Volume 44 - Issue 1
Finessing Independent Attestation: A Study in Interdisciplinary Biblical Criticismby Lydia McGrew
The criterion of multiple attestation is crucial in biblical studies, particularly in historical Jesus studies. While doubts are often conceded about the historicity of a singly-attested incident, when there is reason to believe that an event has been attested in multiple independent sources it is often accepted despite a hesitation to affirm the strong historical reliability of the individual documents.
A problem arises, however, when “independence” is either not defined clearly or not understood with sufficient rigor. Recent work in formal epistemology shows that the correct understanding of multiple independent attestation to an event involves independence of the testimonies given the negation of the hypothesis in question. For example, we should be able to argue that the accounts that affirm some core event would have had to affirm that core in some separate, independent fashion even if the core event did not occur. Moreover, the point of multiple attestation to events should be the argument that the accounts indicate independent access to what really happened, not merely independent access to some common tradition that may or may not correctly represent reality.
This realization makes it important not to confuse literary independence among accounts (e.g., that one account was not copied from another) with the relevant kind of causal and probabilistic independence. The former is helpful but not sufficient. If, for example, two accounts plausibly were derived from the same oral tradition or the same people, then, even if they were not copied from each other, they are not independent in the relevant sense for purposes of confirming the incident. This issue becomes acute when biblical scholars attempt to affirm the perspective of critical scholarship by conceding, even if only for the sake of the argument, that the differences between or among accounts are the result of imagination or invention on the part of the authors. Once that move has been made, those differences cannot confirm the relevant kind of independence.
I will discuss three examples taken from New Testament scholars in which independent, multiple attestation has been alleged without sufficient nuance. These problems in biblical studies illustrate two points. First, since multiple attestation cannot be invoked in the way that some biblical scholars have been in the habit of doing, there will be a need for a less concessive approach to the robust individual reliability of biblical documents if the incidents within them are to be adequately confirmed. Second, probability theory (or commonsense intuition that tracks probability-theoretic insights) should be permitted to inform biblical studies.
1. If the Theory Were False, Would the Accounts Be Dependent or Independent?
Suppose that two witnesses testify that they have seen a bank robbery. Their descriptions of the robbery and the robber have quite a bit of overlap. When detectives make use of their testimonies, they want to be able to argue that they are independent witnesses, so that they can say that they have two reports of the robbery rather than, in effect, only one. But what does “independent” mean in this context?
One explanation of “independent” that is often given is that the witnesses have not colluded or that one has not copied from one another. Either of those scenarios would certainly violate independence, but to say that those scenarios are false does not constitute a general account of the relevant type of independence.
Recently published probabilistic analysis shows that the type of independence needed, perhaps surprisingly, relates to the negation of the hypothesis.1 Suppose that the salient hypothesis is “Jones robbed the bank,” where Jones is a person who closely meets the specific description given by the witnesses. Now suppose that it were false that Jones robbed the bank. To say that the witness testimonies are independent in the relevant sense for confirmation is to say that, given that Jones did not rob the bank and that we have one witness testimony describing the robber, we would have no additional reason to expect the specific content of the other witness’s testimony. One witness’s testimony would not help us to infer the content of the other witness’s testimony given that Jones didn’t rob the bank. If Jones didn’t rob the bank, and if the witnesses are independent in the relevant sense, then in the scenario just described they have both given an excellent and similar description of Jones by sheer coincidence. It is the implausibility of this coincidence that makes their agreement so powerful in the case against Jones.
Obviously, if one person was not even present at the time and copied his report from the other, this copying ruins that sort of independence. Even if Jones were completely innocent, given this theory we would expect the second report to resemble the first report. But that is not the only way for independence to fail. Suppose, instead, that both witnesses got their account of the appearance of the robber from the same third person. Then we would also expect their reports to resemble each other, even given Jones’s innocence. The accuracy of the description, in that case, depends solely upon the reliability of the single source that lies behind both of the (supposed) witnesses we know about. In both of these cases (both copying from one another and copying from a common source) there are not really two sources of information but functionally only one. That is why these both illustrate a failure of the relevant kind of independence.
These are simplified scenarios. One could have partial dependence if both witnesses were really present but one overheard the testimony of the other and subconsciously manufactured some portion of his apparent memories because of that influence, despite trying to be purely factual about what he saw. Again, such partial dependence can be understood best by thinking about the negation of “Jones robbed the bank.” If Witness B subconsciously manufactured a memory of a mole on the side of the robber’s face because he overheard Witness A, then even if Jones (who has such a mole) is completely innocent, both witnesses would be expected to attest to a mole on the robber’s face. One witness’s testimony on that point would, ex hypothesi, give a clue to what the other witness would say.
Those evaluating different witnesses or written sources can use various arguments to try to support independence as opposed to dependence. They can argue that the different authors or witnesses were isolated and would have had no way to know of each other’s statements. They can argue that there was no time or opportunity for a common traditional source to develop from which both could be drawing. They can argue on other grounds that the witnesses really were present, that they have been individually found to be right on other points, that their character as truth-tellers who do not collude is vouched for, and so forth. More interestingly, one can examine the specific contents of the reports themselves for signs of the relevant type of independence. Perhaps the kinds of details on which the reports differ are, as experience has shown, what we would expect from the casual variation found in the testimony of witnesses who are trying to tell the truth. Perhaps the kinds of differences are such as might naturally arise from a different physical vantage point or from entering the scene at different times. One very fruitful type of variation is what is known as an undesigned coincidence; in this type of case, the testimonies vary in such a way that the differences fit together explanatorily—a good indication of independent access to the events themselves.2
Philosophers of science have known for a long time that varied evidence is more valuable to confirmation than evidence that is identical. This is why witnesses who show some variation in their reports are preferable to those that are too similar. William Paley noted long ago that “the usual character of human testimony is substantial truth under circumstantial variety,”3 a fact that also has applicability in courts of law. The understanding of independence as independence given the negation has helped, in recent years, to spell out the reason for the value of varied evidence.4
A more thorough understanding of independence helps us to avoid pitfalls in claiming independent attestation in all historical endeavors, including historical Jesus studies and historical apologetics.
2. The Empty Tomb and 1 Corinthians 15
William Lane Craig has argued in multiple places that Paul’s “creed” in 1 Corinthians 15:3–8 constitutes an independent attestation to Jesus’s empty tomb.
Jesus’ empty tomb is also mentioned in the early sermons independently preserved in the Acts of the Apostles (2.29; 13.36), and it’s implied by the very old tradition handed on by Paul in his first letter to the Corinthian church (I Cor. 15.4). Thus, we have multiple early attestation of the fact of the empty tomb in at least four independent sources.5
We have seen that in 1 Corinthians 15:3–5 Paul quotes from an extremely early tradition that refers to Christ’s burial and resurrection. Although the empty tomb is not explicitly mentioned, a comparison of the four-line formula with the Gospel narratives on the one hand and the sermons in Acts on the other reveals that the third line is, in fact, a summary of the empty tomb narrative…. We have, then, extraordinarily early, independent evidence for the fact of Jesus’ empty tomb.6
Elsewhere Craig says that this independent attestation is “both in the pre-Markan passion story and also in the pre-Pauline formula quoted in 1 Corinthians 15.”7
If we assume that the Gospel of Mark had not yet been written by the time that Paul received the information expressed in these verses of 1 Corinthians, and perhaps not even by the time that 1 Corinthians was written, it is true to say that this creedal affirmation is not literarily dependent upon Mark. If, moreover, we speak of a written “pre-Markan passion narrative,” we might even argue that Paul’s statements in 1 Corinthians 15 are not dependent upon such a written source, though that would be somewhat harder to support. In general, it is plausible that what Paul says in 1 Corinthians 15 is not literarily dependent upon any of the actual Gospel narratives.
But that is not all that is needed for the relevant kind of independence for the apologist’s purposes. It’s important to remember that Craig cannot merely be arguing that 1 Corinthians 15 provides independent attestation that this is what the apostles affirmed. Rather, this passage is supposed to be an independent attestation to the fact of the empty tomb. If it is merely independent attestation to the disciples’ affirmation of the resurrection (and thus implicitly to the empty tomb), then our knowledge of the truth or falsehood of the empty tomb claim depends upon the reliability of the disciples that are the common source behind both the Gospel accounts and Paul’s affirmation in 1 Corinthians 15.
Nor is there any strong reason to assume that the affirmation in Mark (or a “pre-Markan passion story”) comes from a definitely distinct apostle from the affirmation on which Paul is depending in 1 Corinthians 15. We have no special reason to think, for example, that the account in Mark comes from Peter but the creedal affirmation in 1 Corinthians 15 comes from James instead.
Presumably Paul received information from the Christian community in Damascus (Acts 9) and eventually from the apostles in Jerusalem (Gal 1:18). In the end, Paul’s information about the empty tomb was probably received directly or indirectly from the teaching of the apostles. Indeed, Paul’s connection with the apostles is sometimes used to argue that the apostles themselves believed in a physical resurrection since Paul apparently did, and Paul’s gospel was taken to be in harmony with theirs.8 Paul does not indicate that he spoke to one of the women who was present at the empty tomb and who is not a source behind any of the Gospel accounts. Nor is there sufficient detail in Paul’s account to support such a conclusion, as there might be if he had given his own version of the discovery of the tomb. As Craig acknowledges, the empty tomb is not even mentioned explicitly by Paul.
The Gospel accounts of the empty tomb, in contrast, do provide such varying details. They mention different names of women present, which Richard Bauckham has argued may indicate differing human sources used by the authors.9 They give somewhat different accounts of the words of the angels. John’s account gives the perspective of Mary Magdalene very particularly, whereas Matthew seems to follow some others who were with her. Only Luke mentions Joanna (Luke 24:10). These are the kinds of details that help us to make an argument that the authors of the accounts had independent lines of access to the facts attested, not merely to the existence of apostolic teaching that included the affirmation of an empty tomb.10
In short, if the same people lie behind Paul’s affirmations in 1 Corinthians 15 and some or all of the Gospel accounts of the empty tomb (such as Mark’s account), then Paul’s implication of the empty tomb is not independent of those accounts in the relevant sense. It goes back to a common source, though plausibly a human rather than a written source. And nothing in the brief verses in 1 Corinthians 15 permits us to argue otherwise.
The attempt to use 1 Corinthians 15 as an independent attestation to the empty tomb may spring from a confusion between literary independence and the relevant kind of causal independent access to the facts.11 If the tomb was not empty (that is, given the negation of the hypothesis under consideration), if Paul got his understanding of the empty tomb from the apostles who got their information from the women, and if the women’s (in that case inaccurate) story lies behind (say) the Markan account, then we would expect both Paul and Mark to imply or state that the tomb was empty.
In contrast, if Luke’s account of the empty tomb derives in whole or in part from Woman A and Mark’s account derives in whole or in part from Woman B, and if A and B did not agree to collude upon a false tale or copy their claim about the empty tomb from some other source, then Luke’s and Mark’s accounts have a claim to be independent in the relevant sense. Moreover, the differing details of the accounts give us some purchase upon the probability that this is the case. We can attempt to argue that variations of the kind that we find are not likely to arise in accounts that are the result of collusion, copying from a common source, or imaginative variation upon the core of a common source. That is the sort of argument that needs to be made to claim multiple attestation, but it is not available in the case of the creed in 1 Corinthians.
3. The Infancy Narratives as “Midrash”
In a written debate with Bart Ehrman, Michael Licona brings up an hypothesis about the origin of the infancy narratives.
Bart provides the example of the differences between the infancy narratives in Matthew and Luke. In my opinion, those narratives include the most difficult and profound differences in the Gospels…. Here I must acknowledge that I don’t know what’s going on and have no detailed explanations for these differences. I think one can provide some plausible solutions. But I admit they are speculative…. However—even though, as I say, I don’t know what’s going on here to cause the differences—let’s just speculate for a moment and consider the following scenario. Matthew and Luke both agree that a Jewish virgin named Mary who was engaged to a Jewish man named Joseph gave birth to Jesus in Bethlehem. The early Christians all knew this much. However, little else was remembered about this event. So, Matthew and Luke added details to their account to create a more interesting narrative of Jesus’s birth, a type of midrash. I’m not saying this is what Matthew and Luke did. I don’t know what’s going on with the infancy narratives. However, if this occurred, we would have to take the matter of genre—midrash—into consideration and recognize that the historicity of the details outside of the story’s core would be questionable, while the core itself could stand. After all, with such differences between the accounts in Matthew and Luke, one could reasonably argue that the core is attested by multiple independent sources.12
The precise degree of credibility Licona is giving to this suggestion is a little difficult to assess. He says that he can think of “solutions” that are both “plausible” and “speculative” to what he implies is a serious difficulty with harmonizing the infancy narratives, though he does not say in detail why he regards these differences as so particularly difficult. He then invites the reader to “speculate for a moment” and produces only one hypothesis, as just quoted. Though he carefully stipulates that he is not saying that this is what Matthew and Luke did, it is not unreasonable to conclude that this is one of the plausible though speculative hypotheses to which he has just referred.
But I need not decide just how credible Licona considers this theory to be in order to note a problem with his use of “attested by multiple independent sources.” What Licona calls the “core” of the infancy narratives is understood as the overlap in express assertion or clear implication between them. This would be the conjunction of propositions such as, “Jesus was born in Bethlehem,” “The mother of Jesus was a Jewish virgin named Mary,” “Mary was engaged to a Jewish man named Joseph,” and so forth.
Licona’s use of phrases using success verbs such as “the early Christians all knew this much” and “little else was remembered” cannot be taken to mean by definition that this overlapping content is true, for that would be question-begging. We are trying to assess how Luke’s and Matthew’s narratives provide evidence for the truth of that overlap and whether, on Licona’s speculation, they constitute “multiple independent sources.” So the “core” cannot be assumed to be true as part of the theory put forward. Hence, “knew” and “remembered” should be taken to mean something like “commonly believed.” With that adjustment to avoid question-begging in favor of the truth of the overlap, Licona’s speculation amounts to the claim that Luke and Matthew both had access to a common, accepted tradition at the time they were writing their Gospels that included these propositions and “little else.”
On the hypothesis Licona raises, Luke and Matthew then took this commonly accepted set of traditions about Jesus’s birth and imaginatively embellished them to “create a more interesting narrative”; he refers to this embellishment as “midrash.”13 The differences between the two narratives, then, are the result of Luke’s and Matthew’s imaginations. They are not the result of Luke’s and Matthew’s having independent access to the events that actually happened, as would occur in a case where one witness saw or noticed the bank robber’s face while another saw or noticed his gait. If, in contrast, Luke had contact with the family of Mary (a suggestion made by Richard Bauckham14) whereas Matthew had contact with someone who had heard Joseph’s portions of the story, this could constitute genuinely independent access to the events surrounding Jesus’s birth. And if we leave open the possibility that the differences between the narratives are due to that sort of causal independence, with one source reporting some things that the other does not, then the stories might constitute multiple and relevantly independent access to the overlap.
But that is not the conjecture. The conjecture instead is that there was what we might call an epistemic “node” in the form of the accepted tradition and that the two evangelists diverged from that node by taking the content of tradition and elaborating upon it. In that case, there are not two independent sources attesting to the truth of the content of the overlap but rather only one source—the “node” representing the beliefs widely held, to which they both had access. The truth or falsity of the content of the overlap thus comes down to the reliability or otherwise of that tradition.
To return to the probabilistic analysis given earlier, consider the question of independence given the negation of the hypothesis in question. Suppose that some (or all) of the propositions in the overlap between Luke and Matthew were false. On the theory Licona puts forward, can we then say that it is a remarkable coincidence that Luke and Matthew agree on the overlap? Not at all. For the hypothesis in question is that they both had access to that overlapping content in the form of tradition believed by Christians. If the Christians were wrong about that content, Luke and Matthew would not be agreeing on it by astonishing happenstance. Rather, ex hypothesi, they would be agreeing upon it because they were both getting it from the common (partially or wholly incorrect) tradition.
It would be useful if the differences between the narratives could help to solve this problem, and Licona implies that they are useful. When arguing for relevant independence, however, one needs to use the differences between narratives to argue that the different accounts go back, perhaps indirectly, to different sources with some claim to have known what really happened. I illustrated this earlier when discussing the possibility that different Gospel accounts of the discovery of the empty tomb might reflect the perspectives of different women who were there at the time. I illustrated that procedure in this section by raising the possibility that one of the Gospel authors had access to an account that came (perhaps indirectly) from Mary while the other had access to an account that came (perhaps indirectly) from Joseph and that their differences reflected this fact. This is an illustration in practice of the way that varied evidence is helpful to confirmation.
But Licona has blocked that sort of appeal to differences, given the hypothesis in question, by speculating that the differences are not the result of separate access to the real events (whatever those events might be) but rather of separate creative imaginations on the part of the authors in crafting their “midrash” narratives. This would be like having two authors read the same common historical source and then write up partially invented accounts that have some overlap (borrowed from the common source) but are different simply because they chose to invent different added material. There is, in that case, only one source behind both of them. I am not saying that Licona means that “what was remembered” about Jesus’s birth was written down by the time that Luke and Matthew wrote, but the same point applies to a common oral tradition on which each author expands. Their differences do not attest to independent access to what happened.
In short, independence of the authors’ creative personal imaginations is the wrong kind of independence for purposes of multiple attestation to events.
4. “I Thirst” and Synoptic Tradition
Both Daniel Wallace and, following him, Michael Licona have made a rather surprising suggestion about the saying “I thirst” from the cross. Here is Licona’s treatment:
In Jesus’s next-to-last statement on the cross, Mark // Matthew have Jesus say, “My God! My God! Why have you forsaken me?” But John appears to substitute “I am thirsty.” In Jesus’s final statement on the cross, Mark // Matthew report that Jesus then cried out loudly and died; Luke reports that Jesus cried out loudly, “Father, into your hands I entrust my spirit,” then died; and John reports that Jesus said, “It is finished,” then died…. Virtually all specialists of John’s Gospel acknowledge that the evangelist often adapted the traditions about Jesus. These two utterances of Jesus may be an instance when we can observe the extent to which John redacted existing tradition. For the next-to-last logion, it appears that John has redacted “My God! My God! Why have you forsaken me?” (Mark // Matthew) to say, “I am thirsty.” Daniel Wallace proposes that since every occurrence of “thirst” in John carries the meaning of being devoid of God’s Spirit, the evangelist has reworked what Jesus said “into an entirely different form.” It is “a dynamic equivalent transformation” of what we read in Mark // Matthew. Accordingly, in John, Jesus is stating that God has abandoned him. In Mark 15:34, Jesus quotes Ps. 22:1: “My God! My God! Why have you forsaken me?” Thus, John can write, “Knowing that everything had now been accomplished, in order that the Scripture may be fulfilled…, Jesus said, “I am thirsty” (John 19:28, emphasis added). John has redacted Jesus’s words but has retained their meaning.15
Licona addresses independence in an endnote:
If John is independent of the Synoptic tradition as many scholars hold, we may have multiple independent sources pertaining to this logion with Mark’s version being closer to what Jesus may have uttered.16
Following a longer discussion of his theory that “I thirst” is a “dynamic equivalent transformation” of “My God, why have you forsaken me?” and other suggestions in the same vein, Wallace makes a similar comment about independence:
In my own thinking, the thesis put forth here gives fresh impetus to the importance of that question [of John’s relation to the Synoptics], for if John is independent of the Synoptics, then his dominical transmogrifications still need to be counted as yet another vote in multiple attestation.17
Wallace admits that “on the surface, the two utterances do not look at all alike”18 and calls his suggestions “radical repackaging of the dominical material so that it no longer looks like the original saying.”19 He believes, however, that John’s recorded saying “I thirst” bears the same meaning as “My God, why have you forsaken me?” at a deep, theological level, and he provides several pages of discussion to this effect. I find his arguments for the claim that John has made such a repackaging quite unconvincing, but in this essay I will confine myself to discussing the claim made by both Wallace and Licona concerning multiple independent attestation and the implications of their theories for the ability to make that claim.
A major issue that makes this claim of multiple attestation unusual is the alleged radical transformation itself, and there will be no way to avoid discussing the way that that aspect of the theory intersects with the statement that John attests to the same thing that the Synoptic Gospels attest to. Similar considerations apply to the Licona/Wallace suggestion concerning “It is finished.” In these cases both the use of “attestation” and the use of “independent” are problematic, and I will deal with both.
A further potential ambiguity may be present concerning the term “tradition” as used by Licona and Wallace. Licona’s usage appears clear:
Virtually all specialists of John’s Gospel acknowledge that the evangelist often adapted the traditions about Jesus. These two utterances of Jesus may be an instance when we can observe the extent to which John redacted existing tradition.20
The phrase “the traditions about Jesus” leads one to assume that, by “traditions” and “existing tradition” Licona does not mean to refer to what actually happened but rather means to refer to what was believed later on about what actually happened, as he does in the case of the infancy narratives.
The doors that were closed on this issue with Dodd’s Historical Tradition in the Fourth Gospel have been slowly pried open in the last two decades once again…. Thus, even though he is clearly giving a theological interpretation of the life of Jesus, a careful examination of the data gives sufficient evidence that John’s representation is thoroughly grounded in the tradition.21
The reference to C. H. Dodd brings to mind a useful terminological point made by D. A. Carson—namely, that when Dodd says that John drew something out of the tradition, he means that it is historical.22 Sometimes the term “tradition” functions in New Testament scholarship in such a way that its meaning wavers between referring to reality itself and referring to the Christians’ beliefs or teachings about reality. So it is possible that Wallace’s statement that “John’s representation is thoroughly grounded in the tradition” is meant to state that John is attesting to the events rather than only to the existence and content of a tradition (in the narrower sense) similar to that found in the Synoptic Gospels. However, evidence that Wallace as well as Licona is using “tradition” here to mean beliefs about what Jesus said (rather than simply reality) is found in this statement: “I take the last two words in John as this evangelist’s version of two of the utterances found in the synoptic tradition.”23
We must ask what can be meant in these theories when it is said that John and the Synoptics constitute multiple attestation. Whether we take this alleged attestation by John to be to the contents of a tradition about Jesus’s words or to the content of Jesus’s actual words, in either case there is a rather serious problem with claiming that John is attesting to the same thing by “I thirst” that the Synoptic Gospels attest to by, “My God, why have you forsaken me?” Indeed, a question that immediately springs to mind upon reading Wallace’s statement that John will need to be counted as “another vote in multiple attestation” is, “Attestation to what?” Wallace does not say. Licona says that it would be multiple attestation “pertaining to this logion,” not addressing the fact that it is unclear what “this logion” is, given that the facial content of the two sayings is entirely different. Are Licona and Wallace saying that John actually attests that Jesus said (or attests to the existence of a tradition that Jesus said), “My God, why have you forsaken me?”
Again, the discipline of epistemology is helpful here, for epistemic analysis tells us that a crucial part of discussing the confirmational impact of testimony is getting quite clear on what the hypothesis in question is.24 If we want to ask about the epistemic impact of some evidence, we need to have clearly in view the proposition for which (or against which) it is supposed to be evidence. Neither Wallace nor Licona addresses this question. But they both indicate that the earlier tradition (or the reality) was that Jesus said something like, “My God, why have you forsaken me?” instead of “I thirst.” So let us take it that the hypothesis is “Jesus said something recognizably, facially similar to, ‘My God, why have you forsaken me?’ on the cross.” If we wish to loosen up the hypothesis a bit more, we could make it something like, “Jesus made a lament on the cross concerning a sense of spiritual abandonment or emptiness.” Or we could make the hypothesis something parallel to one of these for the contents of tradition: “There existed a tradition according to which…,” and so on.
Let us ask, then, whether John’s record that Jesus said, “I thirst” is an attestation to any of these hypotheses. In particular, we must ask whether it is an attestation to any of these hypotheses given that 1) the event of Jesus’ saying something recognizable as “I thirst” (either these words or something like, “Give me to drink,” etc.) did not occur, and 2) John has radically transformed whatever did occur or the tradition he had heard about what did occur into a form that no longer looks at all like the original saying.
Without those two premises, one might gain some slight indirect confirmation for one or more of the suggested hypotheses. For example, if Jesus really did express physical thirst on the cross as apparently recorded in John’s narrative, perhaps he would also not have been averse to expressing spiritual lament. There might be some weak force to that argument. But that, of course, is an argument that runs “through” the supposition that John is attesting fairly literally to what actually occurred, which is denied by the Wallace/Licona theory.25 Once one has hypothesized that John has engaged in such a radical transformation, how can his record give added confirmation to the saying recorded in the Synoptics? The mere fact that Wallace and Licona conjecture that Jesus did not utter a cry of thirst and that John is metaphorically alluding to a (facially) completely different utterance does not produce confirmation from John of the completely different utterance.
A thought experiment is helpful here: If we knew only what John records and if we were told that it is a radical transformation of reality or tradition into a form that does not look on its face at all like the original, what would we thereby be able to affirm or even reasonably suggest about the content of the tradition or reality? Perhaps that Jesus spoke at all on the cross. Perhaps that he expressed some sort of anguish, though a wordless cry or some Aramaic interjection expressing undifferentiated pain would satisfy that description. It simply is incorrect to say that, on the theory in question, John’s Gospel attests to any specific content in the tradition, much less to the specific content that is also attested to by “My God, why have you forsaken me?” in the Synoptics.
Another thought experiment may make the same point clear: Suppose that we were told that the story of the raising of Lazarus was a radical transformation, into a form that looks completely different, of something that Jesus did or said or a tradition about what Jesus did or said. To what could we then say that it is an attestation? It certainly wouldn’t be an attestation to one particular story that we find in the Synoptic Gospels. It might have been inspired by some event or tradition that we don’t retain otherwise. Or it might have been inspired by a healing rather than a resurrection. Or it might have been inspired by either the raising of the widow’s son at Nain or by the raising of Jairus’s daughter. Or it might be a dramatization of the general proposition that Jesus raised the dead and healed the sick. Such a claim would make it impossible to say what John’s record “attests” to at that particular point and would not single out any particular bit of synoptic content for John to “attest to” independently.
That is the problem with “attestation.” What about independence? Here once again we run into the same issue discussed in the previous sections. Even if John is literarily independent of the Synoptic Gospels—that is, he did not have them available to him or did not literarily rely on them—it does not follow that his account represents independent access to events. Ultimately, of course, the hope is that multiple independent attestation will give us a stronger fix on what the historical Jesus actually said and did. But if John is at this point merely (radically) modifying a tradition to the effect that Jesus said, “My God, why have you forsaken me?” which the synoptic authors also knew of, then this conjecture does not tell us that he has independent access or provides independent testimony to what Jesus actually said on the cross.
This issue would be relevant even if John recognizably recorded, “My God, why have you forsaken me?” as, in fact, he does not. This problem exists therefore in addition to the problem with attestation just discussed. Even if John recorded a saying facially similar to “My God, why have you forsaken me?” one would have to argue for the relevant kind of independence from the Synoptics. If, ex hypothesi, he got his knowledge of that saying from a “synoptic tradition” (Wallace’s phrase) or from “the traditions about Jesus” (Licona’s phrase), and if the synoptic authors also knew those same traditions, then the relevant type of independence does not hold. On that hypothesis, John is not helpful (in addition to the Synoptics) concerning the conclusion that Jesus uttered that saying or something visibly like it. Here we have the “single node” problem discussed in a previous section.
Suppose that Licona and Wallace were to hypothesize instead that John is based upon separate access to the events themselves and is radically modifying the saying, “My God, why have you forsaken me?” as he knew of it in some way other than by a tradition known to the synoptic authors. Licona (perhaps differing a bit from Wallace at this point) actually says that John was “independent of the Synoptic tradition,” which may tend in this direction—implying that John had an entirely separate tradition as his source.
But at this point, the problem of the radical transformation returns. How could Licona and/or Wallace go about arguing for a causal hypothesis of independent access to the events? As in the case of the infancy narratives, one would like to be able to use differences between the accounts to argue that they really represent separate access to the events—to what Jesus said and what occurred in his Passion. This is what one would do if one were arguing for the truth of some clear point of overlap, using other points of divergence to show independence. One might point to the fact that John records “I thirst” while the Synoptics do not and that they record sayings he doesn’t have. One could then use these facts to support the conclusion that (plausibly) John and the Synoptics are based upon different human lines of information about the Passion. Perhaps either John or someone present at the cross heard Jesus say, “I thirst,” while the human sources lying behind the Synoptic Gospels either did not hear or did not report this particular saying and reported others instead. In this way one would be using the differences between the accounts to argue for plausible different access to events, perhaps from different physical vantage points (one person might have been closer to the cross than another, for example), or for different reports from witnesses with direct access based upon differential memory or saliency. Then one could use this independence to support something in the Passion narrative on which John and the Synoptics are clearly agreeing.
But that path is not open to Licona and Wallace. Their hypothesis is that John is not actually attesting to a different saying at all and that the different appearance of the saying in his Gospel is due not to the fact that he and the Synoptics report different aspects of reality but instead to processes taking place in the evangelist’s own mind—his decision to make a theological metaphor in his narrative instead of recording something recognizably like what would have been literally heard by bystanders.
Wallace’s and Licona’s theories make it difficult even to use other differences between John’s and the Synoptics’ crucifixion stories to make an argument for true independence. The saying, “It is finished” in John, like “I thirst,” is allegedly not the result of differential access, memory, or decision to report real events but rather the result of John’s personal decision to make a radical transformation.26 At that point it would seem arbitrary to argue that, since only John records that (e.g.) the soldiers cast lots because one of Jesus’s garments was seamless (John 19:23–24), he probably has some independent access to real events and reports these varied details in an ordinary historical fashion. Perhaps the seamless garment was also a Johannine addition, crafted to create a rationale for the fulfillment of prophecy when the soldiers cast lots.
Wallace suggests that there are still more places where John, though appearing to record a different saying or even discourse, is actually making a radical transformation of something we find recorded in the Synoptics. For example, he suggests that the passage in which Jesus assures his disciples that in his Father’s house are many dwelling places (John 14:1–3) may be a Johannine transformation of the tradition of the Olivet Discourse.27
Licona has conjectured additional places for John to have altered material or added embellishments. For example, he theorizes that John may have added the incident in which Jesus breathes on the disciples and says, “Receive the Holy Spirit” in order to “allude to the event at Pentecost.”28 This would not even be an alleged transformation of some other words of Jesus on earth. He also suggests that John may have “relocated” the appearance to Mary Magdalene.29 But if the appearance to Mary Magdalene really took place along with the appearance to the women recorded in Matthew 28, then the scene in John 20 must include quite a lot of invented dialogue and circumstances. The entire scene between Jesus and Mary Magdalene in John 20 presupposes that she did not already know he was risen and that she is staying near the tomb rather than running away from it. This is why she thinks that Jesus is the gardener and why she is able to look into the tomb while weeping and see the angels.
The point is not that Licona is alone in his various theories. (E.g., Craig Keener also seems to lean in the direction of thinking that John 20:22 is not historical.30) The point is rather that, once unique Johannine material of different types is categorized repeatedly as a result not of differential access to actual events but rather as the evangelist’s personal transformation or addition, John’s unique material can no longer consistently be used to argue for the relevant kind of independence in John’s narrative. John just isn’t being portrayed as the kind of reporter whose unique material is of that sort. For all that such theories permit us to argue, he might be (at the points where there is some overlap) adding to a tradition that he has in common with the Synoptics. As with the infancy stories, so here: Independence of the author’s imagination, transformation, or embellishment is the wrong kind of independence for confirmational purposes. Therefore, even when John does recognizably attest to the same content as the Synoptic Gospels, Licona’s and Wallace’s approach does not retain the epistemic resources to make a convincing case that he does so independently.
5. Broader Implications
A heavy reliance upon multiple attestation has tended to go hand in hand with what one might term apologetic minimalism. This approach is well illustrated by the quotation given earlier from Licona concerning the infancy narratives. There Licona insists that the overlap between the infancy narratives “could stand” even if the non-overlapping material were invented to make a more interesting story. This is a particularly striking claim since there is quite a bit of non-overlapping material between Luke’s and Matthew’s infancy narratives, including the star, the slaughter of the innocents, the shepherds, and so forth. Licona bases his methodological point on the alleged multiple independent attestation to the overlap, concluding, “After all, with such differences between the accounts in Matthew and Luke, one could reasonably argue that the core is attested by multiple independent sources.”31
Claims of multiple attestation, then, serve the purpose of seeming to make it unnecessary to argue for the individual reliability of a particular account. They even are taken to mean that one could concede for the sake of the argument that the account contains a significant amount of non-factual material. William Lane Craig makes such a point in the introduction to the 3rd edition of Reasonable Faith:
Keeping the book at approximately the same length was made possible by the deletion of the chapter on the historical reliability of the New Testament, a chapter which a former editor had insisted, despite my protestations, be inserted into the second edition. The inclusion of this chapter (itself a solid piece of work written at my invitation by Craig Blomberg) perpetuated the misimpression, all too common among evangelicals, that a historical case for Jesus’ radical self-understanding and resurrection depends upon showing that the Gospels are generally reliable historical documents. The overriding lesson of two centuries of biblical criticism is that such an assumption is false. Even documents which are generally unreliable may contain valuable historical nuggets, and it will be the historian’s task to mine these documents in order to discover them. The Christian apologist seeking to establish, for example, the historicity of Jesus’ empty tomb need not and should not be saddled with the task of first showing that the Gospels are, in general, historically reliable documents. You may be wondering how it can be shown that the Gospel accounts of the discovery of Jesus empty tomb can be shown to be, in their core, historically reliable without first showing that the Gospels are, in general, historically trustworthy. Read chapter 8 to find out.32
Multiple attestation is then used in Chapter 8 as one of the ways in which the empty tomb can be established while setting aside the question of whether the Gospels are strongly historically reliable.
If multiple independent attestation is harder to establish than has been previously thought, those who wish to argue for the truth of some “core” or “overlap” in the Gospel narratives may have to shift their emphasis and argue for the reliability of the documents overall rather than merely for the truth of selected passages “mined out of” a document that might, for all we are willing to maintain, be generally unreliable.33 In other words, scholars might need to reconsider the statements made by Craig in the above quotation. Perhaps we need to return to what Craig considers a “misimpression” concerning the interrelated nature of strong reliability and the defense of Christianity.34
Moreover, the analysis here has shown that a certain kind of preemptive concession—namely, hypothesizing explicitly that unique material in a document is the result of the author’s imagination or radical transformation—blocks the road to supporting the relevant kind of independent attestation. At a minimum, the insights of probability theory discussed here show that New Testament scholars who wish to argue for the historical truth of some passage should not build into their theories these sorts of independence-blocking aspects. Nor should they say or imply that multiple independent attestation means that issues like provenance and factual reliability are unimportant. They need to be able to argue at least probabilistically that varied details indicate varied access to the events. Independent attestation must be established by way of a process that, at the same time, tends to support the thesis that the individual documents come from those who were knowledgeable about the facts. Multiple independent attestation does not replace strong, whole-document reliability and ultimate provenance in eyewitnesses; rather, the categories are probabilistically entangled. One cannot concede, even for the sake of argument, that such access and accuracy are not the case and simultaneously try to argue for independent attestation.
Once it is thoroughly understood that significant literary independence is a necessary but not a sufficient condition for multiple attestation, and once it is understood that variation of authors’ imagination or embellishment cannot help to establish independent attestation, New Testament scholars and apologists can see what paths they might explore if the historical claims of Christianity are indeed defensible. The lesson to be drawn is not that the Gospels do not contain real instances of multiple independent attestation. I would argue that they contain a great many, though that argument lies beyond the scope of this paper. The point, rather, is that multiple attestation must be alleged and supported more carefully.
This conclusion can be a constructive one for the defender of Christianity. An older generation of biblical scholars, from William Paley35 to J. B. Lightfoot36 to (in somewhat more recent decades) Colin Hemer37 and Leon Morris,38 believed that it was possible to support the robust reliability of whole documents rather than confining oneself to gleaning a far more limited number of facts out of documents. If conclusions like those drawn in our own time by D. A. Carson,39 Craig Blomberg,40 and Richard Bauckham41 concerning eyewitness testimony and document reliability are a necessity for Christian apologetics rather than a luxury, evangelical scholars should not despair but would do well to see if there is even more evidence of the same kind to be found.
At the meta-level, the issues discussed here indicate the importance of cross-pollination between disciplines. Probability theorists and philosophers of science, not to mention lawyers and detectives, have been interested in the question of multiple independent attestation for a long time. Philosophers have kept working on the issue and making technical progress up to the present, and biblical scholars cannot afford to isolate themselves from these other disciplines. The present essay is offered in the hopes of encouraging such interdisciplinary study.
 See William Paley, Horae Paulinae (London: SPCK, 1877); J. J. Blunt, Undesigned Coincidences in the Writings Both of the Old and New Testament: An Argument of Their Veracity (Birmingham, UK: The Christadelphian, 1847); Lydia McGrew, Hidden in Plain View: Undesigned Coincidences in the Gospels and Acts (Chillicothe, OH: DeWard, 2017); Lydia McGrew, “Undesigned Coincidences and Coherence For an Hypothesis,” Erkenntnis (2018), https://doi.org/10.1007/s10670-018-0050-4.
 William Paley, A View of the Evidences of Christianity, In Three Parts, repr. ed. (Murfreesboro, TN: DeHoff, 1952), 336.
 Branden Fitelson, “A Bayesian Account of Independent Evidence with Applications,” Philosophy of Science 68.3 (2001): S123–S140, https://doi.org/10.1086/392903; McGrew, “Evidential Diversity and the Negation of H”; Lydia McGrew, “Accounting for Dependence: Relative Consilience as a Correction Factor in Cumulative Case Arguments,” Australasian Journal of Philosophy 95.3 (2017): 563–69.
 William Lane Craig, Reasonable Faith: Christian Truth and Apologetics, 3rd ed. (Wheaton, IL: Crossway, 2008), 365.
 E.g., Michael R. Licona, The Resurrection: A New Historiographical Approach (Downers Grove, IL: Intervarsity, 2010), 231–32, 437.
 Richard Bauckham, Jesus and the Eyewitnesses (Grand Rapids: Eerdmans, 2006), 48–51.
 Craig does mention briefly such differences among the Gospel accounts of the empty tomb. Reasonable Faith, 365–66. As long as one does not concede that those differences are the result of invention (see later sections), they can function to support independence. Here I am evaluating only Craig’s claim that 1 Corinthians 15 constitutes independent attestation to the empty tomb.
 While causal independence and probabilistic independence are not identical, in the case of testimony they are closely connected. See, for example, Gregory Wheeler and Richard Scheines, “Coherence and Confirmation Through Causation,” Mind 122.485 (2013):135–70; McGrew, “Evidential Diversity and the Negation of H”; McGrew, “Accounting for Dependence,” 567–69,
 This type of use of the term “midrash” is expressly rejected by N. T. Wright in Who Was Jesus, new ed. (Grand Rapids: Eerdmans, 2014), 92–97. However, the terminology or attempted genre classification does not affect my point about independence.
 Richard Bauckham, “Luke’s Infancy Narrative as Oral History in Scriptural Form,” in The Gospels: History and Christology: The Search of Joseph Ratzinger-Benedict XVI, ed. Bernardo Estrada, Ermenegildo Manicardi, and Armand Puig i Tàrrech (Vatican City: Libreria Editrice Vaticana, 2013), 1:399–417.
 Michael R. Licona, Why Are There Differences in the Gospels? (Oxford: Oxford University Press, 2017), 165–66.
 Licona, Why Are There Differences, 252n119.
 Daniel B. Wallace, “Ipsissima Vox and the Seven Words from the Cross: A Test Case for John’s Use of the Tradition” (paper presented at the Regional Meeting of the Society for Biblical Literature, Dallas, TX, 5 March 2000), 12, emphasis in original.
 Wallace, “Ipsissima Vox,” 4.
 Wallace, “Ipsissima Vox,” 12.
 Licona, Why Are There Differences, 165.
 Wallace, “Ipsissima Vox,” 12.
 D. A. Carson, “Historical Tradition in the Fourth Gospel: After Dodd, What?,” in Gospel Perspectives: Studies of History and Tradition in the Four Gospels, ed. R. T. France and David Wenham (Sheffield: JSOT Press, 1981), 2:106.
 Wallace, “Ipsissima Vox,” 2.
 Lydia McGrew, “Bayes Factors All the Way: Toward a New View of Coherence and Truth,” Theoria 82 (2016): 329–33.
 For the concept of confirmation “through” another proposition, see Lydia McGrew and Timothy McGrew, “Foundationalism, Probability, and Mutual Support,” Erkenntnis 68 (2008): 55–77.
 Licona, Why Are There Differences, 166; Wallace, “Ipsissima Vox,” 10–11.
 Wallace, “Ipsissima Vox,” 12.
 Licona, Why Are There Differences, 181.
 Ibid., 175.
 Craig Keener, The Gospel of John: A Commentary (Grand Rapids: Baker Academic, 2003), 2:1196–1200.
 Licona, “Licona Responds to Ehrman.”
 William Lane Craig, Reasonable Faith, 11–12.
 Although the probabilistic issues are too complex to enter into here, there are different possible meanings of “reliability.” In one sense of the term “reliable,” it is impossible even by independent attestation to get confirmation from multiple sources that are completely unreliable, considered individually. While on one probabilistic construal of “reliability” it is sometimes possible to get confirmation from the independent agreement of multiple sources that are not independently reliable, each source must still have some individual evidential value (in a specific technical sense) for what it attests. See Rodney D. Holder, “Hume on Miracles: Bayesian Interpretation, Multiple Testimony, and the Existence of God,” The British Journal for the Philosophy of Science, 49:52–53; John Earman, Hume’s Abject Failure (Oxford: Oxford University Press, 2000), 55.
 Of course, independent attestation is only one of the criteria of authenticity, but it plays an important role among them.
 William Paley, A View of the Evidences of Christianity, Part I; Paley, Horae Paulinae.
 J. B. Lightfoot, “Internal Evidence for the Authenticity and Genuineness of Saint John’s Gospel,” in The Fourth Gospel: Evidences External and Internal of Its Johannean Authorship, ed. Ezra Abbot, Andrew P. Peabody, and J. B. Lightfoot (London: Hodder & Stoughton, 1892), 131–71.
 Colin Hemer, The Book of Acts in the Setting of Hellenistic History, WUNT 49 (Tübingen: Mohr, 1989).
 Leon Morris, Studies in the Fourth Gospel (Grand Rapids: Eerdmans, 1969).
 For example, Carson, “Historical Tradition in the Fourth Gospel,” 83–145.
 Craig L. Blomberg, The Historical Reliability of the New Testament (Nashville, TN: B & H Academic, 2016); Craig L. Blomberg The Historical Reliability of John’s Gospel (Downers Grover, IL: IVP Academic, 2001).
 Richard Bauckham, Jesus and the Eyewitnesses.
Lydia McGrew is a widely published analytic philosopher and the author of Hidden in Plain View: Undesigned Coincidences in the Gospels and Acts.
Other Articles in this Issue
Speaking in tongues potentially includes three subcategories: (1) known language; (2) unknown language; and (3) language-like utterance—an utterance consists of language-like sounds but does not belong to any actual human language...