‘Stage 1’ is now drawing to a close. During this stage we recruited a team of Expert Volunteers who spent the spring of 2024 transcribing a set of high-resolution photographs of sample pages from wills made between 1540 and 1790.

At the time of writing our team of Expert Volunteers have transcribed 414 pages of wills, consisting of 26,199 lines of text. This is an enormous achievement and a significant milestone in the progress of our project. We would like to offer a very warm thank you to each of our Expert Volunteers for all of their work:

The expert transcriptions produced by these volunteers form the ‘ground truth’ that we are using to ‘teach’ or ‘train’ a Handwritten Text Recognition (HTR) model which can transcribe wills automatically. The next step is to apply the model to our entire sample of wills, enabling us to auto-generate 25,000 will transcriptions.

VOLUNTEER ACTIVITIES

Our Early Modern Research Fellow, Emily Vine, has been working closely with the volunteers during this period to familiarise contributors with the technology we are using, and our unusual transcription conventions.

Many of our volunteers have been transcribing directly onto, and learning how to use, the website ‘Transkribus’. They’ve also been the first to see some of the early outputs of the Transkribus HTR model, and as part of this process, they’ve checked and corrected a further 62 pages of HTR-generated text.

Emily and other members of the project team also answered queries on a dedicated online Forum and met with the volunteers in online workshops. The latter allowed us to introduce our experts to likeminded researchers, to troubleshoot transcribing problems, and to explain in more depth how the volunteers’ contributions fitted into our broader project aims and research questions. Many volunteers were also interested to learn more about HTR, and to consider its potential use in their own historical research.

We were delighted that some of the volunteers were also able to join us in person at one of two in-person ‘What’s in a Will?’ workshops at The National Archives and the University of Exeter. Not only did this give the project team the chance to meet them in person and to find out more about the volunteers interest in wills, it also meant that they were able to share their knowledge and experience of working with HTR with other workshop attendees.

The scale of our project is such that it would not be possible without volunteer involvement, but we intend that this co-creation of data is merely the starting point for broader dialogue with volunteers across the life of the project. So look out for some volunteer blog posts in the future, as well as more events and workshops.

• An archdeacon’s fussiness about the book he was leaving to a library; the bequest of a lead cistern; a necklace with specified number of pearls.
• A London druggist in 1660s left bequests to very many named druggists in the city; a positive roll-call of the profession.
• A massive number of financial bequests in a will that then turned out to be that of Princess Amelia, daughter of King George.
• With regard to goods, adjectival descriptions always make them much more interesting, whether simply described by colour, or more specifically, like a tortoiseshell chest of drawers and “the great Tortoise looking Glasse”, “my Vallence for a Bedd wrought with Tent-stitch” and “my ffanne with a Gold Handle”.
• The ‘sheeps lockes’ – presumably ‘unprocessed’ wool straight from the sheep – left to two of his servants – the female servant got the best sort and the male servant the ‘worser’!

The 25,000 will transcriptions generated by the Handwritten Text Recognition Software must be checked carefully for accuracy and will require correction. Fortunately the formulaic nature of wills (such as repetition of the term ‘Item I bequeath to’) means that we can auto-correct large parts of the text.

What we need help with are the less common words or sentences in the wills – for instance the names of people and places; amounts of money and the items bequeathed. A great benefit of our process is therefore that our volunteers’ efforts will be targeted at those parts of the wills that are most interesting to researchers.

What exactly will this larger volunteer community be doing? Volunteers will contribute to the project on the ‘Zooniverse’ platform. There volunteers will be shown images highlighting the words or sentences from the wills that need checking, alongside the auto-generated transcription that they will be asked to verify, correct or transcribe.

The Zooniverse Community will also be able to access instructions and advice and to see transcription conventions in the same place. We will also run a Forum that will allow volunteers to raise queries with the project team, or to chat with other volunteers about challenges or the interesting discoveries they have made. We plan to engage with our community not just by providing regular updates on progress, but also by setting transcription challenges and holding online and in-person ‘Transcribathon’ workshops where volunteers transcribe while learning more about the key research questions and aims of the project team.

Zooniverse volunteers will therefore be able to hone and improve their palaeography skills at the same time as making a contribution to the success of a cutting-edge research project. They will be able to contribute as much or as little time as is convenient, and they can fit transcription tasks in whenever they have a moment – on the bus or waiting for the kettle to boil. If you are bored of Wordle but have some knowledge of palaeography, then the Wills Project is just the thing!

Stage 2 hasn’t been launched on Zooniverse yet, but if would like to be alerted when the project goes live, please sign up to our mailing list and you will receive updates via email.