Imagine it’s 1800 and you’re writing a letter. You want to keep the contents private, but mass-produced envelopes don’t yet exist (they’ll come along in the 1830s). So what might you do?
You’d likely use a technique called letterlocking, which involves taking your flat piece of paper and turning it into its own envelope. You could choose from a series of manipulations, including both folding and adding an adhesive like sealing wax. An everyday activity for centuries across cultures, borders, and social classes, letterlocking plays an integral role in the history of secrecy systems. You might think of it as the missing link between ancient techniques and modern digital cryptography.
As the conservator for the MIT Libraries, my job is to preserve the knowledge that lies in physical items. I first encountered so-called locked letters in 2000, as a postgraduate Kress fellow in what were then called the Vatican Secret Archives. Faced with these precious and unique documents, I wanted to understand how they’d been locked in the first place, but obviously I couldn’t go around refolding the originals. So I took some non-historic paper, cut it to the same size as the original, and mapped into it details like folds, slits, tears, and discoloration before turning that paper into a model of the original. I’d make the packets and open them back up to see if what I made matched the flat letters I was looking at in the collections. The models were an excellent way to learn that what resembled damage (cut-off corners, slits, folds, or dirt) was actually evidence of the technology. Once I’d made several of these for a single document, I began to really understand it, and I started noticing things I would never have seen otherwise.
Ultimately, our team created the first systematic catalogue of letterlocking practices by studying the folding sequences and locking mechanisms found on 250,000 opened letters. These letters came from around 60 collections, mainly in the United States and Europe, including some of the major institutional archives and many specialist and private ones. We’ve drawn on letters from about 20 countries across a 650-year time span, so this data set provides a widespread look at letterlocking techniques.
The field of letterlocking didn’t officially exist until March of this year, when my colleagues and I published our findings in Nature Communications, but it has really been in development for decades. I began to notice certain details on letters and other historical documents and began keeping records of them. After my colleague Daniel Starza Smith and I began working together, we systematized this information, developed a language for describing it consistently, and started to articulate why we thought our findings were important. I could see the value of this work to conservation; he came at it from a literary and historical perspective.
People from diverse countries, periods, cultures, and walks of life had infinite ways to reduce a rectangular sheet of paper into a small letterpacket that looked like a modern-day envelope.
We wanted to encourage people to use noninvasive techniques and preserve details that might seem inconsequential at first. What we really needed, we realized, was a collection of historic unopened letters that would remain unopened so we could preserve and study the evidence, and categorize the various methods of letterlocking.In 2016 we found what we needed when we learned of the Brienne Trunk, which belonged to a 17th-century postmaster and postmistress in the Netherlands. It contained an archive of 2,600 locked letters from all over Europe that had never been delivered, including more than 575 that had never been opened. With Rebekah Ahrendt, Nadine Akkerman, and David van der Linden, we created what we called the Signed, Sealed, and Undelivered team to explore the collection.
The next step was to find imaging and coding experts interested in virtually unfolding the letters to read their folds as well as the words on the page. We wanted to unlock the letters digitally.
To accomplish this, first the letters were transported from The Hague to London, where they were scanned by x-ray microtomography experts David Mills and Graham Davis at Queen Mary University of London, using equipment originally developed to examine teeth. The data from these scans was sent over to MIT, where algorithm experts were working to develop the virtual unfolding technique they refer to as a pipeline. Our algorithm engineers (Holly Jackson ’22 and Amanda Ghassaei, MA ’19, with supervision from Erik and Martin Demaine and Neil Gershenfeld) used the scanning data created by the dental scanning experts at Queen Mary. They virtually unfolded the letters so the paleography team could read the words, and so Starza Smith and I could read the creases to better understand the locking patterns.
We discovered that people from diverse countries, periods, cultures, and walks of life had infinite ways to reduce a rectangular sheet of paper into a small letterpacket that looked like a modern-day envelope when viewed from the outside but had a huge variety of crease patterns on the inside. Before 1830, when the modern envelope was introduced, letterlocking was pretty much the only way we’d developed to send correspondence since we stopped using cuneiform tablets 4,000 years earlier.
One of the coolest things about letterlocking is that it allowed letter senders to build security into their mail. My colleagues and I found that some letter writers used more than one type of letterlocking and intentionally built tamper-evident characteristics into some letters and none into others. The specific folding configurations and letterlocking mechanisms were obstacles to spies or others trying to break in, because by design they needed to be damaged to grant access to the letter’s contents. Someone who received a letter and saw damage would know that the message had been compromised.
Our project was developed during a period of intense public debate about global communications systems, the role of state interception, and the nature of privacy. Letterlocking shows that these questions have mattered to people for hundreds of years—and allows us to study what they did about it.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
How to fix the internet
If we want online discourse to improve, we need to move beyond the big platforms.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.