Hitler’s Table Talk: Another Update

There’s more news about Hitler! In 2016 I published an update on the saga of Hitler’s Table Talk and its fraudulent or questionable quotations making him out to be an atheist that he wasn’t. That 2016 article contains the background explaining my involvement in this, and the status of the question to that point, which is already quite fascinating, and damning. I highly recommend reading that update. And now today I have another update to report.

Dr. Mikael Nilsson has recently published another article on his research into this inspired by my 2003 article on it in German Studies Review. His new article is in German, so I’ll summarize its salient points for you here. But those who want to tackle the original, you can find it as “Hitler redivivus: Hitlers Tischgespräche und Monologe im Führerhauptquartier—eine kritische Untersuchung,” or “Hitler Reborn: Hitler’s Table Talk and Monologues in the Fuhrer’s Command Headquarters: A Critical Examination,” in Vierteljahrshefte für Zeitgeschichte 67.1 (2019): 105-46.

The official translation of the abstract reads:

This pioneering study of the famous Hitler table talks – “Tischgespräche im Führerhauptquartier” (1951) and “Monologe im Führerhauptquartier” (1980)—uses a lot of previously unknown or overlooked sources to show that historians have been much too uncritical when citing these documents. The texts, the originals of which are now lost, were heavily edited and cannot be cited as being Hitler’s words ad verbatim. They are not more reliable than comparable notes. The notes were written down almost entirely from memory after the conversations they record, often finished long after the first drafts were written. Text was added during the editing process, and the notes contain statements that Hitler most likely did not make. Historians citing these texts are not quoting Hitler, but the words of the authors of the notes.

What’s New

After extensive searching, Nilsson has confirmed the Bormannvermerke is lost. That was the longer, supposedly “complete” edition of the notes of Hitler’s rants taken down in his bunker in Poland from 1941 to 1944, acquired right after the War under shady circumstances by the unrepentant Swiss Nazi banker Francois Genoud, and eventually published in the 1980s under the direction of scholar Werner Jochmann. That text—the “Bormannvermerke” or “Bormann Manuscript,” so-named because it was edited and assembled by Hitler’s secretary Martin Bormann—was supposedly the basis of the only English translation ever produced (and thus the only version typically ever quoted), under the title now famously referenced everywhere, Hitler’s Table Talk.

In my study of this for GSR, and in particular the quotes from it making Hitler out to be an atheist—which is reproduced in Hitler Homer Bible Christ with a new epilogue—I reported finding it said that these notes were produced by two “stenographers” listening in, each in different years, Henry Picker and Heinrich Heim. Picker published his own notes separately, in the original German just after the war (and had transcription mistakes corrected from the originals in a subsequent edition not long after). Heim’s notes were lost, as were Picker’s originals, but edited copies of both their contents ended up edited and even expanded further by Bormann, becoming the so-called Bormannvermerke.

Now Nilsson has confirmed a fourth contributor: Hans Müller produced some entries in 1943 and 1944. So the Table Talk actually comes from multiple authors—Picker, Heim, Müller, and Bormann, all as authors and editors. Moreover, the U.S. captured some pages (only a tiny fraction of the whole, and none on the subject of religion) that I was the first to publish a notice of, and Nilsson confirmed these have found their way back to a museum in Berlin. I only had access to xeroxes in the control of the U.S. Library of Congress. Nilsson got his hands on the originals. He confirmed they are on Bormann’s own paper stock, and adduces evidence they were typed and hand-corrected by Heim.

Nilsson also shows from several examples and direct evidence that the way historians have discussed and used the Table Talk has been routinely rife with errors and misstatements. Despite many declarations to the contrary, in actual fact the Table Talk is not Hitler’s verbatim words. It consists only of paraphrases and summaries composed by people who took notes, often further edited by yet others after the fact, and often unreliably.

It’s particularly important to note that Nilsson now finds the claim that stenographers took notes on the spot is definitely untrue. Rather, recollections of the secretaries present (whether Picker, Heim, Müller, or Bormann) were jotted down, as rough notes to themselves, sometimes a day later, and edited up into fuller statements without consulting Hitler in any substantial respect. Indeed, Nilsson conclusively exposes this myth, and now rightly concludes that we cannot quote the Table Talk as the words of Hitler, but as words put into his mouth by a variety of editors, loosely based on things he might have said, or that they wanted or thought him to have said.

Part of the evidence for this that Nilsson found is a transcript of a tape-recorded interview of Heim himself, and archives of his correspondence, as he recollected first hand how the process actually worked. Nilsson also found a 1951 critique of the Table Talk‘s reliability in an archived letter in Berlin that I also hadn’t known about.

From Nilsson’s article, I vetted and improved Google’s translation:

Hitler’s Air Force Adjutant, Nicolaus von Below, gave a crushing verdict on the Table Talk as early as 1951, after Picker published his edition. He said they were not credible records of what Hitler had said, and try as he might, he could not remember any monologue from Hitler of such length. Instead, he said, Hitler’s remarks happened here and there in the course of discussions, and to understand them one also has to know what the other participants had said. One cannot simply take a person’s statements out of context and then publish them in the way Picker and [Picker’s editor] Ritter do.

In the transcribed interviews, even Heim himself said much the same. I confirmed Google translate already nailed this one from Nilsson’s article:

Heim himself never claimed his notes were stenographic records. According to his own statements, he first typed a version—sometimes based on a few keywords that he had been able to write down during Hitler’s remarks, but mostly from memory—and then made any necessary changes. Subsequently, an “original” and two copies were typed. Bormann retained the “original”, and the two copies were sent to two separate sections of [the Nazi] Party headquarters in Munich.

So the legend of this being a stenograph is false. Worse, it appears to be just cobbled up notes from memory, only occasionally even based on scribbled keywords. So these are the words, the recollections, of different witnesses, and not Hitler’s own direct words—at all, much less in context. Heim even admitted (in a BBC interview taken in 1953) that this happened a day later. So it wasn’t even “right after” any remarks were made. And at most Hitler was aware of and approved this happening; but there is actually no evidence Hitler checked, vetted, or edited any of it.

Nilsson further shows, from comparing alternate eyewitness accounts of some of the same conversations that ended up edited into the Table Talk, that the Table Talk text greatly distorts the context and content of what Hitler actually said on any given occasion. He also proves a great deal of the content came from Bormann’s own hand, weeks or months later in Munich, amounting to numerous interpolations and alterations (and Nilsson still by no means has made an exhaustive examination). Some of these edits, Nilsson shows, even deliberately reverse what Hitler said, to sanitize it for posterity. And he likewise finds examples of the words of others present being confused as the words of Hitler in the text of the Table Talk.


None of Nilsson’s examples relate to Hitler’s views on religion. But his overall findings do deeply diminish the value of the Table Talk as a source for Hitler’s views on anything, religion included. The English translation I already demonstrated to be fraudulent, based on the heavily and bizarrely distorted French translation of Genoud. But even the German text (of which there are multiple disagreeing versions) intermingles the views of each entry’s authors and editors with Hitler’s, takes Hitler’s comments out of context and even creates new context for them, mistakes the comments of others present as Hitler’s, and is drawn at best from loose memory a day after the fact, by three or four different persons at different times, after countless things were discussed by numerous people. And then all was edited and added to by Bormann. (And possibly meddled with by Genoud!)

It’s not looking good for the Table Talk as a source. But it comes to mind that we can compare this even to what historicists want to be true for at least some of the sayings of Jesus.

Here we actually have a source document, a massive collection of sayings, published within mere years of the fact. Unlike the Gospels, which are written up only many decades after the fact. A document whose authors are actual eyewitnesses for whom we can identify their names, roles, biases, interests, verify their presence during what was supposedly said, and so much more. Unlike the Gospels, whose authors are unknown, definitely not eyewitnesses, and about whom we know nothing other than what we can guess at from the Gospel they supposedly wrote. Eyewitnesses, moreover, who wrote down their recollections a mere day after the fact—whereas we have no evidence anything of the kind was done for the sayings in the Gospels. And yet even this document is now to be adjudged deeply unreliable, by standard historical criteria.

Owing to a large amount of eyewitnesses documentary material we have nothing the like of for Jesus, Nilsson was able to verify that this source document confused things other people said for things Hitler said, that it often reflected more the views of each saying’s author’s assumptions and desires and dodgy recollections than what Hitler may have actually said, that it routinely took his sayings out of context, that it was heavily edited by other parties (particularly Bormann) who had their own worldviews and agendas, and was radically distorted when translated into a foreign language, to the point of often saying even the exact opposite of the original.

Imagine if all we had were that English translation, and no access to all that other evidence by which Nilsson and I have uncovered tremendous sources of error, fabrication, and distortion in the Table Talk, even in just a matter of a few years (barely a decade altogether). Would we so foolishly trust it as the reliable record of the words of Hitler? We certainly shouldn’t. And for the same reasons, we shouldn’t trust the Gospels either. For they, too, were just a distorted translation from a lost original that had already gone through multiple redactors, and originated from witnesses (even if only to revelations) whose original recollections we have no access to, and which may have already been heavily distorted by their own worldviews and agendas and flawed memories, even before the further distortions of multiple generations of retelling, inventing, and translating into a foreign language, in a foreign land.

By every measure (eyewitness sourcing, closeness in time, documentary verification, availability in the original language) the Table Talk is a far more reliable source than the Gospels. And yet it’s a shit show of distortions; demonstrably unreliable, or at best only barely reliable. That means the Gospels are necessarily far worse than the Table Talk. And thus far worse than unreliable. Blame Hitler.


  1. Great to see your contributions have created serious retrospecting by all those involved in creating this book.

    I do expect to see a lot of people and scholars unfamiliar with these findings until a revised version of the book will be published. Let’s hope that will happen very soon by the publisher.

    Perhaps we patrons & fans can help by colletively sending in a petition to the publisher to create and publish this revised version?

    1. They’ve already tried. The project for whatever reason keeps falling through. Maybe it’s ongoing and taking a while; but last time they said they’d do this, they came out instead with a reproduction of the bad translation and just added an intro mentioning my findings.

      Possibly really replacing it is too expensive. It’s an enormous amount of translating hours, also involving the need for a new critical edition even of the original German (and that German has a lot wrong with it too, even apart from the fact that a lot of it may have been meddled with by Genoud). Maybe it’s just cheaper to keep publishing the same fraudulent translation and “getting away with it” by simply mentioning somewhere buried in the intro that it’s fraudulent (without actually saying that in so many words).

  2. So in Bayesian terms :
    – the prior probability of Hitler being not an atheist, but a theist, was > 90% (based on the background knowledge that he wrote many a theist-inspired assertion in his most revealing book “Mein Kampf”, e.g. the destruction of the Jews is mere the execution of God’s plan)
    – the evidence of Hitler’s “Table Talks” does not support that thesis, nor does it reject it. The same goes for the anti-thesis “Hitler was an atheist”. Because the evidence does not indicate Hitler’s ideas on those propositions, and if they did, it’s not proven they are really his. So the evidence is of no value whatsoever, it gives a 50/50 chance to either of the hypotheses.
    – that means that the posterior probability of Hitler being a theist remains > 90%.

    I hope I’ve got this right !

    1. Close.

      I’d say some of the evidence in the TT, which all uniformly establishes Hitler a theist (and in fact hostile to atheists), still increases the probability Hitler was a theist, because the scattered data comes from multiple witnesses (more than one author of entries in the TT) and is “contrary to interest,” i.e. it doesn’t have Hitler espouse but in fact elaborately attack Catholicism, despite publicly being a devout Catholic. And all the stuff making him out to be an atheist we have verified in evidence was forged by Genoud, one was added by known-atheist Martin Bormann, and some were demonstrably caused by the out-of-context practice of the other recorders.

      There is a probability these multiple witnesses to Hitler’s theistic anti-Catholic rants all made them up or altered them in the same direction, but it’s not so high as 50/50. It’s also not vanishingly small. But the difference is still enough to effect a slight increase in the proposition’s probability. This is the consequence of using sources that are problematic but not completely useless. But we can only do this because we have so much more information to do it with—none of which we have, for example, for the sayings of Jesus in the Gospels.

      1. I’m glad you didn’t say my reasoning was wrong, only the consequent probabilities I used. Indeed, as you stated in “Proving History” : “The [objections to probability assignments and their derivation from accepted background knowledge] are exactly the kinds of debates historians need to be engaging in.” Knowing nothing much of the TT, I was just being conservative. As you proved here, the reasoning a forteriori holds.

      2. Richard have you considered catering data, and enough of it, to apply actual machine optimisation. I understand much of the data is contextual, subjective etc. But I’m not too sure that to an unsupervised machine learning model this matters too much since it will apply the categories anyway. There are also simple methods to label data using machine learning.

        I’d be pretty sure you have already considered using machine learning on this Jesus topic. Although to my mind there are far more interesting questions in history it could be applied to. The long standing case of the poisoned Agricola springs to mind.

        The main problem I can see here is the quantity of data. That’s what all machine learning revolves around. How much data is there? Can it be structured?

        1. I’m not sure how you imagine this would be done. But it would require millions of dollars of investment in expert teams and resources. Enigma wouldn’t even invest tens of thousands on a new translation. So we won’t be seeing anything yet more ambitious done either.

  3. This is really illuminating.

    In the modern age we have an influencial messianic figure who captivated a whole country, at the height of his power, and his words are still not original but edited re-interpretations of what he might have possibly said. And even those original versions are lost. This happened in a highly literate society, less than 80 years ago in an international world, widely interested (for one reason or another) in a controversial notorious figure.

    On the other hand we have the exact words of an obscure carpenter from 2,000 years ago, from a strata of society that was largely illiterate, in a world that was only partially literate. The claims for the earliest versions of those documents containing his words, after earlier dates (and authorship) have been be proved spurious, are at best a century after his death, at worst perhaps two centuries.
    Yet we have his exact words!

    It’s a miracle!
    Hallelujah brothers and sisters.
    Praise the Lord!

  4. Frederic Christie January 29, 2019, 8:38 pm

    Typo: “Its fraudulent” not “it’s fraudulent”, second sentence.

    I’m glad you and others have been doing this work to correct the public perception of Nazism and Hitler. To me, of course, it never really mattered. Even if Hitler was an atheist, it’s moot. The Nazis combined weird Teutonic revivalism, positive Christianity, atheism, and whatever else, but the bulk of the party was always Christian. As Folding Ideas’ excellent review of The Triumph of the Will notes, their propaganda would use iconography like the great cathedral in Nuremberg to show that the Nazis were the apotheosis of a virtuous history. The argument that “Hitler was an atheist” is always an attempt to try to pretend that Christians don’t have any soul-searching to do because at best the leadership wasn’t Christian, but it certainly disproves that Christians can’t become Nazis. Which tanks the conservative thesis that Christianity is somehow critical to avoiding Nazi-like moral lapses.

        1. Standard modern translations hide the real words underneath loose or deliberately crafted interpretations and thus you cannot read the Bible in translation. So you always have to check the original language. You can’t count on the biased translations of sectarian Christian interpreters.

          Every translation today that says “band” or “gang” or “pack” is hiding the actual word, which is “assembly” or “congregation,” which is what the word “synagogue” means (literally “a bringing together“). Hence that’s exactly the word the ancient Rabbis used to translate this passage into Greek in their expanded Septuagint: “synagogue.” So they knew what the word meant. And Mark (like all the Gospel authors) is using that Septuagint as his base text (as I show and explain throughout Chapter 10 of On the Historicity of Jesus, although this is widely known in the field).

          And again, why are you asking these questions in a thread completely unrelated to these issues?

    1. D’Souza is a liar, a criminal, and a conman. Any claim he may have made that Hitler was an atheist has already been refuted. It is a claim only espoused by Christian apologists, based on no or even fraudulent evidence, and standard Ken Ham style fallacies like quote mining. It is not espoused by any actual Hitler historians, nor in any actual peer reviewed work on Hitler’s religious views.

      1. That’s a mite steep! The Obama state went after him like nobody’s biznis fr a measly $20K. When Obama did it…a slap on the rist and a fine. Just as well Pres Trump pardun’d him.

        But the Nuremberg laws were got from the democrat manifesto – which was even too much for the national socialist german workers party.
        Though Whitman attributes it to ‘America’ rather than the democratic party (there’s a con right there!)

        Hitler’s American Model: The United States and the Making of Nazi Race Law
        by James Q. Whitman (Author)

        1. The FEC says the Obama campaign failed to disclose the sources of 1,300 large donations, which together accounted for nearly $1.9 million. Election Commission rules state campaigns must report donations of $1,000 or more within 20 days of Election Day.

          Obama for America was also fined for “untimely resolution of excessive contributions,” according to the conciliation agreement, FEC says. The campaign accepted more than $1.3 million in contributions that came from donors who had already given $46,000—the maximum allowed by FEC rules. The campaign eventually refunded the excess cash but did not do so within the 60-day window allotted for resolving such cases, FEC said.

          In addition to failing to report big donors and excess donations in a timely manner, the Obama campaign incorrectly dated the filings dealing with $85 million in funds, the FEC claims. This error appears to have been primarily the result of one transfer to the campaign committee from the Obama Victory Fund, a fundraising group that includes money raised by the Democratic National Committee that is earmarked for the presidential race.

          And D’Souza gets shaftid fr a measly 20K by the Obama state stasi.

          The nuremberg race laws were framed on the democratic party manifesto – is this true?

        2. Right. An accounting error that was always on the books and correctly reported and resolved is exactly the same as fraudulently laundering money to conceal the origin of election funds to bypass election laws, and lying to the government to conceal that fact. One is legal. The other is a felony.

Add a Comment (For Patrons & Select Persons Only)

This site uses Akismet to reduce spam. Learn how your comment data is processed.