sources have any unified agenda in that sense. Belief in his account will not grant him or anyone else eternal salvation from the pit of Hell. Also, Herodotus often gives various versions of each account, sometimes he examines them critically, and he outright admits that he does not vouch for the truth of anything he says, but is merely writing down what others have told him [9]. We find none of this honesty and critical thinking in the nameless Gospel authors, making them even more prone to perpetuating legends than Herodotus.
Moreover, there is another detail Sherwin-White omits, and it is worth noting that his audience was the historical community (the book is a collection of his lectures on an advanced topic) who would already be expected to know this: Herodotus is not alone. We have other written sources confirming at least the basic details of the events of the Persian Wars, including, we should not forget, the actual physical remains of the war dead at Marathon, and inscriptions commemorating related battles–the most spectacular of which, a bronze three-headed serpent-column commemorating those killed at Plataea, was moved to the Hippodrome in Istanbul by Constantine the Great and is still there today. Aeschylus–unlike Herodotus an actual participant in the Persian wars–composed a tragedy about the events (Persian Women) only five years after the war’s end, and Thucydides refers to some details as well (1.74, 1.138, etc.), as do Aristophanes, Lysias, Isagoras, and several others, only a generation or less after Herodotus–some also we know but whose works are lost, e.g. Ion of Chios, and another who lived during the war and wrote on it even before Aeschylus: Phrynicus, who composed two related tragedies (Capture of Miletus–even Herodotus cites this work, cf. 6.21–and Phoenician Women) [10]. Thus, not only do we have independent checks on Herodotus, something totally lacking in the Christian Gospel tradition, but we know he was not alone in recounting these events in writing (and thus even later historians would have had written sources we lack).
Ignoring these crucial differences, Sherwin-White contends that “Herodotus enables us to test the tempo of myth-making, and the tests suggest that even two generations are too short a span to allow the mythical tendency to prevail over the hard historic core of the oral tradition” (p. 190), the obvious origin of Craig’s characterization of his argument. As I’ve already noted, few doubt that Jesus and certain other characters, and cultural, geographic and other details of these texts, form a genuine “historical core” worth mining for data. This is generally not in question. What is in question is what mythical and other distortions have entered the account, and almost all historians agree that a great deal of this is present in the Gospels, something Sherwin-White does not dispute (though Craig would have us believe otherwise). Sherwin-White only disputes the notion that myth will destroy the historical core. But even Homer has not done that, for we know that many core details in the Iliad are historically correct, not the least of which being the Trojan war, which he has clouded greatly by myth, invention, and chronological confusion–yet facts are there, and historians recognize the value of Homer despite the extent of fiction in his works. This does nothing to restore historicity to the more unbelievable details of the Gospels, such as the resurrection, nor does it help us to test the reliability of details like the tomb burial for which we cannot be certain of the source.
Nevertheless, despite the irrelevance of his point to Craig’s argument, Sherwin-White’s analysis still shows a central fault in even his own comparison: his “tests” consist of nothing more than one single example, a legend that we already know was circulating in the time of Herodotus (Histories 6.120-3), yet Herodotus recounts the “truth” rather than the legend [11]. But this is not proof against the rapid creation of a legend–for the legend was already there, as Sherwin-White concedes (p. 190), and the fact that Herodotus has to argue against it proves irrefutably that legends do rise within a generation. So this example is only proof of the relatively critical acumen of Herodotus, entirely lacking in the NT authors, or perhaps his lack of any relevant agenda in this particular case–or indeed, even an opposing agenda, to highlight another heroic agent, as Sherwin-White also notes (p. 191). This tells us nothing about the Gospels–the analogy is not even portable. And it does not prove that legends do not rise within a generation–it actually proves the opposite! It is curious how this single, poor, irrelevant example becomes a plural notion of scientific-sounding “tests” in Sherwin-White’s argument, a grand hyperbole that Craig buys hook, line, and sinker–it appears that he only read the one sentence from Sherwin-White, and didn’t actually check to see if the plural (or even the word “test”) was really warranted. He has thus fallen victim to his source’s own rhetoric.
It is also worth pointing out that this “example” is a story that happened in the very city in which Herodotus is writing, whereas we have no evidence that any of the Gospel authors composed their works while in Jerusalem. And, unlike Athens, a major and devastating war had destroyed many witnesses and a great deal of physical evidence by the time the Gospel authors composed their accounts. Sherwin-White is not unaware of this, and couches his conclusion more carefully than Craig implies, saying that “this 0 suggests that, however strong the myth-forming tendency, the falsification does not automatically and absolutely prevail,” emphasis mine, “even with a writer like Herodotus, who was naturally predisposed in favour of certain political myths, and whose ethical and literary interests were stronger than his critical faculty” (p. 191). In other words, all he claims to have proved is that facts are not entirely and automatically replaced by myth even in uncritical authors–although Herodotus is far more critical than the Gospel authors, who never even express a single word of doubt, in contrast to numerous instances of this in the Histories. But almost all historians agree that some facts can survive the crucible of distortion in any source–this does not solve the real question of which ones.
Finally, Craig’s adulteration and misrepresentation conceals the fact that Sherwin-White’s only objective in composing this argument was to defend everything that precedes in his book, namely his historical analysis of the trial of Christ. Thus, his concluding remarks are all about how one cannot dismiss the possible historical core of the facts of the trial. He says nothing about miracles or the resurrection, much less anything to do with his burial or appearances or the empty tomb. Sherwin-White is explicit about the point of his argument in his closing words, “The point of my argument is not to suggest the literal accuracy of ancient sources, secular or ecclesiastical, but to offset the extreme skepticism with which the New Testament narratives are treated in some quarters.” We would never know this qualifying feature of his argument from Craig’s presentation.
“Recent” Legendary Developments in Herodotus
Craig’s argument is that “there simply was insufficient time for significant accrual of legend by the time of the gospels’ composition. Thus, I find current criticism’s skepticism with regard to the appearance traditions in the gospels to be unwarranted.” But this conclusion is based solely on the badly mischaracterized and, in fact, quite irrelevant argument of A.N. Sherwin-White. When we look at Herodotus ourselves, we find that Craig does not have a leg to stand on. It is believed that Herodotus wrote his account between 450 and 420 BC, so we will examine the least believable accounts of events after 490, on the grounds that 40 to 70 years is roughly the gap that falls between the death of Jesus and the Gospels.
Consider the astonishing suicide of Cleomenes, King of Sparta (490 BC). Herodotus tells us without a hint of skepticism that he went “mad” (though his only “mad” behavior was warding off untrusted noblemen with a stick) and was for this locked in a pillory. Then he demanded his guard give him a knife, threatening him repeatedly until he did so, at which he slowly mutilated himself in detailed fashion from the feet upwards, eventually dicing his own belly before dying (6.75). As A.R. Burn observed, “It was officially said that he had intimidated his helot jailer into giving him a knife, and had so mangled himself. The story reeks of the dark mystery of what went on behind the austere, Doric facade of Sparta” [12]. One imagines that Herodotus, born today, would probably repeat without a bit of doubt the report that a dissident in Russia really died by “falling down the stairs.”
Then consider the lengthy speeches of Persian courtiers and their debate with Xerxes about making war on Greece (c. 486 BC), which span ten pages of printed English (7.5-18). It is patently impossible for Herodotus to have any sources for any of these speeches, not even their gist, much less their details–yet the content is cleverly constructed to convey Greek tragic and moral thinking. This is clearly his own “plausible invention” of what sort of debate might have happened, colored by his own moral agenda, but we are not told this–he presents it as if it is a factual story, pure narrative, just as we are given the conversations of Jewish councils and leaders in the Gospels. More astonishing is Herodotus’ record, shortly thereafter, of a horse giving birth to a rabbit (7.57). Then there is a fulfilled prophecy from the God Apollo: the priestess at Delphi accurately foretold the abandonment and burning of Athens, and a decisive naval victory at Salamis (7.140-143). This is clearly a prophecy invented after the fact to fit the actual, and unexpected, course of events, and thus a legendary development within a span of only 40 or so years, which Herodotus repeats as fact [13].
In 480-479 BC we have the account of the second Persian War, with several legends attending–and these were developed within less than forty years, and within the lifetime of Herodotus himself, even in his own city. There is the example of a legend being believed for patriotic reasons: Herodotus, being unbiased, gives both accounts (of the Argives and their political enemies), but the Argives clearly developed, in a very short time, a more face-saving myth explaining their inaction in the war (7.148-152). A similar face-saving myth is reported, with only mild skepticism, at 7.167. There was a popular legend in Herodotus’ day that Apollo bade them to summon the North Wind in a naval battle, and that it came upon their bidding, a story which Herodotus shies from outright challenging (7.189), then the Persian Magi cast a spell to make it stop on the fourth day, and it did (7.191-192). Similar legendary motifs attend descriptions of the battles and other events of the war: Xerxes “leaped thrice from his throne in fear of his army” (7.212); Scyllias deserted to the Greeks by swimming ten miles under water (8.8–Herodotus doubts it, but the story is still proof that many believed it); the temple of Delphi magically defended itself with animated armaments, lightning bolts, and collapsing cliffs (8.37-38); the sacred snake on the Acropolis would not eat its honeycake, confirming that the Athenians should desert the city (8.41); the sacred olive tree which had been burned up by the Persians grew a new shoot an arm’s length in a single day (8.55); a disembodied chant was heard in the holy city of Eleusis, then a dust cloud spontaneously arose from there and drifted toward Greece signaling that they would win the war, in accord with the prophecy of Dikaeus (8.65); a charming but possibly invented account of Artemisia’s incredible good luck in a naval battle (8.88–the punchline gives the story away as a possible myth); a story about Xerxes on his retreat that Herodotus doubts, but was clearly believed by others (8.118-120); a miraculous flood tide wiped out a Persian contingent that had desecrated an image of Poseidon (8.129); a morality tale about Persian decadence vs. Greek frugality may be apocryphal (9.82); and, finally, after the war, there was a mass resurrection of cooked fish (9.120).
William Lane Craig wants you to believe that, based on Herodotus, “there simply was insufficient time for significant accrual of legend by the time of the gospels’ composition.” Oh, really? You decide.
[1] The argument described here, and all quotations, originally appeared in “Contemporary Scholarship and the Historical Evidence for the Resurrection of Jesus Christ,” Truth: an International Inter-Disciplinary Journal of Christian Thought 1 (1985), pp. 89-95, and are here taken from the version of this article provided on the web at http://www.leaderu.com/truth/1truth22.html.
[2] In regards another defense of Craig’s position in In Defense of Miracles, see my review of that work. In regards the entire question of the resurrection of Jesus, see my comprehensive essay on that subject.
[3] I discuss the entire subject of miracles in great detail in my review of In Defense of Miracles. I also remark on the subject in one section of my comprehensive essay on the resurrection.
[4] I discuss proper historical method in great detail in another part of my review of In Defense of Miracles, and comment on the general failings of Christian apologists in doing history in the conclusion of my summary of that review.
[5] Craig cites the source as Roman Law and Roman Society tn the New Testament [sic]. He must have forgot to double check his reference. The misspelled “in” is clearly just a type-o, but the real title of Sherwin-White’s book is Roman Society and Roman Law in the New Testament, 1963.
[6] loc. cit., “The Historicity of the Gospels and Graeco-Roman Historiography,” p. 189.
[7] A fact little-understood by non-historians is the usefulness of texts such as the NT for details of social history. In this respect, they are invaluable even if they are completely false, because any fiction must necessarily reflect the social realities or beliefs of the author and his audience. In this respect, even the most overtly fictional texts from antiquity are historically valuable.
[8] ibid., pp. 189-90.
[9] “As for the stories told by the Egyptians, let whoever finds them credible use them. Throughout the entire History it is my underlying principle that it is what various people have said to me, and what I have heard, that I must write down,” 2.123; see also 1.5, 4.195, and for his giving of different accounts, see, e.g., 1.3-5, 2.20-27, 5.86-87, 6.53-54, 7.148-152, for naming his sources, see, e.g., 1.20-21, 2.29, 4.14, 4.29, 5.86-87, 6.53-54, 8.55, 8.65, and for expressions of healthy skepticism, see, e.g., 2.45, 3.16, 4.25, 4.31, 4.42, 4.95-96, 4.105, 5.86, 7.152. Despite all this, Herodotus is notoriously called, today as in ancient times, the “Father of Lies” (in mockery of his more respectable title as the father of history), since in the final analysis he is not very reliable, and invents a great deal in order to create symmetries and allusions and illustrations of his moral beliefs, or passes on as fact a lot of bogus, sometimes absurd information. In this respect, the Histories does indeed serve as a good analogy for the Gospels, although the Gospels are worse, lacking even what Herodotus has in the way of critical thinking evident in his storytelling.
[10] The decisive starting points for sources of the Persian War are C. Hignett, Xerxes’ Invasion of Greece (1963); A.R. Burn, Persia and the Greeks, 2nd ed. (1985); J.F. Lazenby, The Defence of Greece (1993); also, the Cambridge Ancient History, 2nd ed., 5.2 (1992).
[11] Sherwin-White is being a little overly rhetorical here, and Craig would have noticed this if he had actually bothered to read the passage in Herodotus that he draws his example from. For the reasoning of Herodotus is entirely subjective, not based on any falsifying evidence, but only his analysis of the interests of certain parties: “I do not accept the story” he says (6.121), but the story itself is true, he insists, only “who it was that was the agent I cannot say” (6.124). This is not a legend Herodotus is doubting, but the truth behind a political smear campaign. It is worth noting that if Herodotus had heard the Gospel stories, he would have been equally doubtful, for the same subjective reasons, if we can judge from his reception of an account of a Thracian resurrection religion that was similar to Christianity in important respects (in having a man resurrected as a god in the flesh–after three years, rather than three days–granting immortality for believers, 4.94-6; Plato ascribed healing magic to him in the Charmides 156d-158b). Since Herodotus presumes this man to have been a slave of Pythagoras, the date of this story would be sometime around 500-530 BC, only fifty to one hundred years before Herodotus recorded the details.
[12] The Penguin History of Greece (1990) p. 170.
[13] In fact, these oracles were very likely political inventions by military leaders (such as Themistocles) during the war itself, making the time of their invention almost immediate, yet no one challenged them. There are many other fulfilled prophecies reported by Herodotus (cf. 8.96, 9.43).