Over recent years we’ve been treated to a series of fascinating advances in the world of x-ray imaging, as  researchers have developed their x-ray microtomography techniques and equipment to the point at which they can probe and then computationally reconstruct written material within objects such as letters or scrolls in museum collections whose value or fragility means they can’t be opened and read conventionally. There is more to this challenge than simply extracting the writing though, in addition to detecting the ink the researchers also have to unpick the structure of whatever it was written upon. A particular challenge comes from letterpackets, the art of folding a letter into its own envelope, and a newly-published Nature Communications paper details work from a team of academics in the USA, the UK, and the Netherlands in tackling it.

Letterpackets were more than a practical method of packaging a missive for the mail, they also had a security function often called Letterlocking. A packet would be folded in such a way as to ensure it was impossible to open without tearing or otherwise damaging the paper, and their structure is of especial interest to historians. The researchers had a unique resource with which to work; the Brienne collection is a trunk full of undeliverable mail amassed by a 17th century postmaster couple in Den Haag in the Netherlands, and now in the possession of the Beeld en Geluid museum in that city. In it were a cache of letters including 577 never-opened letterpackets, and the x-ray technique promised a means to analyse these without compromising them.

The researchers have developed an entirely computational technique for the virtual unfolding process. Starting with a 3D volumetric x-ray scan of the unopened packet they then identify the various layers of paper and the bright spots which denote the ink. Their algorithm has to cope with areas in which two or more layers are tightly in contact, for example when multiple levels are folded, and then unpick the resulting 3-dimensional mesh into a 2-dimensional sheet. Their process for mapping the crease pattern involves applying a colour map representing the mean curve radius at a given point. The final section of the paper looks at the multiple different methods of letterlocking, and attempts to categorise them all including a security rating for each. It’s evident that this could be a highly personalised process, indeed they give as an example a letter from Mary Queen of Scots that used an intricate spiral folding technique to identify its sender.

It’s clear that this technique will reveal many more fascinating historical documents as it is both refined and extended across the many more collections of further artefacts that have lain waiting for it. As they say, individual letters do not necessarily contain earth-shattering historical discoveries, but taken together they shed an important light on the social history of past centuries.

One of the names on the paper is [David Mills], whose work has featured here before.