Legendary Development

"Legends do not courteously wait to arrive until their protagonists are long dead and gone." — Dale Allison, 2021¹

"The point of my argument is not to suggest the literal accuracy of ancient sources, secular or ecclesiastical, but to offset the extreme scepticism with which the New Testament narratives are treated in some quarters." — A. N. Sherwin-White, 1963²

Summary

The theory of legendary development says the resurrection stories grew in the telling:  early claims that Jesus had been "seen" became, over decades, scenes of an empty tomb, angels, guards, and a risen Jesus who talks, is touched and eats.  The question here is narrower than "did it happen?"  It is how much of the tradition is attested early, how much of the later narration may preserve earlier memory or may have grown, and what evidence can tell those apart.  Hallucination Theories of the Resurrection and The Empty Tomb weigh what caused the claims.

1. The earliest evidence attests claims, not scenes, and some of the claimants. Paul's letter of the mid-50s, his teaching at Corinth about AD 50, and the older formula he quotes (1 Corinthians 15:3b-5) show that within a few years of the crucifixion Jesus' followers said he had died, been buried, been raised on the third day, and "appeared to Cephas, then to the twelve." Gerd Lüdemann, who explained the appearances as visions, dated the formation of these appearance traditions to AD 30-33. Paul says he stayed fifteen days with Cephas and met James about three years after his own conversion, and that he and they preached the same thing. None of these sources describes what anyone saw or heard. So the early date carries the claims back, and with them the claimants Cephas and James, but no Gospel scene.
2. The appearance tradition is early on the transmitted text; early growth remains possible. Paul already claims to have seen Jesus in 1 Corinthians 9:1. Robert Price argues for a later insertion of the list in 15:3-11, not necessarily a late origin for every appearance claim. His proposal lacks direct manuscript support but must be assessed on its internal arguments, rather than dismissed from that absence alone. The early list, read as Paul's, constrains late-invention theories. It does not exclude early growth, including changes within the claimants' own retelling.
3. Where Mark survives, the later Gospels visibly rework it, but the labels differ by example. Assuming that Matthew and Luke used Mark, their changes can be seen: Matthew adds a guard, an earthquake, a descending angel and an appearance to the women, and turns Mark's silent women into women who run to tell; Luke turns the promise that Jesus "goes before you into Galilee" into a reminder of what he said "when he was still in Galilee," and narrates no Galilee appearance. What is visible is rewording, addition and omission. Whether an added item came from another source, was the writer's framing, or was invented is a further question, settled differently for each item. Matthew's guard and descending angel are probably secondary; the women's meeting with Jesus and Luke's Jerusalem appearances are undecided.
4. The proposed physical trajectory partly fits. Richard Carrier states it plainly: from "revelatory" experiences in Paul, to a vanished body in Mark, to a "vaguely physical" encounter in Matthew, a "very physical" one in Luke and an "incredible physical encounter" in John. Physical demonstration does increase from Mark through Matthew to Luke, and Luke and John have explicit invitations to touch and displays of hands or wounds. Luke also narrates eating as a bodily proof; John has Jesus provide a meal without saying he eats. These scenes are framed as answers to doubt. Two parts of the claim are weaker. Whether Paul's "spiritual body" was non-physical is disputed, and John is not clearly more physical than Luke. The features often cited against the trajectory (vanishing, not being recognised, locked doors) do not count against it, because a transformed body can do all of these.
5. Second-century texts continue the same lines, but they are weaker controls than they look. Ignatius (writing under Trajan, by tradition about 108) has the disciples "immediately touched him and believed," and the Epistle of the Apostles (most often placed in the mid-second century) adds a footprint test. But Ignatius' saying is not Luke's wording, and Origen and Jerome each assign it to another book. The Gospel of Peter narrates the rising itself. Most current scholars date it to the second century and think it used the canonical Gospels; John Dominic Crossan argued that an early source inside it underlies all four. Its only substantial copy is a late-antique manuscript, and its identity with the book Bishop Serapion read about 200 is a probable assumption.
6. Legend can attach fast, cores can survive, and neither fact settles a particular claim. A. N. Sherwin-White argued from Herodotus that two generations were "too short a span to allow the mythical tendency to prevail over the hard historic core." He did not claim that legend needs two generations to appear; in a footnote he reported a colleague's point that myth about Alexander grew "within the lifetime of contemporaries," and judged it "encouraging rather than the reverse," because the core survived. Dated ancient cases show marvel claims appearing within days (Lucian on Peregrinus, AD 165) or about a decade (Josephus' Temple portents), or dated by later writers to the moment of the event (the sworn ascent of Augustus). In each, the public event survived. The best-documented later case fits the same pattern. Legendary miracle reports about Sabbatai Sevi appeared in letters within months in 1665, and a few years after the first reports of his death in 1676 believers said his burial cave had been found empty on the third day; his conversion to Islam and his death were never erased. The nearest ancient parallel to the appearances, Apollonius of Tyana's appearance to one unnamed youth after his death, is first attested about 120 years after it. These cases show what can happen and how fast; most of them cannot show whether the marvel was false, and none measures how often.
7. What the evidence can decide is limited, and it cuts both ways. It supports an early claim that named leaders and groups saw Jesus, and probably an early story of women at the tomb. It shows probably secondary features in Matthew's guard story and his descending angel. For the conversation, touching and eating scenes, the evidence that would tell memory from growth (an early independent witness, an author's stated source, a dated control from the same line of tradition) is missing. Their late first attestation gives more room for change, which is a real point against relying on them, but it does not establish that they were invented.

Verdict.  The stories demonstrably changed in the telling where a source survives, and kept changing in the second century.  On the transmitted Pauline text, the appearance claims already exist in the mid-50s, with an older core; an origin only decades later cannot explain that evidence.  Legend in the sense of early growth around early claims is not ruled out, and fits the evidence for several details.  The popular picture of a simple progression from Mark to John fits the narrow sequence of physical proof from Mark to Luke better than anything else, and fits John and Paul less well.  The specific scenes on which the group-hallucination debate turns cannot be shown to preserve the first reports, and cannot be shown to be inventions.  Sherwin-White's argument supports the survival of a core, which here includes the claim that Cephas and others saw Jesus.  It does not certify the details, and it does not certify the claim itself.

Part 1: The models, in their proponents' words

"Legend" is used loosely in this debate, so the models need to be kept apart.  Four questions separate them:  when the claims began, what started them, how fast the stories changed, and which details are supposed to be late.

A proposed later origin for the list

Robert Price argues that the list in 1 Corinthians 15 is not Paul's at all.  "In the present article I will be arguing that this pericope presents us instead with a piece of later, post-Pauline Christianity," he writes, and concludes that "The most probable solution, however, is simply that vv. 3-11 constitute an interpolation."³  On his reading the appearance to James belonged to the list, which as a whole is post-Pauline, "while the note about the 500 is later still, an interpolation redolent of much later legendary extravagance."³  This challenges the date of this particular list.  It does not by itself mean that every claim to have seen Jesus began after Paul.

Price concedes that "There are no extant copies of 1 Corinthians which lack my passage," but answers that "There simply are no texts at all for the period in which I suggest the interpolation occurred."³  Agreement among the surviving manuscripts therefore counts against his proposal without logically excluding an insertion before their common ancestry.  His internal case includes the apparent tension between receiving this gospel and Paul's insistence in Galatians that his gospel came by revelation, and his judgment that verse 12 follows verse 2 more naturally without verses 3-11.  Neither argument uniquely establishes interpolation: a received formula can differ in style from its surrounding argument, and Paul's revelatory commission need not exclude his learning information from other people.  Price's proposal is not the usual skeptical position.  Richard Carrier, who thinks Mark invented the empty tomb, answers that "arguments that the whole creed was interpolated (e.g. Price makes the best attempt in The Empty Tomb, pp. 69-104) are just built on possibiliter fallacies."  His reason comes from the text itself:  the awkward fit of verse 7 "actually argues for the creed not being an interpolation, and for verse 7 being an interpolation, since if the whole thing were, it would be coherent."⁴  Carrier suspects a smaller insertion at verse 7 instead.⁴  One critic's judgment against another's does not settle the question, but it shows that the early date of the list does not depend on a Christian reading.  The following discussion uses the text as transmitted, while distinguishing what would remain if Price's proposal were granted.

Growth from early experiences

Dale Allison, who believes Jesus appeared to his followers, argues that the canonical stories nevertheless contain first-century inventions.  On Matthew's guard, earthquake and sealed tomb he writes:  "One can understand someone adding, for theological and apologetical ends, the guard (Mt. 27:62-66; 28:4, 11-15), the sealing of the tomb (28:66), and an earthquake (28:2). But how do we explain someone subtracting those things, which are also missing from Luke and John? I am unable to conjure a satisfactory motive."¹  His conclusion is that "Mark's far simpler account of Jesus' burial and resurrection commends itself as being earlier," while Matthew's "much more elaborate and apologetically oriented narrative ... impresses one as later, as full of secondary developments, as indeed being on its way to the Gospel of Peter."¹  He does not think speed is a defence:  "Legends do not courteously wait to arrive until their protagonists are long dead and gone."¹  Yet Allison also holds that the appearance stories, "despite their myriad disagreements with each other and their late and legendary features," may be "reminiscent in certain particulars of the original experiences, although delineating those particulars is an uncertain business."¹  His model is early experiences plus growth, not growth from nothing.

Bart Ehrman, an agnostic, puts the general point bluntly:  "Of course, the Gospels contain nonhistorical information and stories that have been modified and exaggerated and embellished."⁵  He describes the mechanism as ordinary retelling:  "Storytellers were apt to add details to stories that were vague, or to give names to people otherwise left nameless in a tradition."⁵  He explains the proof scenes as answers to doubt:  this "would explain why there is such a strong doubt tradition in the Gospels, and why there is such an emphasis (in Luke, John, and especially Acts) on the fact that Jesus had to 'prove' that he was raised."⁵  He also allows growth inside Paul's own list:  "I am not saying that Paul necessarily made up the story of the five hundred himself; he may well have inherited it from an oral tradition. ... Sometimes stories just get exaggerated or invented."⁵  His reconstruction keeps visions by Peter, Paul and Mary at the start;  that part is weighed in Hallucination Theories of the Resurrection.

Gerd Lüdemann makes the physical scenes the clearest case of late growth:  "Those reports within the Gospel narratives that stress the bodily nature of the risen Christ are presumably of later origin. Their emphasis on the reality of the resurrected body of Jesus shows signs of secondary apologetic concern in the face of the docetic challenge, according to which Jesus only seemed to have risen."⁶  Docetism was the belief that Christ only seemed to have a real body.  On Luke's fish-eating scene he writes that Luke's "'crude materialism' in this story is no mere derivation from naive popular belief but has the explicit aim of repressing any possible magical or demonological interpretation."⁶

Richard Carrier states the strongest form of the growth model, and he is the clearest proponent of a progression across the Gospels.  He argues that the empty tomb is a symbolic story that Mark created "even if with entirely righteous intentions," and that "Just as Matthew could invent guards, an earthquake, and a descending angel, so could Mark invent an empty tomb."⁷  The tomb claim is weighed in The Empty Tomb.  On the appearances he sees a sequence:  "from nothing more than 'revelatory' experiences in Paul, to a vanished body in Mark, to a vaguely physical encounter with Jesus in Matthew, to a very physical encounter in Luke, all the way to an incredible physical encounter in John (and if we go beyond the canon, the next stage is reflected in the Gospel of Peter: actually witnessing Jesus rise from the grave)."⁷  His core is not empty:  "a vision of some mysterious kind inspires or informs someone (perhaps Peter or Mary) with the basic outline of the Gospel."⁷  Part 5 tests his sequence.

The defence of early narrative tradition

N. T. Wright argues that the narratives have four features a later legend would not have produced:  "virtually no embroidery from the biblical tradition," no mention of the believer's own future resurrection, a Jesus who does not shine "like a star" yet passes through doors and is not recognised, and women as the first witnesses.⁸  On the physical scenes in particular he writes:  "It seems to me totally incredible that stories like these, especially those in Luke and John, represent a late development of the tradition in which for the first time people thought it appropriate or necessary to speak of the risen Jesus being solidly embodied."⁸  He grants that "The more the stories go on, particularly of course in the fuller accounts of Luke and John, the more they reflect the style, and the theological interest, of the particular evangelist."⁸

William Lane Craig argues from time:  "One of the major problems with the legend hypothesis, however, which is almost never addressed by skeptical critics, is that the time gap between Jesus' death and the writing of the gospels is just too short for this to have happened."⁹  The claim he defends is about the core.  His first point is that "there was insufficient time for legendary influences to erase the core historical facts," and he concludes, "In order for the gospels to be legendary at their core, more generations would be needed between the events they record and the date of their composition."⁹  A separate argument of his concerns details:  eyewitnesses and the apostles were "a natural check on tendencies to elaborate the facts in a direction contrary to the one preserved by those who had known Jesus."⁹  His authority on time is A. N. Sherwin-White, examined in Part 7.

Michael Licona explains many differences as compositional conventions also found in Plutarch, such as compression:  "That Luke compressed the events in this manner is clear, since in the sequel to his Gospel, Luke says Jesus appeared to his disciples over a period of forty days before ascending to heaven (Acts 1:3-9)."¹⁰  He argues from Paul that "none of them invented the core story."¹⁰

Distinctions the models need

1. Three clocks. The date of an event, the date a later writer assigns to the first telling of a story about it, and the date of the earliest surviving witness are different things. Earliest attestation is not first origin. A detail first written down in the 80s may be older, and a detail written down early may still be invented.
2. Kinds of change. Where a later writer's source survives, five different things can be going on, and they need different evidence: (a) rewording and rearranging what the source says; (b) adding material from another source; (c) framing the story to make a theological or apologetic point; (d) inaccurate growth, where a remembered event gains wrong details in retelling; (e) invention of an episode. Comparing texts shows (a), and shows that something was added. Telling (b) from (d) or (e) needs further evidence. Labelling every addition "legend" uses evidence for the weakest kind of change to claim the strongest. Refusing ever to infer invention ignores the evidence that sometimes exists.
3. A surviving core is not an accurate detail. Sherwin-White, Craig and Licona argue for a core. The disputed scenes are details.
4. An early claim is not an explained claim. An early date limits the time for growth. It does not show what produced the claim: a real meeting, a vision, a mistake or an invention by the first claimants are all early.
5. Legend is not fraud. Allison, Ehrman and Carrier propose growth in good faith: retelling, interpretation, apologetic answers to doubt, symbolic composition. None of them proposes a conspiracy. Wright's argument, that a later inventor "would" have added scripture or glory, is aimed at deliberate composition and has less force against slow unplanned drift.
6. Convention is not error, and not memory either. Licona's compression is a literary choice, not a mistake. But a compressed or rearranged scene is not evidence for its own details.

Part 2: What is early, and what the early sources do not say

Paul's four anchors

On the transmitted text, Paul's report in 1 Corinthians 15 supplies the earliest securely dated account of the appearance list.  The following anchors assume that the passage is Paul's; Price challenges that assumption.  The anchors should not be merged.  The sibling articles Paul's Letters as Historical Evidence and The Resurrection Belief: Event or Process? give the evidence for each date.

1. The letter itself, in the mid-50s. Everything in 1 Corinthians 15:3-8 is securely attested by this date, including the five hundred, James, "all the apostles" and Paul.
2. Paul's teaching at Corinth, about AD 50. Paul says he "delivered" this to the Corinthians when he founded the church.¹¹
3. The older formula, 15:3b-5. The wording "that Christ died ... that he was buried, that he was raised on the third day ... and that he appeared to Cephas, then to the twelve" is strongly inferred to be older than Paul's letter, from its un-Pauline wording and the Aramaic name Cephas. Whether every item in verses 6-8 belonged to it is not known; Allison's own surmise is that the older formula "probably ended with v. 5."¹
4. Proclamation before Paul's conversion. Paul persecuted people who already proclaimed Jesus, so the proclamation of a raised Jesus probably existed one to six years after the crucifixion. This dates a claim. It does not date the formula's wording, and it does not show which appearances that early proclamation named.

Scholars who reject the resurrection put the formula's material at the very start.  Gerd Lüdemann, who explained the appearances as visions, wrote:  "We can assume that all the elements in the tradition are to be dated to the first two years after the crucifixion of Jesus."  His conclusion was that "the formation of the appearance traditions mentioned in I Cor.15.3—8 falls into the time between 30 and 33 CE, because the appearance to Paul is the last of the appearances and cannot be dated after 33 CE."¹²  Carrier calls verses 3-5 "almost certainly a pre-Pauline text composed within a few years of when Jesus was believed to have died."⁴  Lüdemann's reason dates the experiences more firmly than the wording.  If Paul's experience came last and no later than 33, the others' came earlier, but the list that joins them could still have been put together some years later.  Paul's Letters as Historical Evidence sets out that distinction.  For this article the point is narrower:  Lüdemann and Carrier, who both argue that the Gospel scenes grew, do not argue that the appearance claims themselves came late.

What the four anchors share is also what they lack.  The Greek verb for the appearances, ὤφθη, "he was seen" or "he appeared," is repeated four times for Cephas, the five hundred, James and Paul, with εἶτα or ἔπειτα ("then") linking the further groups.¹³  Nothing is said of where, how long, what was said, or whether anyone touched anything.  The list names no women and does not mention the tomb, though "he was buried" comes before "he was raised."

What Paul's contacts constrain, and what they do not

Evidence outside the list adds something the formula alone does not.  Galatians says that about three years after his conversion Paul went to Jerusalem "to visit Peter, and stayed with him fifteen days," and saw "James, the Lord's brother"; on a later visit he met "James and Cephas and John, those who were reputed to be pillars."¹¹  In 1 Corinthians 9:1 he asks, "Haven't I seen Jesus Christ, our Lord?"¹¹  These passages fall outside Price's proposed insertion.  The agreement in preaching stated in 15:11, "Whether then it is I or they, so we preach, and so you believed,"¹¹ falls inside the verses Price challenges.  It supports the ordinary reading of the passage but cannot independently refute his proposal.

1. What they constrain. Taken with the list, Paul's contacts make it plausible that Cephas and James themselves claimed appearances, and that Paul knew their claims at first hand. Galatians does not report their conversations or say when the wording of an appearance claim became fixed. If the list were removed, the contacts alone would not recover its specific claims. Paul's own claim to have seen Jesus in 9:1 would remain, so a theory that every appearance claim began after Paul would still face early evidence.
2. What they do not constrain. They do not show what Cephas or James experienced. They do not show that Paul met any of the five hundred, so that item is secure as Paul's claim by the mid-50s but not tied to a person he names; Ehrman's suggestion that it was inherited and exaggerated, and Carrier's that it is a version of Luke's Pentecost,⁷ are both early-growth proposals that this evidence cannot exclude. Lüdemann, who also rejected the resurrection, judged the opposite: "the appearance to more than 500 brethren was told for no other reason than that it had really happened."¹² By that he meant a real event of another kind, "the report of a mass ecstasy that took place in the early period of the community."⁶ Critics who agree that the visions came first therefore disagree about whether this item grew. And the contacts do not reach any Gospel scene.
3. What follows. A growth model needs to explain the early material it starts from. Paul's contacts constrain a proposal that people much later simply attached appearance stories to these men's names. They do not exclude development within the claimants' own accounts. If Cephas or James gradually reinterpreted an experience, or changed how he described it, the surviving itinerary would not tell when that happened. A full explanation must address both the initial experience or belief and any subsequent retelling; the categories can overlap. Whether a vision, a real meeting, a mistake or some other event began that process is the subject of the sibling articles.

So the early sources establish an early claim that Jesus appeared to named leaders and to groups, made in part by the named leaders.  An early list does not attest the later touching, eating and conversation scenes.  It also does not rule out rapid narrative growth around the claim in the same years.

The Gospels: order, not a calendar

The Gospels give narratives, and their dates are softer.  When Were the Gospels Written? gives the working estimates:  Mark between the 40s and the early 70s, Matthew in the 70s or 80s, Luke in the early 60s or the 80s-90s, and John in the 80s or 90s, each on different evidence.  Four cautions from that article matter here.

1. Markan priority gives order, not years. If Matthew and Luke used Mark, their resurrection chapters are later than Mark's, but this does not say by how much.
2. Paul is earlier than Mark on most dates, not all. Paul's letter is securely dated to the mid-50s. On most proposed dates Mark is later. On the earliest date in that range, the 40s, Mark would be earlier than the letter, though probably not earlier than the formula Paul received. So "Paul before Mark" is a statement about the probable dates, not a fixed point.
3. John is dated on separate grounds, and its independence is a question for each passage. Mark Goodacre's 2025 book argues, in its publisher's summary, that John "knew and used the narratives of Matthew, Mark, and Luke";¹⁴ its detailed argument was not read for this article. Allison calls the independence or partial independence of Luke 24:1-12 and John 20 from Mark "a live option."¹ Two texts can share material without one copying the other, through a common oral or written source, and an author who knew a Gospel can also use other material. So "independent" has to be decided episode by episode, not for John as a whole. John 21 supplies no fixed year.
4. Luke comes before Acts. A firm latest date for Acts bounds Luke, but a late earliest date for Acts does not make Luke equally late.

Mark 16:8, read exactly

The earliest recoverable Mark ends at 16:8.  Bruce Metzger's textual commentary records that "The last twelve verses of the commonly received text of Mark are absent from the two oldest Greek manuscripts (א and B)," that "Clement of Alexandria and Origen show no knowledge of the existence of these verses," and that "Eusebius and Jerome attest that the passage was absent from almost all Greek copies of Mark known to them."¹⁵  He judges that "the earliest ascertainable form of the Gospel of Mark ended with 16.8," and that the longer ending "was excerpted from another document, dating perhaps from the first half of the second century."¹⁵  Irenaeus quoted Mark 16:19 about 180, in a book written while Eleutherus was bishop of Rome.¹⁶¹⁷  Whether Mark meant to stop at 16:8 is unknown;  Metzger lists three possibilities and thinks a lost last leaf most probable.¹⁵

What the verse says is narrow.  The women "went out, and fled from the tomb, for trembling and astonishment had come on them. They said nothing to anyone; for they were afraid."¹¹  That is narration about the women.  It is not a statement that no one ever saw Jesus.  The same chapter has the young man say "He goes before you into Galilee. There you will see him, as he said to you," pointing back to Jesus' prediction at 14:28.¹¹  So Mark's story twice promises a meeting in Galilee, and presents the promise as Jesus' own.  That shows Mark expected his readers to take the promise as reliable.  It does not show that Mark had a narrative of the meeting, or what his readers already knew about it.  Three cautions follow.

1. A missing scene is not a denial. Mark narrating no appearance does not show that Mark's community knew none.
2. A disagreement is not automatic corroboration. When Matthew has the women run to tell the disciples and Mark has them say nothing, the difference does not make them two independent witnesses; Matthew probably rewrote Mark.
3. A possible harmony is not a demonstrated one. Licona suggests that "said nothing to anyone" means they spoke to no one on the way, comparing Mark 1:44.¹⁰ That is possible. It is not evidence that the women did tell the disciples at once.

The second-century additions show what later readers felt was missing.  The longer ending summarises an appearance to Mary Magdalene, an appearance "in another form" to two walking into the country, a rebuke of the eleven "for their unbelief and hardness of heart," a commission, promised signs, and the ascension.¹¹  Each element has a counterpart in Luke, John or Acts.  This is a clear case of a later writer supplying narrative to fit a text that seemed to lack it.

Part 3: The details, side by side

The table sets the reported details in rough order of writing.  The order of the middle columns rests on Markan priority and the working dates above;  Paul's column is the only one with a secure early date, and the last two are second-century.  A dash means the source does not mention the item.  Verse numbers are given so each cell can be checked.

Sources for the table:  the World English Bible and SBL Greek New Testament for the canonical texts,¹³¹¹ and M. R. James's translation of the Gospel of Peter.¹⁸  The "recipients or scenes" row mixes two units on purpose and says so.  Paul lists recipients:  Cephas, the twelve, the five hundred, James, all the apostles and Paul.  The Gospels narrate occasions:  Matthew's two are the women and the eleven;  Luke's three are Emmaus, Simon (reported in one line) and the eleven;  John's four are Mary, the disciples, the disciples with Thomas, and the seven at the sea.  A list of names and a set of narrated scenes are not the same kind of thing, so the numbers cannot be lined up to test anything.  No author cited in this article predicts that the number of appearances should rise steadily with date, and the table is not offered as a refutation of one.

Five things stand out.

1. No column is simply an expanded copy of the one before. Matthew adds a guard and a touch; Luke narrates no Galilee appearance and adds eating; John adds Mary's individual meeting and Thomas, and keeps a Galilee scene in an added final chapter.
2. The canonical Gospels never narrate the rising itself. Only the Gospel of Peter does.
3. Physical demonstration rises from Mark to Luke. Mark has no appearance; Matthew has the women grasp Jesus' feet as they worship; Luke has Jesus invite touch, show his hands and feet, and eat to prove he is not a spirit. John has the most invasive invitation (Thomas' hand in the side) but no explicit eating and no statement about flesh and bones.
4. Touch does not begin with John or Luke. The first narrated contact is Matthew's women at 28:9: "They came and took hold of his feet, and worshiped him."¹¹ The Greek ἐκράτησαν αὐτοῦ τοὺς πόδας means they grasped his feet.¹³ This fits Carrier's own ordering, which already calls Matthew "vaguely physical"; it limits only a model in which touch first appears in the latest Gospel.
5. Strangeness travels with solidity. The texts that make Jesus most physical are the same texts in which he is not recognised, vanishes, comes in while the doors are locked, and ascends. Part 5 asks what this does and does not show.

Some details have a length that can be measured.  Counting Greek words from the women's visit to the end of each account, Mark 16:1-8 has 136, Matthew 28 has 329, Luke 24 has 809, and John 20 and 21 together have 1,163.¹⁹  The boundaries are not identical:  Matthew's count leaves out his guard scene in 27:62-66 but includes the guards' report in 28:11-15;  Luke's includes the ascension;  John's includes the closing notes of 20:30-31 and 21:24-25.  What the numbers measure is length.  Longer narratives are what one would expect if stories grew.  They are also what one would expect if later authors had more sources, more space or a different purpose:  Luke and John wrote longer Gospels than Mark.  Length alone does not distinguish the two, so it is not used as evidence here.

Part 4: Where the source survives: what the editing shows

Most theories of legendary development rest on inference about lost oral stages.  One part of the evidence does not:  if Matthew and Luke used Mark, their resurrection chapters show what two first-century writers did with a source we still have.  The Synoptic Problem gives the case for that dependence;  most scholars accept it.  Wright is more cautious here, writing that "Because we cannot tell what literary relationship, if any, there was between the resurrection narratives, it is impossible to say whether, and if so to what extent, any of the gospel writers have consciously modified their sources in the interests of their own theology or other agendas."⁸  That caution is unusual.  Elsewhere in the Gospels Matthew and Luke follow Mark's wording closely for long stretches, and the same writers are unlikely to have worked differently in their last chapter.  The comparisons below assume Markan priority, and they lose their direction for anyone who rejects it.

What a comparison shows is limited, and it matters to say so for each example.  It shows that the later writer reworded, added or left out.  It does not show by itself where an addition came from.  Each change below gets a label from the list in Part 1 and a short statement of what supports it.

What Matthew did

1. Added a guard story. Matthew alone has the chief priests ask Pilate to secure the tomb "lest perhaps his disciples come at night and steal him away," a guard who "shook, and became like dead men," and the bribe that explains why "This saying was spread abroad among the Jews, and continues until today."¹¹ The story answers a specific accusation that Matthew says was current in his day. Label: an addition with an apologetic frame, probably secondary. Basis: it is absent from Mark, Luke and John; Allison's point that adding it is easy to explain and subtracting it is not (Part 1); and the same motif grows in the Gospel of Peter. The Empty Tomb judges the guard probably not historical, while the rumour of theft was probably real by Matthew's day.
2. Made the scene more dramatic. Mark's stone is already rolled back. In Matthew "there was a great earthquake, for an angel of the Lord descended from the sky and came and rolled away the stone from the door and sat on it. His appearance was like lightning, and his clothing white as snow."¹¹ Label: an addition, probably Matthew's elaboration. Basis: Mark's young man is already sitting inside; Luke and John, whatever their relation to Matthew, have no descent or earthquake; and Matthew uses an earthquake at the death of Jesus too (27:51-54). This is a judgment about Matthew's telling, not a proof that nothing unusual happened.
3. Reversed the women's silence. Mark: "They said nothing to anyone; for they were afraid." Matthew: "They departed quickly from the tomb with fear and great joy, and ran to bring his disciples word."¹¹ Matthew keeps Mark's fear and adds joy and obedience. Label: rewording of Mark's ending. Basis: the verbal overlap. Whether Matthew read Mark's silence as temporary (Licona's reading of Mark) or changed it is not decidable from the text.
4. Added two appearances. Jesus meets the women, who take hold of his feet, and his words to them largely repeat the angel's: "Go tell my brothers that they should go into Galilee, and there they will see me."¹¹ Then the eleven see him on a mountain in Galilee, "but some doubted," and receive the commission with the trinitarian formula.¹¹ Label: additions whose source is undecided. The meeting with the women has a partial parallel in John 20, discussed below. The mountain scene fulfils Mark's promise. Its wording is Matthew's: Allison notes the trinitarian formula is unattested before Matthew.¹ "Some doubted" can be read two ways. Defenders count it as an odd detail for an apologist to invent at his climax. Carrier reads it as a rhetorical device that brands doubters, with Moses at Sinai in view.⁷ Either way, a writer with a persuasive purpose can include a difficulty; the detail alone does not show that Matthew was or was not shaping the scene.

The first two changes have no source outside Matthew.  The fourth is different.  John 20 also has Mary Magdalene meet Jesus near the tomb and be sent to "my brothers."¹¹  The two scenes share a setting, a recipient and a message to "brothers," and differ in nearly everything else:  Matthew has two women who grasp Jesus' feet and worship;  John has Mary alone, who does not recognise him until he speaks her name.  Three explanations fit:  John developed Matthew's scene;  both drew on an earlier report of an appearance to Mary;  or both drew on a common oral tradition and shaped it differently.  Goodacre's case that John knew the Synoptics favours the first, but even an author who knew Matthew could have had other material for this scene.  So the weight of the appearance to the women depends on this episode, not on a verdict about John as a whole, and on present evidence it is undecided.  They are not one bare individual report, and neither scene is evidence for the other's details.

What Luke did

1. Turned a promise about Galilee into a reminder. Mark's young man says, "He goes before you into Galilee. There you will see him" (προάγει ὑμᾶς εἰς τὴν Γαλιλαίαν). Luke's two men say, "Remember what he told you when he was still in Galilee" (ἔτι ὢν ἐν τῇ Γαλιλαίᾳ).¹³¹¹ What follows in Luke is the passion prediction, "that the Son of Man must be delivered up into the hands of sinful men and be crucified, and the third day rise again," which Luke had placed in Galilee (9:22, 9:44).¹¹ Luke also has no counterpart to Mark 14:28; Galilee is not mentioned in Luke 22. Label: rewording of the angel's saying. Basis: the verbal parallel, under Markan priority. Licona allows two explanations, that "Luke 24:5-9 either paraphrases Mark more freely than Matthew or has drawn from a different source," and adds that since Luke compresses everything into one day, "there is no time to have the disciples go to Galilee. Thus, Luke may have redacted the message of the angel."¹⁰
2. Narrated only appearances in and near Jerusalem before the ascension. Luke then has Jesus tell the disciples to "wait in the city of Jerusalem," and Acts repeats it: "Don't depart from Jerusalem."¹¹ The appearances Luke narrates before the ascension are on the road to Emmaus and in Jerusalem; Jesus then leads the disciples out to Bethany, and Acts places the ascension on the Mount of Olives. This is not true of Luke-Acts as a whole: Acts later narrates the risen Jesus appearing to Paul near Damascus and speaking in visions at Corinth and Jerusalem.¹¹ Acts also says Jesus "was seen for many days by those who came up with him from Galilee to Jerusalem," which names Galilee as where the witnesses came from, not where they saw him.¹¹ So Luke omits Galilee appearances; he does not say there were none. A Galilee appearance during the forty days of Acts 1:3, before the command of Acts 1:4, is not excluded by Acts, though Luke 24 read alone has no room for one. What can be said is that Luke selected and framed his material to fit a plan that begins in Jerusalem and moves outward. Whether he also contradicts what happened depends on what happened in Galilee, which Luke's text cannot settle.
3. Compressed the time. Luke 24 reads as one day, from dawn to the ascension, while Acts 1:3 has "a period of forty days."¹¹ Licona takes this as deliberate compression by the same author, and that reading is strong, since one writer produced both.¹⁰ Label: a literary convention, not an error, and also not evidence for any scene's timing.
4. Added proof. Luke's eleven think they see "a spirit"; Jesus answers, "Touch me and see, for a spirit doesn't have flesh and bones, as you see that I have," and eats broiled fish "in front of them."¹¹ Label: an addition whose source is undecided, framed as an answer to doubt. Part 5 weighs it.

What the edits do and do not show

1. They show that Matthew and Luke treated Mark's resurrection chapter as material to be reshaped: to answer an accusation, to fit a geographical plan, to heighten the scene, and to resolve Mark's open ending. This does not depend on speculation about oral stages. Defenders such as Licona describe the same moves as accepted compositional practice in ancient biography; Ancient Historiography and Compositional Convention examines that claim.
2. They do not show that every addition is invented. Matthew and Luke had sources besides Mark, and an added scene can come from one of them. For the guard and the descending angel there is positive evidence of secondary growth. For the appearance to the women and Luke's Jerusalem appearances there is not, and the honest grade is undecided.
3. They do show that late first attestation is not neutral. A detail found only in a later Gospel, when its Markan source lacks it, has had more opportunity to change, and its first witness is a writer seen reshaping his source elsewhere in the same chapter. That is a real reason for caution, not a proof of invention.
4. They do not show steady growth. Luke narrates no Galilee appearance, which Mark promised and Matthew narrated.

Part 5: Did the risen Jesus become more physical?

The strongest specific claim of the growth model is that the stories moved from a heavenly, visionary Jesus toward a solid one who can be handled and who eats, and that this move answered doubters and docetists.  Its proponents state it in two forms.  Lüdemann's is a claim about layers:  reports that "stress the bodily nature of the risen Christ are presumably of later origin."⁶  Carrier's is a sequence across the texts, from Paul's "revelatory" experiences through Mark, Matthew and Luke to "an incredible physical encounter in John."⁷  Allison reports that "Much modern scholarship ... regards the texts just cited as relatively late and apologetical, perhaps even directed at an emerging docetism," and that "A trajectory from less literal to more literal is, then, plausible."¹

The model makes four predictions that can be checked.  Each observation below is matched to the prediction it tests.

Prediction 1: explicit physical demonstration appears only in the later texts

Holds.  Luke's Jesus says "See my hands and my feet, that it is truly me. Touch me and see, for a spirit doesn't have flesh and bones, as you see that I have," then asks for food and eats broiled fish "in front of them."¹¹  John's Jesus tells Thomas to "Reach here your finger, and see my hands. Reach here your hand, and put it into my side."¹¹  Acts has Peter say the witnesses "ate and drank with him after he rose from the dead."¹¹  Nothing of this kind is in Paul or Mark.  Matthew has contact without demonstration:  the women "took hold of his feet, and worshiped him" (28:9).  Nobody doubts, nothing is proved, and the touch is an act of worship.  That places physical contact in a Gospel usually dated before John, which limits only a version of the model in which touch is John's or Luke's invention.  Carrier's version already ranks Matthew as "vaguely physical" and reads the grasp as "probably a didactic or dramatic embellishment in a description of a purely revelatory episode."⁷  That is his interpretation;  the text itself says only that they held his feet.

Prediction 2: the demonstrations are framed as answers

Holds, but it does not date the question.  Luke's disciples "supposed that they had seen a spirit";  John's Thomas refuses to believe without touching;  Acts speaks of "many proofs."¹¹  An answer implies a question.  It does not say whose.  The question "was it a ghost?" could be the disciples' own in the scene, as Luke tells it;  a question the writer expected his readers to ask;  or the charge of an outside opponent.  The first two need no late date.  The third points to docetism, and docetism is named clearly only later:  Ignatius, under Trajan, writes against people who "say, that his Passion was merely in semblance."²⁰  The Johannine letters warn against those "who don't confess that Jesus Christ came in the flesh";¹¹  whether that is the same teaching, and when the letters were written, are both debated.  So the framing fits apologetic composition, and it also fits a narrator reporting doubt that he believed was really felt.  It does not, by itself, show that the scenes were composed late or invented.

Prediction 3: the motif keeps growing after the canon

Holds for growth;  weaker for growth from Luke.  Ignatius gives a version close to Luke's:  "'Take, handle me and see that I am not a phantom without a body.' And they immediately touched him and believed ... And after his Resurrection he ate and drank with them as a being of flesh."²⁰  He goes beyond Luke by narrating the touch that Luke and John only offer.  But his wording is not Luke's.  Luke has "a spirit doesn't have flesh and bones";  Ignatius' Greek says literally "not a bodiless demon."²⁰  Eusebius did not know where the words came from:  "taken I know not whence."²¹  Origen assigns the saying "I am not an incorporeal demon" to a book called "The Doctrine of Peter,"²² and Jerome to "the Gospel which I have recently translated."²³  So Ignatius shows the touch motif circulating about 110 in a written form not taken from Luke's wording.  That could be a rewriting of Luke's scene or a parallel tradition, and it cannot be dated earlier than Ignatius.  Ignatius' own date is itself not fixed:  Eusebius' chronicle puts his martyrdom about 108, and Kirsopp Lake reports that "Modern critics are by no means unanimous as to the correctness of this date," though most place it under Trajan, 98-117.²⁰  The Epistle of the Apostles, most often placed in the middle of the second century (Part 6), has Jesus tell Peter and Thomas to put their fingers in the wounds and Andrew to "look on my feet and see whether they press the earth; for it is written in the prophet: A phantom of a devil maketh no footprint on the earth," after which "we touched him, that we might learn of a truth whether he were risen in the flesh."¹⁸  Here the apologetic purpose is explicit and the proof has grown a new test.

Prediction 4: the texts line up from least to most physical in the order they were written

Partly holds.  From Mark to Matthew to Luke the order fits:  no appearance, a grasp in worship, a demonstration with flesh, bones and fish.  Three points weaken the rest of the sequence.

1. Paul's end of the scale is contested. Paul speaks of a body (σῶμα) that is "sown a natural body" and "raised a spiritual body," and says "flesh and blood can't inherit God's Kingdom."¹¹ Whether "spiritual body" means a body made of spirit or a body animated by God's spirit is disputed; Paul's Letters as Historical Evidence weighs the readings and finds that Paul's words favour a transformation of the same body that was sown. This article does not re-argue that verdict. What matters here is narrower. Paul says "body," so his account is not of a bodiless ghost, and Luke's Jesus denies being exactly that (a πνεῦμα, "spirit"). Luke's phrase is "flesh and bones," not Paul's "flesh and blood," and nothing in either text shows that one writer had the other's phrase in view. If Carrier's reading of Paul is right, the gap from Paul to Luke is large; if the transformation reading is right, it is smaller. The trajectory's starting point depends on a disputed reading.
2. John is not clearly more physical than Luke. John has the most invasive invitation (a hand in the side) and has Jesus breathe on the disciples and prepare a meal. But John narrates no eating by Jesus and has no "flesh and bones" statement.¹¹ He ends the Thomas scene with "Blessed are those who have not seen and have believed,"¹¹ which points readers away from physical proof. Carrier's ranking of John above Luke rests on reading John as "to some extent an embellishment" of Luke,⁷ which is the question of John's sources again.
3. The order of Luke and John is itself uncertain. On the dates in Part 2, Luke could be earlier or later than John by some decades, or close to it.

What does not count against the trajectory

Three observations are often offered against the physical trajectory.  They do not test it.

1. "Don't hold me." John's Jesus tells Mary, "Don't hold me, for I haven't yet ascended to my Father" (Μή μου ἅπτου).¹¹ The present-tense command is commonly read as "stop holding me"; Licona renders it "stop 'clinging'."¹⁰ On that reading Mary is already holding him. The verse limits the purpose of contact, not its possibility.
2. Locked doors. John says that "when the doors were locked where the disciples were assembled ... Jesus came and stood in the middle."¹¹ John does not say how he came in, and does not say that he passed through the material of the doors.
3. Vanishing and not being recognised. The two at Emmaus walk and talk with Jesus for miles, but "their eyes were kept from recognizing him"; when they do, "he vanished out of their sight."¹¹ Wright puts the point sharply: "The same Lukan text that tells us that Jesus ate broiled fish, and invited his followers to touch him and see he was real, also tells us that Jesus appeared and disappeared at will, that at one of these appearances two close friends and colleagues did not recognize him, and that in the end he was taken up into heaven."⁸ This is a real feature of the texts. But a body that can be touched and can also appear, vanish and go unrecognised is what a transformed body would be, and a writer elaborating a transformed body could include both. These features are not the opposite of touchability. They count against a model in which later writers turned Jesus into an ordinary revived man. No proponent cited here holds that model.

Two further considerations, one each way

1. Wright on the direction of change. Wright argues that the move from a non-bodily to a bodily view would run backwards: "If there was likely to be development, the model we find in Josephus, for example, suggests that we might expect a hellenistic-style 'spiritualizing' of the tradition, not a re-Judaizing of it."⁸ Allison replies that Wright "assumes the unitary nature or undivided origin of the relevant stories. An equally sensible view is that those stories preserve early elements right beside later ones."¹
2. Allison on solidity. Allison, who grants the trajectory is plausible, then offers "dissent": most apparitions reported by the bereaved are "opaque rather than transparent," and some visionaries report touching the figure.¹ If experiences themselves are often perceived as solid, a report of solidity is not by itself a mark of late development.

What the test shows

The broad claim, that the stories grew steadily more physical from Paul to John, is half right.  The sequence fits from Mark to Luke, its starting point in Paul depends on a contested reading, and John does not clearly continue it.  The narrow claim, that the explicit proof scenes (invitation to touch, wounds displayed, eating to answer the fear of a spirit) belong to later layers shaped by apologetic needs, has real support:  they first appear in the later Gospels, they are framed as answers, and the motif kept growing in the second century in writers who name their opponents.  These features give a reason to consider development, but do not distinguish invention from the selection and shaping of earlier reports.  No earlier surviving source describes the scenes, and silence in a formula or an unfinished narrative carries limited weight.  Allison's position, that the stories "preserve early elements right beside later ones," fits the evidence better than either a purely legendary or a purely documentary reading.  It also makes the job of sorting the two, scene by scene, harder than either side usually admits.

One text-critical detail matters for anyone who builds on Luke 24.  The verse in which Jesus "showed them his hands and his feet" (24:40) is printed in brackets in the SBL Greek text.¹³  Codex Bezae and some Old Latin and Syriac witnesses lack it.  In its 1971 commentary the editorial committee kept it with its lowest confidence rating, {D}, judging that a copyist borrowing from John would have written "side" rather than "feet";  the committee's later rating was not checked for this article.¹⁵  The invitation in 24:39 is not in doubt.

Part 6: What later growth looks like, and how strong the controls are

The best control for "legendary development" is a text that is clearly later, whose growth can be traced to a known earlier text.  The second century supplies candidates.  Each needs four things kept apart:  when the work was composed, the date of the copy that survives, whether the surviving text is the work an ancient writer names, and whether it took its material from the canonical Gospels or from somewhere else.

The Gospel of Peter.  Its only substantial text is a fragment found in 1886 in a grave at Akhmim in Egypt, in a codex now dated by most scholars to late antiquity.  The estimates vary widely.  The first editor, Urbain Bouriant, put it in the eighth to twelfth centuries, and the introduction to the old Ante-Nicene Fathers translation still gives that range.²⁴  Helen Jacobus, writing in 2026, lists later judgments:  "The codex has been dated to the late fifth century by Cavallo and Maehler, and to the fifth to sixth centuries by Grenfell and Hunt."²⁵  Paul Foster, in his 2010 edition and commentary, concludes that "a date of the late 6th to the early 9th century provides a highly probable period."²⁶  Jens Schröter says "probably from the sixth or seventh century."²⁷

The work itself is older.  Serapion, bishop of Antioch at the end of the second century (Foster gives the widest range as 189-211), found a "Gospel of Peter" read at Rhossus and wrote against it, saying it contained "many things in accordance with the true doctrine of the Saviour, but some things added to that doctrine."²⁸  That the Akhmim text is Serapion's book is, in Foster's words, "an assumption of modern scholarship," since the fragment has no title;  he still judges the identification "the most likely explanation."²⁶  A small Oxyrhynchus papyrus of about 200 (P.Oxy. 2949) may be part of the same work;  Schröter calls this "possible, but not undisputed."²⁷

The date of composition and the work's sources are disputed, and the dispute matters here.

1. The majority view: second century, using the canonical Gospels. Foster writes that "The majority of critical scholarship, despite the challenges raised by Crossan and other, still prefers to locate the text in the second century," and concludes that "a date of composition during the period 150-190 CE seems the most sensible suggestion."²⁶ He judges dependence on the canonical accounts a better explanation of the text than Crossan's early-source proposal.²⁶ Schröter calls it a creative "recreation" of the Jesus story "from a second-century perspective" that "builds on an already developed tradition of the passion story."²⁷ M. R. James had judged in 1924 that it "uses all four canonical Gospels."¹⁸
2. The early-source view. John Dominic Crossan argued that the fragment contains an earlier document, his "Cross Gospel," and that "the original Cross Gospel is the one passion and resurrection narrative from which all four of the intracanonical versions derive."²⁹ On his view the canonical resurrection stories depend on the Gospel of Peter's source, not the other way round. Joel Marcus has argued for an origin in the first century, as Jacobus reports.²⁵ Foster judges that "the case for the dependence of the canonical gospels upon material embedded in the Gospel of Peter is not particularly strong."²⁶

On the majority view, the Gospel of Peter is a later writer's expansion of the canonical tradition.  On Crossan's view it preserves an early layer, and its spectacular rising would then be early too.  Either way, its resurrection scene reads:

"there came a great sound in the heaven, and they saw the heavens opened and two men descend thence, shining with (lit. having) a great light, and drawing near unto the sepulchre. And that stone which had been set on the door rolled away of itself and went back to the side ... they saw again three men come out of the sepulchre, and two of them sustaining the other (lit. the one), and a cross following after them. And of the two they saw that their heads reached unto heaven, but of him that was led by them that it overpassed the heavens. And they heard a voice out of the heavens saying: Hast thou (or Thou hast) preached unto them that sleep? And an answer was heard from the cross, saying: Yea."¹⁸

Compared with the canonical texts, the Gospel of Peter does four things.  It narrates the moment of resurrection.  It makes the guards, a centurion and Jewish elders the witnesses, so that the proof comes from enemies.  It adds a cosmic scale and a speaking cross.  And it still keeps the women's visit, the young man in the tomb and the women's flight from Mark.¹⁸  Allison calls Matthew's account "on its way to the Gospel of Peter, with its spectacular, colorful details that nobody mistakes for history."¹

The longer ending of Mark (first half of the second century by Metzger's estimate, quoted by Irenaeus about 180) summarises Luke, John and Acts and adds promised signs:  "they will take up serpents; and if they drink any deadly thing, it will in no way hurt them."¹¹  Its dependence on the canonical texts is the clearest of any of these controls.

The Epistle of the Apostles.  Carl Schmidt, its first modern editor, dated it to "about A.D. 160."¹⁸  Julian Hills, in his 1990 study, describes "the present consensus, that the Epistula was composed in the mid-second century in Asia Minor, in Egypt, or in Syria," and notes that a disputed verse in it (17.2) has often been used to date it, "but beyond the general range of date indicated (120-180 CE) there has been no firm advance."³⁰  It adds a footprint test and has the apostles touch Jesus "that we might learn of a truth whether he were risen in the flesh."¹⁸

These controls cut two ways, and both should be stated.

1. They show restraint in the canonical stories. The canonical Gospels do not narrate the rising, do not make the guards the witnesses of Jesus himself, do not give Jesus a giant or shining form, and do not stage a test by footprints. Craig makes this point: the Gospel of Peter shows "how real legends look."³¹ Wright's observation that the canonical Jesus does not shine "like a star" points the same way.⁸ On Crossan's view of the Gospel of Peter this contrast weakens, because the spectacular account would then be early.
2. They also show continuity. The second-century texts develop the lines that critics see beginning in the canon: guards and hostile witnesses (Matthew), physical proof against a phantom (Luke and John), and summaries that fill gaps (Mark's endings). Allison's comment is that the second-century inventors were "not without first-century predecessors who, among other things, contributed to the canonical traditions about Jesus' resurrection."¹ Restraint relative to the Gospel of Peter is not evidence that no development took place; it is evidence that the canonical writers did not go as far.
3. They are not all controls of the same strength. The longer ending of Mark is a demonstrable later summary of known texts. The Gospel of Peter is one on the majority view, but its date and sources are argued, not documented. Ignatius' touch saying comes from an unknown source. The Epistle is dated only within a range of decades. Later, more elaborate attestation documents a more elaborate version; it does not by itself show which earlier text was being elaborated, or when a particular detail was first told.

Part 7: How fast do legends form? Sherwin-White and the ancient cases

What Sherwin-White argued

A. N. Sherwin-White was an Oxford historian of Roman law and administration.  His Sarum Lectures of 1960-1961, published in 1963, end with a short section, "The historicity of the gospels and Graeco-Roman historiography," inside the last lecture.  His starting point is that ancient historians usually work from "derivative sources of marked bias and prejudice composed at least one or two generations after the events which they describe, but much more often, as with the Lives of Plutarch or the central decades of Livy, from two to five centuries later," and still believe "a hard core or basic layer of historical truth can be recovered."²  The reason, he says, is "the existence of external confirmations, and the working of the synoptic principle," which reveals "a basic unitary tradition beneath the manifold divergences of detail in rival narratives."²  External confirmation is part of his method, not an afterthought:  "From time to time external contemporary evidence ... confirms the conclusions drawn from the critical study of literary sources. Hence we are bold to trust our results in the larger fields where there is no such confirmation."²

His argument about time is aimed at "form-criticism of the extremer sort" and at "the presumed tempo of the development of the didactic myths."²  He puts it as a matter of likelihood:  "The agnostic type of form-criticism would be much more credible if the compilation of the Gospels were much later in time, much more remote from the events themselves, than can be the case."²  He grants the first step:  "Certainly a deal of distortion can affect a story that is given literary form a generation or two after the event, whether for national glorification or political spite, or for the didactic or symbolic exposition of ideas. But in the material of ancient history the historical content is not hopelessly lost."²  Then comes the sentence at the centre of the modern debate:

"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."²

His test case is the Athenian tyrannicides.  "The true story was that they assassinated Hipparchus in 514 B.C., but the tyranny lasted another four years before the establishment of the Athenian democracy. Popular opinion created a myth to the effect that Harmodius and Aristogeiton destroyed the tyranny and freed Athens. This was current in the mid-fifth century. Yet Herodotus, writing at that time, and generally taking the popular view of the establishment of the democracy, gives the true version and not the myth."²  His count of generations follows from those dates:  from 514 to the middle of the fifth century is about sixty years, two generations at about thirty years each.  His conclusion is modest:  "the falsification does not automatically and absolutely prevail even with a writer like Herodotus."²  Applied to the Gospels, his claim is that those with "a passionate interest in the story of Christ ... would not be led by that very fact to pervert and utterly destroy the historical kernel of their material."²

Alexander comes into the argument in a footnote, not in the main text.  The lecture ends on p. 192, and its last footnote, running onto p. 193, reads:  "Mr. P. A. B. Brunt has suggested in private correspondence that a study of the Alexander sources is less encouraging for my thesis. There was a remarkable growth of myth around his person and deeds within the lifetime of contemporaries, and the historical embroidery was often deliberate. But the hard core still remains, and an alternative but neglected source—or pair of sources—survived for the serious inquirer Arrian to utilize in the second century A.D. This seems to me encouraging rather than the reverse."²  So Sherwin-White accepts that myth about Alexander grew fast and deliberately, and reads the case as support, because the core survived and a sound source reached a later historian.  The same footnote states his purpose:  "The point of my argument is not to suggest the literal accuracy of ancient sources, secular or ecclesiastical, but to offset the extreme scepticism with which the New Testament narratives are treated in some quarters."²

Read in full, Sherwin-White claims four things and denies a fifth.

1. A mythical tendency is at work within one or two generations.
2. In his test case it did not "prevail" over the core within two generations.
3. A determined inquirer can recover detail even a century later "by inquiry within the inner circle of the descendants of those concerned."²
4. Myth about Alexander grew within his contemporaries' lifetimes, deliberately, and the core survived anyway, which he counted in his favour.
5. He does not claim literal accuracy for the Gospels, and he does not claim a minimum time for legend to appear.

How the argument has been used, and the critics

Julius Müller's 1836 review of D. F. Strauss made an older version of the argument:  "Most decidedly must a considerable interval of time be required for such a complete transformation of a whole history by popular tradition, when the series of legends are formed in the same territory where the heroes actually lived and wrought."³²  Müller challenged Strauss to show that "within thirty years, on a clear historical scene, not strange fables, — for thirty years are not requisite for that, — but a grand series of legends, the most prominent elements of which are fictitious, have anywhere gathered round an important historical individual."³²  The words "not strange fables, — for thirty years are not requisite for that" are Müller's own concession that individual fables can arise quickly.  His claim was about a whole series replacing the history.

William Lane Craig quoted Müller in 1981 and added Sherwin-White, writing that "According to Professor Sherwin-White, generations are required for legends to prevail over historical facts."³³  In On Guard (2010) he wrote that the gap is "just too short for this to have happened" and that "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."⁹  As Part 1 shows, Craig's stated claim, like Müller's, is about erasing a core.  Single marvels that attach quickly to a remembered event therefore do not refute that claim as he states it.  They bear on his separate argument that eyewitnesses checked elaboration.  Craig's use of Sherwin-White has three problems of a different kind, each about wording or attribution.

1. Alexander. Craig writes, in a passage introduced as Sherwin-White's point, that "The fabulous legends about Alexander the Great did not develop until the centuries after these two writers," meaning Arrian and Plutarch.⁹ Sherwin-White's footnote says the reverse: "a remarkable growth of myth around his person and deeds within the lifetime of contemporaries."² Diodorus, writing in the first century BC, more than a century before either, already tells the story of the Amazon queen who came to Alexander "for the purpose of getting a child."³⁴ This is a factual and attributional error, not a matter of emphasis.
2. "Unbelievable." Craig's 1981 book says that Sherwin-White "comments that for these stories to be legends, the rate of legendary accumulation would have to be 'unbelievable'; more generations are needed," citing pages 188-91.³³ Only the single word "unbelievable" is inside quotation marks; "more generations are needed" is Craig's own paraphrase, and a fair one of Sherwin-White's "would be much more credible if the compilation of the Gospels were much later in time" and his two-generation sentence. The quoted word itself does not appear in Sherwin-White's book as far as a search of three separate scans can reach, including its footnotes. The nearest wording is "much more credible." So the substance of Craig's sentence is a defensible paraphrase, while the adjective he marked as a quotation remains unverified. The details are in this article's quarantine file.
3. "Show" for "suggest," and the Acts sentence. Sherwin-White wrote that the tests "suggest"; Craig's 2010 version says they "show." Craig also quotes Sherwin-White as saying "Any attempt to reject its historicity even in matters of detail must now appear absurd," where Sherwin-White wrote "its basic historicity" and said it of Acts, which he called in the same breath "no less of a propaganda narrative than the Gospels, liable to similar distortions."²

Critics have pressed the gap from the other side.  Allison quotes the two-generation sentence and replies:  "Yet these words deny only that the mythical tendency will prevail, not that it will fail to manifest itself at all."¹  His quotation drops the word "oral" ("the hard historic core of the oral tradition"), which does not change his point.  Richard Carrier argues that Sherwin-White "freely admits" distortion within two generations, that his argument "rests merely on a single case," that "Herodotus believed without question many obvious legends," and that the Athenian truth could be recovered partly because of an inscription that the Gospel writers lacked.⁷  Some of this needs correcting.  Carrier says Sherwin-White's remarks "are confined solely to the trial of Jesus";  the section discusses the Gospel narratives and the "actions and sayings of Christ" generally, though its worked example is the trial.²  He faults Craig for claiming that "tests" (plural) were made, but the plural is Sherwin-White's own word.⁷  On the inscriptions he is partly right:  Thucydides does cite an altar inscription and a pillar on the Acropolis, but to prove that Hippias was the eldest son and the ruling tyrant, not to establish the killing itself.³⁵  His point that the core itself is disputed stands:  he suggests the "hard historic core" may be that Christ "was executed, buried, and then, through epiphanies and scripture, believed to have been raised from the dead."⁷

Seven dated cases, with three clocks

The table sets out seven cases, six ancient and one from the seventeenth century, chosen because each has a datable event and a datable witness.  For each it gives the three clocks from Part 1:  the event, the time a source assigns to the change, and the earliest surviving witness.  It also gives the status of the change in the sources:  whether anything in them shows it to be false.  It is a small, selected set:  it shows what can happen and how fast, not how often.

The details:

1. The tyrannicides. Herodotus says that after the killing "the Athenians continued for four years after this to be despotically governed no less than formerly,--nay, even more," and that the Alcmaeonids "set Athens free much more than Harmodios and Aristogeiton."⁴⁰ He was writing after 430 BC, since he mentions the Spartan envoys put to death by the Athenians "during the war between the Peloponnesians and Athenians," an event Thucydides places in the war's second year.³⁵⁴⁰ Thucydides calls the popular account an error: "The general Athenian public fancy that Hipparchus was tyrant when he fell by the hands of Harmodius and Aristogiton," and tells the story "to show that the Athenians are not more accurate than the rest of the world in their accounts of their own tyrants and of the facts of their own history."³⁵ He also says that the Athenian commons of his own day had heard that the tyranny had been ended "not by themselves and Harmodius, but by the Lacedaemonians."³⁵ The myth and the correct account lived side by side, and a careful inquirer could still find the second about a century after the event.
2. Alexander. Plutarch reports that "the queen of the Amazons came to see him, as most writers say," naming five, while nine others "say that this is a fiction." He adds that Alexander's own letter to Antipater mentions the Scythian king's daughter "but he makes no mention of the Amazon," and that "the story is told that many years afterwards Onesicritus was reading aloud to Lysimachus, who was now king, the fourth book of his history, in which was the tale of the Amazon, at which Lysimachus smiled gently and said: And where was I at the time?"⁴¹ Lysimachus took the title of king with Alexander's other successors in the archon year 307/6 BC.⁴² Arrian says Onesicritus was the pilot of Alexander's ship, "who, in the narrative which he composed of Alexander's campaigns, falsely asserted that he was admiral," and names Lysimachus among "the confidential body-guard."³⁶ Arrian also reports a different Amazon story, of a hundred armed horsewomen given to Alexander in Media, notes that it was "recorded neither by Aristobulus nor Ptolemy," and suggests the women shown, if any, were "some other foreign women trained in horsemanship."³⁶ So a participant wrote a story that other participants did not confirm, the story survived, and it may have grown from a real encounter. The anecdote of Lysimachus' laugh is itself told by Plutarch as a story.
3. Augustus. Suetonius records that "There was even an ex-praetor who took oath that he had seen the form of the Emperor, after he had been reduced to ashes, on its way to heaven."³⁷ Dio names him and says Livia "bestowed a million sesterces upon a certain Numerius Atticus, a senator and ex-praetor, because he swore that he had seen Augustus ascending to heaven after the manner of which tradition tells concerning Proculus and Romulus."⁴³ Both writers place the oath at the time of the deification, but both wrote a century or more later. The payment is Dio's report two centuries later. Nothing in these sources tests whether Atticus saw anything.
4. The Jerusalem portents. Josephus lists a light round the altar, a heifer that gave birth to a lamb in the Temple, the Temple's brass gate opening by itself, and, a few weeks later, "chariots and troops of soldiers in their armor were seen running about among the clouds, and surrounding of cities." He says: "I suppose the account of it would seem to be a fable, were it not related by those that saw it."⁴⁴ The Greek for "were seen" is ὤφθη, the verb Paul uses of the appearances.⁴⁵ Josephus says he offered the War "to the emperors themselves, when the facts were almost under men's eyes," and that Titus signed the books and ordered them published.⁴⁶ That is Josephus' own claim, and even if true it does not show that anyone checked the portents. The witnesses "related" what they saw; Josephus does not say they swore to it. Tacitus has his own version, with "Contending hosts were seen meeting in the skies" and the shrine doors opening as "a superhuman voice cried: 'The gods are departing'";⁴⁷ whether he drew on Josephus is not settled here. What the case shows is that a history written within about a decade, by a participant, carried reports of marvels attributed to eyewitnesses. It does not show that the marvels were invented. Some could have been misperceived natural events, some exaggerated, some reinterpreted after the war; the sources do not say.
5. Apollonius of Tyana. Apollonius, a wandering Pythagorean teacher, died in old age under Nerva. Philostratus wrote his Life at the request of the empress Julia Domna, and it "cannot have been published before the year 217," in the words of its Loeb translator, F. C. Conybeare. Conybeare lists his sources: memoirs by a disciple, Damis, a history of Apollonius' time at Aegae, letters, and stories Philostratus gathered on his travels. Charles Bigg argued in 1910 that Damis was "a mere man of straw invented by Philostratus," a view Conybeare thought "unnecessarily sceptical."³⁸ On the end of Apollonius' life Philostratus writes: "with regard to the manner in which he died, if he did actually die, there are many stories, though Damis has repeated none." He gives deaths at Ephesus and Lindus, and on Crete an entry into a temple whose doors closed behind him while maidens sang "Hasten thou from earth, hasten thou to Heaven, hasten."⁴⁸ Then comes an appearance. "Apollonius had already passed away from among men, but people still wondered at his passing." At Tyana a young man who denied that the soul was immortal said Apollonius was "so utterly dead that he will not appear to me in response to my entreaties." Five days later he woke, cried "I believe thee," and said the sage was present reciting verses. The others asked, "But where is he? ... For we cannot see him anywhere," and he answered, "It would seem that he is come to converse with myself alone." Philostratus adds that he never met "any tomb, however, or cenotaph of the sage."⁴⁸ Of the cases here, this is the nearest in kind to the Easter claims: a follower-type report that a dead teacher appeared. Its clocks differ from Paul's. The one who saw him is unnamed, the people beside him saw nothing, and the earliest written account comes more than a century after the death, from an author who treats the death itself as uncertain. What the case cannot show is when the story began. Philostratus had older sources, and earliest attestation is not first origin.
6. Peregrinus. Lucian, who watched the cremation, says he told the gullible that "a vulture, flying up out of the midst of the flames, went off to Heaven," a story he says he made up. Back at the festival he met "a grey-haired man whose face, I assure you, inspired confidence," who said he had seen Peregrinus "in white raiment a little while ago," and "Then on top of it all, he put the vulture, swearing that he himself had seen it flying up out of the pyre, when I myself had just previously let it fly to ridicule fools and dullards."³⁹ This is satire by a hostile witness, and the old man cannot be checked. It is still a dated report, by someone who was there, that an invented marvel was sworn to within days, and that a sighting of the dead man was claimed.
7. Sabbatai Sevi. The seventeenth-century Jewish messianic claimant is the best-documented case of fast legend among believers. Gershom Scholem, the historian of the movement, writes: "The transition from history to legend took place with extraordinary rapidity in what are practically eyewitness accounts. Already the earliest documents confuse dates and chronologies, and abound in legendary accounts of miracles." His example is "A letter written in Gaza early in July, 1665 ... purporting to give an account of the events of the preceding four months."⁴⁹ Sabbatai converted to Islam in 1666 and died in 1676; "The Sabbatian traditions all agree that Sabbatai died in Dulcigno." The story of his death then grew in stages that Scholem's sources let him date. At first: "Samuel Gandoor does not yet say that Elijah Sevi or other believers and rabbis had been present at his deathbed. Israel Hazzan already tells of the presence of a group of believers, but makes no mention of Sabbatai's brother." Then "A few years later Baruch of Arezzo recounts the full-blown legend," in which the brother came to the burial cave "on the third day" and "found the cave empty. 'Neither Our Lord nor anything else was in the cave, but it was full of light.'" Scholem traces the legend to the believers' feeling that "perhaps this was no 'real death' at all," and to the earlier belief that at his apostasy the messiah had become "hidden."⁴⁹ A few years after the first reports, then, believers told a third-day empty-tomb story that those reports did not contain. In Scholem's account of how the legend formed, no one claims to have seen Sabbatai after his death, and Nathan of Gaza, the movement's prophet, "possibly understood the mystery" but "preferred to keep silent."⁴⁹ Allison mentions later "sightings of the postmortem Sabbatai Sevi," about which he says he was "unable to learn much."¹

What the cases show, and what they do not

1. Marvel claims could attach to public events and persons almost at once. In two cases a near-contemporary dates them within days (Lucian) or about a decade (Josephus). In two the early date comes from writers a century or more later (Augustus, the Amazon). In the best-documented case, Scholem finds legendary miracle reports about Sabbatai Sevi in a letter written within about four months of the events. So a story's early date does not, by itself, protect its details. Müller conceded as much in 1836 ("not strange fables, — for thirty years are not requisite for that").
2. Only one case shows an invention by the inventor's own admission, and that admission comes from a satirist. The Amazon story was disputed in antiquity and may have a real seed. The tyrannicide belief was contradicted by better-informed writers. The ascent of Augustus and the Jerusalem portents are reports whose truth these sources cannot settle. The Sabbatai case shows growth another way: the dated first reports of his death lack the empty cave, and a later account has it. The cases show how fast claims can appear, not that each claim was false. An extraordinary report is not shown false by being extraordinary.
3. Public cores survive. In six of the seven cases the public event (the killing, the campaign, the funeral, the war, the cremation, the apostasy and the death at Dulcigno) is preserved in the same sources that carry the marvels. Sabbatai's believers kept even his conversion to Islam, the most damaging fact in their story, and built a theology to explain it; Scholem describes the need "to explain and rationalize the painful contradiction between historical reality and faith."⁴⁹ The exception is Apollonius, whose place and manner of death, and even whether he died, Philostratus could not settle more than a century later. A century's distance blurred that core. This is Sherwin-White's real point, and the cases support it.
4. Correction did not eradicate these stories, and no rate can be drawn from them. One participant's reported laugh did not remove the Amazon story; contemporary correction did not end the tyrannicide belief; Josephus' readers, if they read critically, left the portents in. That shows that correction by participants can fail. It does not measure how often it succeeds, because successful corrections leave no story behind to count.
5. Compared like with like, the cases cut both ways. Public death against public death: the crucifixion, like the funeral of Augustus and the cremation of Peregrinus, is the kind of core that survives. A named man's claim to have seen the dead or departed against a named man's claim: Numerius Atticus, named by Dio, claimed at once and on oath to have seen Augustus rise, and Paul names Cephas. In both, an early named claim shows that the claim was made, not why. Claims by groups: Josephus' unnamed "those that saw it" and Lucian's festival crowd are reports about crowds by a single writer, while Paul's list names a group (the twelve) whose leaders he knew. Post-death claims against post-death claims: Apollonius' appearance is to one unnamed youth whose companions saw nothing, first written down more than a century later. Sabbatai's death produced a third-day empty-cave story a few years after the first reports, but in Scholem's account of it no one claims to have seen him. The Sabbatai case is the closest parallel here to the growth of a tomb story, and the strongest case in this set against the claim that there was too little time for one. A skeptic will add that its pattern is formally like the Christian one: Paul's formula says Jesus "was buried" and "was raised on the third day" without describing a tomb, and Mark, decades later, is the first to narrate the empty tomb. Whether the Gospel tomb story was early is argued on other grounds in The Empty Tomb. No case examined here shows a series of named recipients, including groups, making their own claims within a few years and then having them elaborated. That absence is a limit of the comparison. It supplies no measured likelihood either way.

Applied to the resurrection, Sherwin-White's argument supports what Part 2 found from Paul:  the core claim that Jesus died and was seen by named followers is early and is the kind of thing that survives.  The cases show that a core can survive alongside fast-growing marvels, and that an early named claim of a post-death sighting can be made whether or not it is true.  So the argument does not certify a single detail of the Gospel scenes, and it does not certify that the core claim itself was true.  Whether Jesus ate fish in Jerusalem or the women held his feet is exactly the kind of detail that, in these cases, can grow within the first generation.  The argument also leaves the main dispute open, because what counts as the "core" is itself contested:  Carrier puts the empty tomb outside it,⁷ and The Empty Tomb argues for it on other grounds.

Part 8: What the evidence can decide, element by element

Four kinds of evidence can separate early memory from later growth.

1. An earlier witness that has the element, or lacks it while having reason to mention it. Paul's list has reason to name recipients and no reason to describe scenes, so his silence on scenes is not evidence against them. Mark's ending stops before any appearance, so his silence on appearance scenes is not evidence either.
2. Visible editing, where a later writer's source survives.
3. A dated control that shows how a motif behaved later, from the same line of tradition.
4. Features that cut against a writer's interest. This is the weakest: it shows that a detail was probably not invented for that particular purpose, not that it is accurate.

Date of writing is a fifth consideration, and it is not neutral.  A later first attestation leaves more time and more hands for change, so it lowers confidence in a detail's wording and arrangement.  It does not show that the detail was invented, because a later writer can also preserve older material.

The grades in the table use these tests.  Each is a judgment about the evidence, not a probability, and no numbers are attached because no data supply them.

1. Early as a claim: attested by Paul or by inference before the 50s. Secure that the claim was made; says nothing about content.
2. Probably early: first attested in a Gospel, with a specific positive reason to think it older, such as an earlier text that presupposes it.
3. Undecided: first attested in a later Gospel, with no specific evidence either way. This is not a verdict of "equally likely"; it is a statement that the evidence that would decide is missing.
4. Probably secondary: positive evidence of later addition, such as a visible addition to a surviving source together with a reason the addition serves the writer, or a dated control showing the motif expanding.
5. The writer's wording: language and arrangement characteristic of the author. This says who shaped the telling, not whether a report lay behind it.

The best case on each side, scene by scene, comes down to this.  For the proof scenes, the skeptic's best case is the combination of late first attestation, apologetic framing and later expansion;  the defender's best case is that a demonstration of bodily reality answers a question the first witnesses would themselves have faced, that Luke presents doubt as the disciples' own, and that apparitions are often perceived as solid.  For the women's meeting with Jesus, the skeptic's best case is that Matthew's words merely repeat his angel's and John may have developed Matthew;  the defender's is that two quite different scenes sharing one recipient point to an older report.  For the guard, the skeptic's case is strong and the defender's, as The Empty Tomb finds, is weak.

What follows for the hallucination debate

Hallucination Theories of the Resurrection found that no collective case it examined shows a larger group independently reporting a familiar person conversing with them at length, and that the force of this point depends on whether the Gospel scenes of conversation, touching and eating preserve the original experiences.  This article's answer is that they cannot be shown to.  The group appearances themselves are early as claims (the twelve, the five hundred, all the apostles).  Their content is unrecorded in the early sources.  The scenes that supply content are written decades later, by writers who are seen reshaping their sources elsewhere, and in the case of touching and eating carry an apologetic framing that continues into the second century.

That cuts in both directions.  A critic cannot treat invention as established by the late date and framing alone, and Allison, no friend of harmonisation, thinks the stories preserve "early elements right beside later ones."¹  A defender likewise cannot treat the scenes' first-century origin as independently confirmed, and the clearest motif in them is one that later writers kept developing.  The comparison in the sibling article therefore stays where it was left:  a limit of the comparison, not an impossibility proof and not a confirmation.

Mixed models

Allison's position combines early experiences with later growth.  Other mixed models are possible:  a few visionary experiences with social spread and later narrative development (Ehrman),⁵ revelatory experiences reinterpreted through scripture and then embellished (Carrier),⁷ or a real encounter whose report was later shaped by apologetic needs (a traditional reading that accepts redaction).  Each is a legitimate possibility.  None is established because each of its parts has some modern or ancient example.  A combination of individually documented steps is still only as good as the evidence that each step happened in this case.

Two cautions apply in opposite directions.  The absence of any exact ancient or modern parallel to the whole series of claims supplies no measured likelihood against a natural account, because no one has a sample of comparable series to measure.  And a traditional account is not "a bare resurrection":  it claims both that Jesus was raised and that he intentionally appeared to particular people.  A bare resurrection does not predict who would see him;  the traditional account does, and it should be judged as the claim it actually makes.  Neither naturalistic nor supernatural causation is assumed here, and historical inquiry is not ruled out for either.  What the evidence cannot supply is a set of numbers for comparing them.

Objections and responses

1.  "The Gospels were written while eyewitnesses were alive.  They would have corrected any legend."  This is Craig's fourth argument, and it is aimed at details.  Living witnesses plausibly constrained the core:  who claimed to have seen Jesus, and roughly when.  For details, the dated cases show that correction by participants can fail:  the Amazon story outlived a participant's reported laugh, and Josephus' history, which he says he gave to men who had fought the war, still reports armies in the sky "related by those that saw it."⁴⁴  Those cases cannot say how often correction works.  Modern memory research measures something related, though not the growth of legend in a community.  William Hirst and colleagues surveyed the same people about the attacks of 11 September 2001 four times over ten years.  They found that "the forgetting observed during the first year can be traced to the non-critical items," and reported "the excellent retention of the critical items."⁵⁰  Among the 202 who answered all four surveys, accuracy for the crash sites went from .95 in the first week to .89 after ten years, and for the number of planes from .95 to .88, while accuracy for the airline names fell from .88 to .56.⁵⁰  Errors also behaved differently by kind:  "Inconsistent flashbulb memories were more likely to be repeated rather than corrected over the ten-year period; inaccurate event memories, on the other hand, were more likely to be corrected."⁵⁰  A flashbulb memory here is a person's memory of his own circumstances when he heard the news.  The pattern fits the line this article draws.  The central public facts held, and errors about the event tended to be corrected, while personal particulars drifted in the first year and the drift then persisted.  These are surveys of individuals about a heavily reported event, not a measurement of stories passed on for forty years;  Oral Tradition and Memory weighs them in detail.  They support the survival of a core such as who claimed to have seen Jesus.  They give no protection to the particulars of the conversation, touching and eating scenes.  Paul's remark that most of the five hundred "remain until now" invites a reader to check that an appearance was claimed;¹¹ it does not check the details of a story told decades later in another city.

2.  "Sherwin-White showed that legend needs at least two generations."  He did not.  He said two generations are too short for myth "to prevail over the hard historic core," granted "a deal of distortion" within one or two generations, and accepted that myth about Alexander grew "within the lifetime of contemporaries," a case he thought "encouraging rather than the reverse" because the core survived.²  His argument supports the survival of a core.  It sets no minimum time for a legendary detail.

3.  "Then Sherwin-White and Craig are refuted by fast-forming legends."  Not as they state their claims.  Both claim that a core is not erased within two generations;  Müller conceded that "strange fables" need much less than thirty years.  A vulture at a pyre or an Amazon in a campaign narrative does not erase a core.  What the fast cases refute is the inference, sometimes drawn from Sherwin-White, that an early date secures the details.

4.  "Paul's list is a later insertion, so there is no early evidence of appearance claims."  Price argues for an insertion of 15:3-11.  He acknowledges that no extant manuscript lacks it, but replies that no manuscripts survive from the period when he places the insertion.³  His hypothesis would remove the list and verse 11, not Paul's claim to have seen Jesus in 9:1 or his contacts with Cephas and James in Galatians.¹¹¹¹  Those contacts alone do not reconstruct the list, and its authenticity must be judged on the textual and internal evidence rather than inferred merely from the visits.

5.  "An early claim cannot be legend, so the appearances must be historical."  Early date rules out late invention.  It does not rule out early growth, and it does not identify the cause of the first claims.  Numerius Atticus claimed at once, by name and on oath, to have seen Augustus rise;  an early named claim shows that a claim was made.  What caused the claims of Cephas and James is the subject of the sibling articles on hallucination and the empty tomb.

6.  "Early growth can explain everything, so the appearances need no further explanation."  A complete account must explain both the early material and how it changed.  Paul's contacts make a purely late attachment of names less plausible on the transmitted text.  They do not exclude changes in the claimants' own interpretation or retelling, and do not reveal what began the process.  Invoking growth names a possible process; it does not document its particular steps here.

7.  "The differences between the Gospels show independent eyewitness testimony."  Where Matthew and Luke differ from Mark in the tomb story, the best explanation is that they rewrote Mark, not that they independently saw the events.  Differences count as independent evidence only where independence can be shown, episode by episode.  The best candidate is John, and its relation to the Synoptics is disputed.  Licona, who defends the accounts, notes that John shares several items with only one Synoptic Gospel and argues:  "If John wrote independent of the Synoptic traditions as most scholars think, these multiple points of contact with unique Synoptic traditions caution us against dismissing John."¹⁰  Goodacre's book argues the opposite premise.

8.  "The contradictions prove the whole story is legend."  Disagreement in detail is normal in ancient sources that preserve a real event.  Sherwin-White's example is Tiberius, whose four sources "disagree amongst themselves in the wildest possible fashion," yet no one doubts his reign.²  Licona himself lists five differences in the resurrection narratives that "leave us bewildered" and concludes, "Therefore, the tensions remain."¹⁰  Unresolved tensions count against the accuracy of the details in tension, not against the existence of the core claim, which is attested earlier than any of the narratives.

9.  "Mark ends at 16:8 because the earliest Christians knew of no appearances."  Mark's own text promises an appearance:  "He goes before you into Galilee. There you will see him, as he said to you" (16:7), pointing back to 14:28.¹¹  Paul's list, which on most dates for Mark is earlier, names many appearances.  The ending at 16:8 shows that Mark chose not to narrate an appearance, or that his ending was lost;  it is not evidence that none was known.

10.  "Luke contradicts Mark and Matthew by keeping everything in Jerusalem."  Luke's narrated appearances before the ascension are all in or near Jerusalem, and he turns Mark's Galilee promise into a reminder.  Acts later narrates appearances near Damascus and visions elsewhere, so the restriction is to Luke's pre-ascension narrative.  Luke omits Galilee appearances;  he does not deny them, and the forty days of Acts 1:3 leave room for them.  Whether Luke's selection misrepresents what happened depends on evidence Luke does not supply.

11.  "Wright's four strange features prove the stories are early and unembellished."  The features are real.  The canonical stories do not quote scripture as the passion narrative does, do not mention the believer's future resurrection, do not make Jesus shine, and begin with women.⁸  They argue against one kind of origin:  a deliberate composition from scripture or doctrine.  They do not show that every detail is early, and Wright himself grants that the fuller accounts "reflect the style, and the theological interest, of the particular evangelist."⁸  The "silence" also needs a qualification.  Luke 24 makes scripture a theme, having Jesus explain "in all the Scriptures the things concerning himself" and say "Thus it is written," though without quoting a text.¹¹  Shaping by scripture without quotation is also claimed:  Carrier reads Matthew's mountain scene as Jesus as "the new Moses" and the women's grasp of the feet as an echo of a psalm.⁷  Those readings are arguable, not established.  And as Allison replies, early and later elements can sit side by side in the same story.¹

12.  "The Gospel of Peter shows what legend looks like, and the canonical Gospels are not like it."  The contrast is real and is evidence of restraint (Part 6), on the majority dating of that Gospel.  It is not evidence that no development occurred.  Matthew's guard and Luke's proofs are smaller steps in the same directions that the Gospel of Peter and the Epistle of the Apostles take further.

13.  "The touching scenes answer docetists, so they must be late."  The scenes answer the question whether the risen Jesus was a spirit.  That question could be the disciples' own, a reader's, or an opponent's;  Luke presents it as the disciples' own.  Ignatius explicitly attacks the semblance teaching, traditionally around 110, though his date is disputed.  So the apologetic framing is real, but it does not date the scenes by itself.

14.  "Since the physical scenes are late, the first experiences must have been non-physical visions."  A late first attestation does not show what the first experience was like.  Matthew's women already hold Jesus' feet without any proof motive, Paul speaks of a raised body, and modern apparition reports are often of solid figures.¹  The early sources simply do not say how the appearances were perceived.

15.  "Lucian is a satirist, so the Peregrinus case proves nothing."  Lucian is hostile and the old man is unverifiable, so the case is weak as a report about Peregrinus.  It is still a contemporary claim, by a man who was at the pyre, that a marvel he invented was being sworn to within days.  The other cases do not depend on satire, but they also cannot show that their marvels were false.

16.  "Every difference can be harmonised."  Many can be, and some harmonies are natural:  Luke's one-day narrative and his forty days in Acts come from the same author, so compression is the obvious explanation.  But a possible harmony is not a demonstrated one, and the defenders' own best treatment leaves differences unresolved.¹⁰  For this article's question, a harmonised detail still has the same late attestation it had before.

17.  "This article assumes Markan priority."  It does, in Part 4, and says so.  Without Markan priority the comparison in Part 4 loses its direction:  the changes could run the other way.  The case for Markan priority is made in The Synoptic Problem.  The other parts do not depend on it:  Paul's list, the second-century texts and the ancient cases stand on their own dates.

18.  "Sabbatai Sevi shows that a third-day empty-tomb story can arise within a few years, so the Gospel story can be legend too."  As to speed, yes.  Scholem's dated sources show believers telling, a few years after the first reports of Sabbatai's death, that his brother found the burial cave empty "on the third day," when the first reports had no such scene (Part 7).⁴⁹  It is the best evidence in this article that the shortness of time alone cannot rule out a tomb legend.  What the case does not contain is the rest of the early Christian evidence.  In Scholem's account no follower, not even Nathan the prophet, claimed to have seen Sabbatai after his death, and believers explained the death as "a second and even deeper occultation," a hiding.  Whether the Gospel tomb story is early is argued in The Empty Tomb.

Issues and unanswered questions

1. John's sources, episode by episode. Whether John's meeting of Mary with Jesus, and his Jerusalem appearance, depend on Matthew and Luke decides how much those scenes weigh. Goodacre's 2025 book argues that John knew all three Synoptics; only its publisher's description was read for this article. Allison treats John's independence as "a live option" and Licona as the majority view.
2. The Gospel of Peter. Its date (Foster 150-190; others earlier), its sources (Crossan's Cross Gospel against the majority view of dependence) and the date of its one substantial manuscript (estimates from the late fifth to the ninth century) are all argued. Crossan's own case was read only in his statement of it, not in its detailed arguments. Joel Marcus's first-century proposal is known here only through Jacobus.
3. The Epistle of the Apostles. Dated to 120-180 by the range Hills reports. Charles Hill's 1999 argument for an early date, and Francis Watson's 2020 study, were not read.
4. Ignatius' source. Eusebius did not know where Ignatius found the "bodiless demon" saying; Origen and Jerome name other books. If the saying came from a tradition independent of Luke, the touching motif had two written forms by about 110; if from a rewriting of Luke, one. Nothing read here decides it. Ignatius' own date is disputed within and beyond Trajan's reign.
5. No measured rate. The seven cases are selected examples of what can happen. No sample exists from which to calculate how often a marvellous detail attaches to a remembered event within a given time, or how often participants succeed in correcting one. Memory studies measure something narrower: how individuals retain the facts of a public event and their own circumstances over ten years (Objection 1). Their results fit this article's line between core and detail, but they are not a rate of legend formation in a community.
6. Luke 24:40. The committee's 1971 rating {D} is reported. Its rating in the 1994 second edition of Metzger's commentary was not read; every copy reachable for this article was the first edition.
7. The physicality debate in Paul. This article relies on the sibling Paul article for the meaning of "spiritual body." If Paul meant a body made of spirit, the gap between Paul and Luke widens; if he meant a transformed physical body, it narrows.
8. What counts as the core. Sherwin-White's argument protects a "hard core," but which elements belong to it is the point in dispute. Nothing in his method decides whether the empty tomb or a group meal belongs to the core of the resurrection tradition.
9. Unread sources. Erik Manning's public argument against a Mark-to-John progression, a social-media post, could not be retrieved; the progression is tested here against Carrier's statement of it instead. James D. G. Dunn's Jesus and the Spirit is cited only through Allison. Carrier's further examples (Herodotus' account of Delphi defending itself, 8.35-39) were not checked.

Verdict

The resurrection narratives demonstrably changed in the telling where a source survives.  Assuming Matthew and Luke used Mark, they reworded, added and left out, for apologetic, dramatic and geographical reasons, and the second century carried several of the same lines further.  What the comparison shows is change;  whether a given change is invention needs further evidence.  That evidence exists for Matthew's guard and descending angel, which are probably secondary.  It is missing for the appearance to the women, Luke's Jerusalem scenes and the proof scenes, which are undecided, though their later first attestation is a real reason for caution about their details.

On the transmitted Pauline text, appearance claims already exist by the mid-50s, with an older core, so an origin only decades later cannot explain them.  Price challenges the list's Pauline authorship, but his proposed insertion does not remove Paul's own claim to have seen Jesus in 9:1.  Galatians confirms contacts with Cephas and James, not the content of their conversations.  Early development remains possible within their own accounts and in reports about groups such as the five hundred.  The increase in physical demonstration from Mark through Luke is observable; deciding how much reflects inherited memory, selection or invention requires more than the sequence alone.

Between those findings lies the question that matters for natural explanations.  The scenes of conversation, touching and eating are written decades after the events, show apologetic framing, and belong to a motif that later writers kept developing.  They may preserve early memory, as Allison and Wright in different ways believe.  The available evidence cannot show that they do, or that they do not.  Sherwin-White's two-generation argument, read as he wrote it, supports an early core, including the claim that Cephas and others saw Jesus.  It says nothing in favour of these details, and it does not show that the core claim was true.  That question is not decided by how fast legends grow.

Sources