In Obsolete Paradigm (37) I noted that Elizabeth Myers made a recent case for the authenticity of 1 Peter that is data driven and not a product of doctrinal reasoning. I have since spent time examining her case closely and I must say I’m convinced. Even if I am not certain, this still has consequences for the probability that Jesus existed that I am sure Myers did not anticipate and probably would not want to entertain. But we should. So I am composing a two part series on this: first, I will discuss Myers’ study and why I find it persuasive; then, I will lay out the consequences I believe this has on the probability of historicity.

Backstory

Many experts agree Hebrews was written in the 60s AD. And those who don’t, argue solely from dubious tradition, not “evidence” in the way other fields of history would call it. The actual evidence for Hebrews being early is in my opinion conclusive—even stronger than for How We Can Know 1 Clement Was Actually Written in the 60s AD, and that’s pretty solid too. This is dangerous to admit because an early date for Hebrews looks odd for an early movement text unless Jesus wasn’t a real historical personage, but only known of mystically (Historicity 538–40, 594). Not so odd as to slam the dunk; but worrying (540–52). 1 Clement also famously overlaps Hebrews (see Myers, Probability of Literary Dependence 210–11, cited below) and while debate remains which was using the other, if they both published in the same five-or-so year period that distinction would become less pressing to resolve.

I lead with this, because while an ardent historicist can just dismiss this as too weak a point to matter (and by itself, they’d be right), it becomes a five alarm nightmare for them if they admit 1 Peter was written before Hebrews. And there is now pretty good evidence it was. I’ll get into all that. But first, you need a crucial point of context:

Everyone rightly agrees that 1 Peter and 2 Peter have different authors (their stylometry is wildly discordant; and not in any way that can be explained by Peter “using scribes”), and that 2 Peter acknowledges and is kind of a response to 1 Peter. So we know for a fact 2 Peter is a later forgery; and it’s concerns and style fit best the second century (see my discussion, with abundant citations of scholarship, in chapter seven of Paradigm). But we don’t actually know either about 1 Peter. There have never really been any good arguments that it is a forgery or late. It could be the earliest written Christian document surviving. The Argument from Illiterate Fishermen is really the only argument against that conclusion and it completely falls apart on any close examination (as I summarized in Things Fall Apart but perfected with more discussion and citations to scholarship in Paradigm 125–28). That’s really dogma, not an argument.

This is important because 2 Peter invents exactly the eyewitness testimony to the real historicity of Jesus that should instead be in 1 Peter—and specifically to defeat deniers of it. So if the evidence were the other way around—if 1 Peter were the obvious late forgery and 2 Peter the early authentic writing of the founder of Christianity (the launcher of the resurrection cult that acquired that name)—this would be very strong evidence for historicity. If we were certain it was early and authentic, it would definitely get the odds well above 50/50, as then it would be thousands to one more likely Peter actually met Jesus even if he was making things up about what he saw—historicity would nonetheless be presumed in his audience and all but impossible that early for him to fabricate (see Historicity 53–55 and §6.7). So if we were certain 2 Peter was authentic its effect could be thousands to one, leaving a margin of error on Jesus existing of, say, 33% to near 100%. And that low end is below 50% only because attesting to what could be a vision is still compatible with mythicism; if 2 Peter contained testimony more assuredly referring to a real man, that would be tens of thousands or more against it, and then the lower error margin would rise well above 50%.

The letter’s effect is limited to its uncertainty of authenticity, however. So if we were only 10 to 1 sure of its authenticity, then an authentic 2 Peter’s effect could not exceed 10 to 1 in favor of historicity. Then the Paradigm probability of historicity (379–86) would still be near zero on the low end—but over 80% on the high end, which would keep mythicism in play but leave historicity more likely. So the fact that this is all reversed (2 Peter is the fake and 1 Peter authentic) is evidence against historicity. And I’ll explain how bad this is for historicity after here explaining why we can now say it’s more likely than not an authentic early letter of the real founder of the religion: Cephas (a.k.a. Peter, who may or may not have really been named Simon).

The Myers Studies

What Elizabeth Myers has proved to my satisfaction across several studies is that the many parallels between Hebrews and 1 Peter that have long been puzzled over (she surveys scholarship noting and attempting to explain them spanning over a hundred years back and all the way into the present) are evidence sufficient to conclude that one of them used the other for ideas and material. Which entails one definitely wrote with knowledge of the other, and in sequence.

Myers also proves to my satisfaction that it is not as certain but nevertheless probable that Hebrews used 1 Peter and not the other way around. Which would entail 1 Peter must pre-date Hebrews. And that is what I am sure will spark an “all hands on deck” effort to now “prove” Hebrews cannot be early so as to escape the consequences of 1 Peter being early—or else to “destroy” Myers for making this necessary (“How dare she!”). Because as my readers will have seen countless examples of here by now, biblical scholars care more about their emotionally attached dogmas than with changing their mind in response to evidence. So as soon as facts become uncomfortable, those facts must be made to go away—rather than the ideas they threaten abandoned. But as you know, I don’t play that game. I’m going where the evidence leads. Myers of course (I assume) would not support all the consequences of her thesis that I lay out (she’s not a mythicist). But that her finding of “before Hebrews” does have those consequences is what I will argue.

The most essential book here is Elizabeth Myers, Intertextual Borrowing between 1 Peter and Hebrews: Probability of Literary Dependence and the Most Likely Direction of Borrowing (2020). That lays out the case for dependency and direction. Although if you want to adopt, test, or critique her mathematical apparatus and its reliability as a method, you will need its preceding volume, Probability of Intertextual Borrowing: A Methodology for Determining the Likelihood of Literary Dependence and the Direction of Borrowing between New Testament Books (2020). These appear to comprise either her PhD dissertation (or postdoctoral research thesis) in biblical studies, which she completed after a long career in electronics engineering. Which explains her facility with math and probability. Ironically, I, too, come from an electronics engineering school (in the military), and (probably like her) got my math chops before that in advanced courses in high school (and took statistics in college). Historians tend to be terrified of math unless they have a math background.

There is also a third, more affordable book, Authorship of 1 Peter and Hebrews: New Evidence in Light of Probable Intertextual Borrowing (2020), that tries to argue beyond those volumes’ results that Hebrews was “therefore” written by Priscilla and Aquila, but, IMO, not convincingly (members can read the SBL Review). Myers does not present enough evidence to overcome Base Rate Neglect here (the Fallacy of Suspicious Coincidence): we almost certainly will not be so lucky as to have the author of this text named somewhere and yet coincidentally not as its author (or as an author at all). There are dozens of available authors most of whose names won’t chance to survive in our scant few texts. So the base rate expectancy is that we won’t know the name of the person who wrote this text. That is a far more probable hypothesis. But we can know some things about them, such as that they were a well educated member of the movement alive in the 60s AD who knew Timothy and was eager to extend the logic of Paul’s ministry further than he dared.

This is because Paul avoided outright preaching that Jews could abandon strict Torah observance (quite carefully in fact). He argued only that Gentiles need not adopt it. But the author of Hebrews carries his same logic to its full conclusion: that perhaps even Jewish Christians should abandon the strict requirements of Torah and not return. Hence it is written to “the Hebrews” as Christians thinking about “returning” to Torah observance and its reliance on Temple service. Which means Hebrews must be written before the Jewish War (as it has no knowledge the temple cult doesn’t even exist to return to) and after Paul died (as its audacity requires him to be unaffected by it), a death 1 Clement references, and which Paul’s letters entail must have been in the early 60s at the latest. Which solidly places Hebrews mid-60s.

Notably, the author of Hebrews and 1 Peter are on opposite sides on its table-flipping expansion of Paul’s argument. Which is one more reason why I think the direction of borrowing most likely runs from 1 Peter to Hebrews and not the other way around. And here I will assume this means Peter actually wrote it (in Aramaic, the name is Cephas—neither a real name, but both meaning “Rock,” and thus evidently some kind of cult moniker: see Historicity 416n62 and 524–25n29 with revisions [1] and [2] now in the Errata). Peter would not respect a work like Hebrews nor dare attach himself or his arguments to it; indeed, he’d be disinclined even to read it other than to refute it, yet the parallels are almost all respectful and agreeing, not polemical. By contrast, Hebrews would use 1 Peter as a base text, in fact would be inclined to, as it would then carry a supreme authority, be well known, and never directly contradicts what Hebrews argues.

Someone arguing for abandoning Torah would slaver at the opportunity to use ideas and motifs and arguments from 1 Peter to bolster their case. They would want to take advantage of the image of Peter being amenable to this possibility that Paul inadvertently painted in Galatians (and yes, there has long been a suspected reliance on Galatians by the author of Hebrews as well). It’s a perfect example of taking a well-trusted text and using it to pivot to a new idea contrary to the original author’s intention or expectation. 1 Peter is not doing this to Hebrews. But Hebrews is doing this to 1 Peter. Many of Myers’ examples even carry this implication (where Hebrews “expands” on points simply taken for granted in 1 Peter).

The Case in General

In her second edition commentary, Karen Jobes argues 1 Peter is probably from around the 40s or 50s AD because (1) it is not aware of later Roman state persecution (e.g. in Pliny’s letter of around 112 AD), which is true; (2) its Greek is peculiarly Semitic, indicating a Jewish author for whom Greek was a second language, less probable for a second century forgery thesis, which is also true (in contrast to Luke-Acts, whose semitisms appear to derive from deliberate emulation of the Septuagint more than the author’s natural language); and (3) traditions imagining Peter illiterate are not that trustworthy, which is again true.

Indeed, Jobes notes that if the Gospels imagine him having a brother with a Greek name (Andrew = Andreas), they are assuming he knew Greek, and if he was no mere coxswain but running a fishery business, he could be literate in Greek as well (as noted above, I have presented evidence for this myself: all Rabbis were educated and required to ply a trade, and evidence in Paul entails Peter had to be a literate Rabbi). Which matches her second point as well. And to her first point we can add the letter’s ignorance of all the Gospels and every story and saying in them (by contrast to 2 Peter), and its seeming ignorance of anything Paul wrote—to combat, reinterpret, or support (also by contrast to 2 Peter).

On the other side there are really no arguments against 1 Peter’s authenticity or early date. The ineffective Argument from Illiterate Fishermen doesn’t cut it. Other arguments are circular (e.g. assuming it hails from the second century, interpreting its contents accordingly, and then using those results to argue it hails from the second century). Ironically this includes the point that it strangely omits any personal knowledge or reminiscences of Jesus—the very oddity that makes it a danger to historicity if admitted to be early.

Enter Elizabeth Myers. If Hebrews used 1 Peter as a source, that supports Jobes conclusions even more. So to establish they knew each other, Myers lists the parallels between them (there are dozens), and develops a complex method of estimating the comparative probability that a given parallel is accidental (inclusive of sharing a common source like the Septuagint or a common vocabulary like Christian “missionspeak,” and the like), the converse of which is the probability the parallel is a product of direct borrowing instead (either the author of 1 Peter adapting material from Hebrews, or the other way around). She looks at other Christian and non-Christian literature as controls (how often does a parallel occur anywhere else for example, and thus not due to adapting either source). But ultimately her approach is to list criteria (like length, rarity of content, structural similarity, functional similarity, etc.) and guesstimates how improbable a given conjunction of criteria is by chance. And then she runs the probability of none of these being due to borrowing—the converse of which is the probability of borrowing.

In outline her approach is valid. One can question her assigned weights (how much should a certain amount of rarity or similarity count in terms of the probability of that occurring by any version of chance). But I do not think doing so would make much difference. Because she is already using overly skeptical estimates of a “direct” vs. “not direct” connection throughout. She builds a lower and upper margin of error and roughly finds the probability of direct relationship to be between 80% and 90% (for odds between 4 to 1 and 9 to 1). I think she’s being far too lenient. The actual probability is surely higher. I’d put it above 100 to 1 (> 99%).

But this isn’t the relevant probability for historicity (more on that in Part 2). The real question is direction of borrowing.

For this Myers finds a 33% chance that 1 Peter borrowed from Hebrews, which for the upper bound I will use as a likelihood. Most might say the probability that 1 Peter would look as it does if it borrowed from Hebrews is closer to 50/50, as there is nothing distinctive of that direction of borrowing. Myers is essentially saying that we should have more evidence of this if it were the case; there is only one item of such evidence (the awkward disruption of the digressive pesher from Isaiah in 1 Peter 2:22–25: Probability 166–69); but that is less than we expect; so she gets a probability of seeing this of only 33%. Her method would have gotten a much higher probability with more such evidence, as one should expect. But suppose you think the base rate of borrowers showing no distinct signs of borrowing is appreciably high. Then you might say the probability that 1 Peter would look like it does should be closer to 50% than 33%. So I’ll use that as my upper bound on that likelihood. And as that would be the same whether 1 Peter is connected to Hebrews at all or not (it looks the same either way) it stays at 50%.

Myers then finds a 98% chance Hebrews borrowed from 1 Peter, owing to a whole stack of evidence for this direction of borrowing. Her method in both cases is essentially to list 13 generic features that would indicate this (in any context, so her method is not limited to just comparing these two texts; she tests the method on other known texts in her first volume), and takes the percentage of those features (minus a rightly disregarded epsilon) as the probability of that direction of borrowing. Then she calculates the total probability of all these instances nevertheless not being that, the converse of which is the probability of it then being that. For the 1 Peter direction she had only 5 data points leaning that way (getting her the 33% expectancy); for the Hebrews direction she has 30, many of which leaning even harder (getting her the 98% expectancy). One can argue she is under-weighing the data points in every or many cases. So it should be even higher than 98%. But if we went the other way and halved all her weights instead, her 98% would become (say) 86%. I think this is the lowest it can honestly ever be claimed to be. Which gets a likelihood range of 86/50 (low end) to 98/33 (high end), so from 1.72x to 2.97x for her thesis, or roughly 1.7x to 3x.

The Case in Particular

The question then is how does Myers arrive at this result for Hebrews borrowing from 1 Peter. And if you are confused by the math, it will help to just see the reasoning it quantifies.

Scholars have long observed that Hebrews appears usually to expand on ideas in 1 Peter rather than 1 Peter abbreviating or assuming acquaintance with Hebrews; and other features that are unexpected. For example, 1 Peter 3:22, Hebrews 2:5–8 (cf. 10:13), and Ephesians 1:22 all conflate the same two Psalms (110:1 and 8:6) and in the same way. This could be due to all three using a common (now lost) Christian pesher. Or it could be because each is citing the other. And in that case, 1 Peter seems the more probable originator. Scholars have catalogued some 200 parallels between Hebrews and 1 Peter like these, but Myers concludes most can be readily explained by commonalities of text and context (these are Christian books, for example, so they already use common ideas and vocabulary). Myers narrows these 200 down, therefore, to some 80 parallels that are not as explicable without direct influence; of which 35 are peculiar enough to suggest even a direction of borrowing.

I think Myers’ second book, Probability of Literary Dependence (or PLD, on direction of borrowing), would have benefited from adding to its appendix her equations of procedure by which every table’s data can be reproduced. This is explained in her first book. But readers of the second need it in one easy reference point. Similarly, her tables need keys (symbols and abbreviations are only explained deep in the text; they should be reproduced in the appendix for easy reference). But for example, her directional evidence consists of 30 items in favor of 1 Peter → Hebrews. For each one she counts up a score of “hits” on criteria and gives the probability as a straight percentage of those criteria hit. Which assumes each criterion carries the same weight as evidence. But any mod to that is unlikely to make much difference. It also assumes hitting all 13 would max the probability at 100% (functioning as a likelihood or expectancy rather than “the probability,” because it is then compared to the alternative scenario’s probability, which could also be 100%). But correcting that to account for some epsilon (some small chance otherwise) also wouldn’t make much difference. And in any event, no item scores more than 4 anyway (for a 31% probability of her thesis being correct on just that one item), and most score only 1 (at 8% each).

Her thirteen criteria break down into four categories: disturbance (the text appears to disturb something about the letter’s flow, which is more likely if it is being intruded into the text from another source rather than an original thought), consistency of practice (all the proposed borrowed material is integrated in a consistent way, in distinction to the source text or the rest of the borrowing text), logical progression (suggesting awareness of the source text), and external source (when one document cites its source, like a passage of scripture, and the other does not, that will happen slightly more often when the first document originated the parallel), each with subcategories. She lists 13 total subcategories (all the ways borrowing could be indicated), but only some are relevant in the Hebrews/Peter case, as follows (some items scoring on a few, others on only one):

Each example thus gets a Borrowing Index (BI), a count of how many of these criteria out of all 13 a single passage hits—in this case ranging from 1 to 4. From that she derives an Independent Probability (IP_bor) of borrowing for each item, derives the converse (IP_nbor), runs a calculation for the Combined Probability that it wasn’t borrowed (CP_nbor), and derives the converse of that, which is the Combined Probability (CP_bor) of this being the direction of borrowing.

Note that this means had Myers combined subcategories, and thus only checked (say) one to five boxes instead of one to thirteen, her result would have been an even higher probability. So by expecting a borrowed passage to have so many indicators at once she is substantially lowballing the probability here. After all, no example in this case checked more than 4 boxes, and only 8 boxes were ever checked. So she could have just “dropped” all the other metrics and goosed her result to a higher probability. She didn’t. So her result is a definite undercount of the resulting probability. And that is also why trying to make some subcategories weigh more than others is unlikely to make a substantial difference. If anything such jiggering would increase her probability again—unless you gerrymandered your weights to deliberately undermine her thesis, which would be foul play.

Best Case

Her best example (the only one with a BI of 4) is her thirtieth parallel out of the assessed eighty: the proposed adaptation of 1 Peter 2:5c into Hebrews 13:5, scoring DL, CS, PS, and PF. For example, Hebrews uses two words for “offer up,” anapherômen and propherô, but only uses anapherômen when paralleling 1 Peter using it (170), while 1 Peter never uses the other word. That’s not a proof, but it is slightly less likely on any other theory than that Hebrews is using 1 Peter and not the other way around (she would score it alone as only 1/13th or 8% likely). This is also part of a larger structure of proposed borrowing and thus part of a consistent practice (177–78), and in a logically expected way if Hebrews is expanding on 1 Peter rather than 1 Peter abbreviating Hebrews (185). And Hebrews converts Peter’s exposition to exhortation which is more typical if that is the direction of borrowing. As Myers explains, “the text in 1 Pet 2:5 explains what Christians are established by God to do, [while] the parallel text in Heb 13:15 instructs them to do it,” which typically requires the explanation of the thing to precede the exhortation to it. And here, the preceding text is in 1 Peter, with Hebrews thus presuming or exploiting an acquaintance with it. All these four features converging make the evidence on her thesis 31% likely on just this parallel alone.

The structural argument here is important. Myers locates more than one example of “fan structure” in Hebrews, where a passage in 1 Peter gets “fanned out” in the same order across multiple passages of Hebrews, in a way hard to explain as the product of an abbreviating process (it would be a weird way to abbreviate or summarize a text) but not hard to explain as the product of an exegetical process—Hebrews is in effect functioning like a midrash or “a commentary on” 1 Peter, or an expanded “reinterpretation” of it, which presumes it already exists to comment or expand on. The parallel described above is part of the most extraordinary example of this, here diagrammed from Myers, PLD 69 (and this isn’t the only example of her discovered fan structure, it’s just the largest):

The single line “you also, like living stones, are being built into a spiritual house to be a holy priesthood, offering spiritual sacrifices acceptable to God through Jesus Christ” has three ideas in order, emulating Jesus: Christians as a house, as a priesthood, and as a priestly sacrifice. These get expanded with exegesis pretty much in the same order across (a) Heb 3, (b) Heb 5–7, and (c) Heb 8–10. This is significantly more likely if Hebrews is expanding than if 1 Peter is abbreviating. And there are many more fans like this across Hebrews (ten in all); none across 1 Peter. Moreover, Hebrews often does this same thing with scripture, indicating a common practice: this appears to be just a way the author of Hebrews used sources generally. And the fans all consist of what can be described as a reinterpretation of something said in 1 Peter, making Hebrews a covert response to 1 Peter (e.g. Probability 179–89).

Myers’ argument is that no single example proves her case. So you may think that one is tantalizing but not decisive. She agrees. Her point is that once you stack some thirty parallels that all have some if not the same indications, forty-five indicators in all, the overall case becomes harder to maintain that 1 Peter is using Hebrews. All the evidence points in the other direction. Even the single parallel that has any indication of the converse order, counting as five distinct parallels for five total indicators, is far too weak to signify it. It is overwhelmed by all the other data.

For example, throughout her analysis Myers finds a lot of odd choices other scholars have noted about Hebrews who sought various ways to explain them, but which are all explained by her single theory that these oddities result from using 1 Peter as its base text with intent to comment on or reinterpret it. Some examples are even well-attested-but-rejected variant readings that Myers’ theory recovers as the authentic text, because it was previously thought inexplicable, but on her thesis its oddity is expected. I was astonished to discover one of these (inserting an autô into Hebrews 13:21: Myers, PLD 174) isn’t even mentioned in the leading critical edition despite being a strong reading (Greek New Testament 748); I had to follow Myers’ quotation to Metzger’s Textual Commentary to confirm it (and checked other sources to boot). Their reason for omitting it is the want of any reason for it—an argument Myers’ theory refutes, and as the lectio difficilior, her theory now makes it more likely authentic. This all shows she has really done her homework, which marks the value of her study.

Overall, I found her case overwhelming and thus her concluding probability generously low. But I can’t “prove” it should be higher. So I will stick to that until I (or anyone) can prove better.

Concluson

You need to read Myers to truly appreciate the merit of her reasoning and all her evidence. She herself is (I am sure) a historicist and, like Karen Jobes, does not see how perverse 1 Peter is as a letter from Jesus’s closest companion and right-hand man. So I don’t think they see the consequences coming from their own conclusion that 1 Peter is authentic. But I do. And I will explore that in Part 2. Myers is also conservative in her proposal of date ranges for 1 Peter. She believes late legends placing the death of Peter under Nero are believable, which puts a terminus ante quem on 1 Peter of 64 AD. However, I do not believe they are credible at all. But I do believe 1 Clement dates before the Jewish War and thus before 66 AD, yet after (the letter itself says) Paul has died, and best estimates place Paul’s authentic letters ending near 60 AD. 1 Clement also says Peter has died—and it sounds like something that also happened at least a year or two before. So Myers is inadvertently right: 1 Peter must predate 1 Clement, which was written between 60 and 66 AD.

Myers thinks 1 Peter was written toward the end of Peter’s life but gives no good reason for believing that. I think it’s far more likely it dates to the 40s AD, before any of Paul’s letters were written, as it shows no knowledge of them or any conflict they would have created. Myers also believes (as do most scholars; she cites many: 210–11) that 1 Clement used Hebrews as a source. They are definitely related, and Hebrews definitely also predates the war for the same reason as 1 Clement (Myers mentions this possibility but is unaware of the correct argument for it, which I present in Historicity §11.5). But if 1 Clement used Hebrews and dates between 60 and 64 AD, Hebrews must date nearer to 60—but not, in my opinion, earlier, as it radicalizes Paul’s mission in a way Paul did not in his last letter to Rome nearing the year 60, and since Paul would have been in panic mode over that (as it undoes his reconciliatory argument in Romans) it must post-date Romans and thus date no earlier than 56 (the earliest Romans could have been written) and more likely to the early 60s, with 1 Clement then in the mid-60s. This all puts 1 Peter no later than the 50s AD anyway. Either way, 40s or 50s, that’s during the life of most of the companions of Jesus (had he lived) and the first generation of Christians (and all their quarrels over, and hunger for, news about the charismatic life of Jesus). Which also assures us of 1 Peter’s authenticity (as Peter or his friends would call out so soon a forgery).

The unexpected significance of all this I shall explain in Part 2 (that link will go live in a day or two).