You’re reading Age of Invention, my newsletter on the causes of the British Industrial Revolution and the history of innovation. This edition went out to over 24,000 subscribers. To support my work, you can upgrade your subscription here:
I was quite overwhelmed by the response to my last piece, on whether history has a reproducibility crisis — all the more overwhelmed because I posted it just before moving house. But I’ve been sent so many interesting things as a result of it, that I’d like to share a few of them that stood out. And to make a public commitment.
The concern I expressed in the piece is that the field of history doesn’t self-correct quickly enough. Historical myths and false facts can persist for decades, and even when busted they have a habit of surviving. The response from some historians was that they thought I was exaggerating the problem, at least when it came to scholarly history. I wrote that I had not heard of papers being retracted in history, but was informed of a few such cases, including even a peer-reviewed book being dropped by its publisher.
In 2001/2, University of North Carolina Press decided to stop publishing the 1999 book Designs against Charleston: The Trial Record of the Denmark Vesey Slave Conspiracy of 1822 when a paper was published showing hundreds of cases where its editor had either omitted or introduced words to the transcript of the trial. The critic also came to very different conclusions about the conspiracy. In this case, the editor did admit to “unrelenting carelessness”, but maintained that his interpretation of the evidence was still correct. Many other historians agreed, thinking the critique had gone too far and thrown “the baby out with the bath water.”
In another case, the 2000 book Arming America: The Origins of a National Gun Culture — not peer-reviewed, but which won an academic prize — had its prize revoked when found to contain major errors and potential fabrications. This is perhaps the most extreme case I’ve seen, in that the author ultimately resigned from his professorship at Emory University (that same author believes that if it had happened today, now that we’re more used to the dynamics of the internet, things would have gone differently).
It’s somewhat comforting to learn that retraction in history does occasionally happen. And although I complained that scholars today are rarely as delightfully acerbic as they had been in the 1960s and 70s in openly criticising one another, they can still be very forthright. Take James D. Perry in 2020 in the Journal of Strategy and Politics reviewing Nigel Hamilton’s acclaimed trilogy FDR at War. All three of Perry’s reviews are critical, but that of the second book especially forthright, including a test of the book’s reproducibility:
This work contains numerous examples of poor scholarship. Hamilton repeatedly misrepresents his sources. He fails to quote sources fully, leaving out words that entirely change the meaning of the quoted sentence. He quotes selectively, including sentences from his sources that support his case but ignoring other important sentences that contradict his case. He brackets his own conjectures between quotes from his sources, leaving the false impression that the source supports his conjectures. He invents conversations and emotional reactions for the historical figures in the book. Finally, he fails to provide any source at all for some of his major arguments
But I think there’s still a problem here of scale. It’s hard to tell if these cases are signs that history on the whole is successfully self-correcting quickly, or are stand-out exceptions. I was positively inundated with other messages — many from amateur historical investigators, but also a fair few academic historians — sharing their own examples of mistakes that had snuck past the careful scholars for decades, or of other zombies that refused to stay dead.
Take this charming story of Euler embarrassing Diderot at the court of Catherine the Great, by “proving” God existed with some algebra. Once the layers of accumulated error through repetition have all been peeled back by tracing the citations back, then even if the original story is to be taken at its word, it doesn’t seem as though Diderot was bamboozled, or even that Euler was involved at all! This particular story has been periodically debunked by scholars since at least the 1950s, but it continues to shamble on.
Or take this case, shared by the military historian Eamonn O’Keeffe, about the Duke of Wellington drinking “to the corpse of India” when he heard of the death of Tipu Sultan in 1799. When tracing the citation back, O’Keeffe discovered that it actually comes from the dialogue in a work of historical fiction! By being cited and re-cited through increasingly prestigious sources, and accruing even more errors along the way, a line in a novel had somehow achieved the status of historical fact — an astonishing case of citogenesis, as the comic writer Randall Mundroe (xkcd) calls it.
Many cases of persistent error may perhaps seem minor or pedantic, but some of the examples I was sent appear to have serious implications (and besides, if historians aren’t pedantic, what are we?) Among the serious cases, take the concern expressed by a group of scholars in New Zealand — Michael Stevens, Te Maire Tau, Atholl Anderson, Puamiria Parata-Goodall, and Tā Tipene O’Regan — about an article that claimed Polynesian explorers in the seventh century had sailed into Antarctic waters and “perhaps even the continent”, which was widely reported on at the time. They’re concerned that the publication of work that is not careful or systematic, and which then ends up being repeated and exaggerated uncritically by the media, ultimately threatens to discredit the integrity of Māori tradition as a historical source — and thus will undermine the diligent work of generations of scholars who have been trying to improve its credibility. You can read their fuller argument here.
Now, some historians contended to me that the scholarly world is functioning just fine, but did grant that there may be problems in so-called popular history — that is, history not done by professional historians, and for wider audiences. I’m not convinced by this distinction, quite frankly. The worst cases of all are when trust-conveying labels like “peer-reviewed publication” provide cover for major errors or worse. Such work instantly crosses into the popular realm when it gets picked up by the press. And even if the distinction is meaningful — and perhaps I’m a bit biased here, straddling the two worlds — then I still think it’s a problem worth trying to solve. We should still develop better incentives and institutions to combat historical inaccuracy, regardless of who commits it. The more tools at our disposal, the better.
And even if you still think I’ve over-exaggerated the problem, then there’s a case to be made for making history more open anyway. As the science writer Ananyo Bhattacharya — author of a fascinating book about John von Neumann — put it in a comment: “what you’re proposing here is an ‘Open History’ initiative, much like many funders and fields in science are now committed to ‘Open Science’ … Not only would this help correct the historical record, it would be a potential treasure trove for other historians and writers”.
Yes. Even if you think history is rooting out the bad work just fine, we should still be doing more to make our sources accessible. I have been a huge beneficiary of the work that’s already been done in this regard. For primary sources I use resources like Early English Books Online almost daily, as well as Google Books, the Internet Archive, and various digital genealogical and newspaper collections. Had it not been for the Digital Humanities Institute’s work in digitising and transcribing the letters by and related to the seventeenth-century intelligencer Samuel Hartlib, I would never have discovered forgotten links in the history of the steam engine. And when the Hartlib papers in one case contained a transcription of a letter I suspected had been misattributed — a letter I consider a smoking gun — the fact that the Beinecke Rare Book and Manuscript Library had high-quality images of it online meant that I could instantly examine the handwriting and confirm my suspicions. Research that would once have taken months, if not years, can be performed in mere minutes.
I was sent many more examples of initiatives that we might call Open History: an online repository of datasets to do with the history of early North America; a repository of transcripts of primary sources related to slavery in the British Empire and the newly independent United States; and a new peer-reviewed journal called The New American Antiquarian, which invites the publication of sources to do with the pre-1825 Americas: transcriptions from unpublished manuscripts, new translations of texts into English, and more.
And I’ve been inspired by what some scholars have been doing by themselves. You can, for example, view the photographs of various archival sources taken by Kurt Schuler for his book on a precursor to the 1944 Bretton Woods conference. He has even assembled a repository of various other primary sources relating to the conference. And the military historian Bruce Ivar Gudmundsson has been publicly logging the errors that can be found in his published works.
Inspired, that is, to practise what I preach. From now on, I will start maintaining a public log of corrections I’ve made to previous works (poetically, the first on the list will be that I accidentally wrote “faster” rather than “slower” in the opening paragraph to the previous piece all about how there are too many errors!) And I will begin to upload to somewhere freely accessible the transcripts and notes that I use or cite in the work that I publish — initially for future works, and eventually going back over previous work — as well as maintaining a public inventory of what photographs of sources I have, so that people can get in touch to request them. And, of course (I always intended this), when I publish my book on inventors of the British Industrial Revolution, I will make the underlying database freely and easily accessible online too.
This is just a start. Suggestions of more Open History problems, best practices, and examples welcome.
If you’d like to support my work, please consider upgrading your free subscription to a paid one here: