Research Assistant Posting 1

Position 1:

Who We Are:

The Digital Tokugawa Lab is a digital humanities lab focusing on early modern Japanese history. For detailed description of the team and our project, please see link.

Your Role:

Establishing village boundaries on ArcGIS

Over the summer, a collaborator of the lab ran a neural net on a corpus of about 1500 maps from the General Staff Office, surveyed between ca. 1890 and ca. 1940. These are the most highly resolved maps of Japan that have full coverage, down to individual trees and wells and power lines. The neural net excerpted the municipal (町村) boundaries at the time. It worked amazingly well, but the boundaries have gaps. Mostly, they also appear on the original maps, for example when a boundary coincided with a river and the mapmakers decided that it would be obvious to readers that it kept following the river. There are also small gaps where labels obscure the boundary. The nnet was generally better than humans at not overlooking boundaries, but there are a cases where it missed a boundary that human eyes can discern. We now need to close those gaps manually. It is the sort of work one can do while listening to an audiobook. It can be pleasantly hypnotic in our experience. The maps are also very interesting to look at, too. They are black and white, but show individual footpaths and fields and buildings, so looking at them may make you to feel like a color-blind bird winging its way across the landscapes of Imperial Japan. You will also learn a lot about how villages and towns were organized. Some had exclaves or enclaves. Others had fantastically complex boundaries, for example following the banks of a stream while the higher ground already belonged to another village.

Once we have closed the gaps, we will divide these municipalities into the Edo-period villages, based on other maps that we have collected. They are often beautiful, but the quality of the survey often falls far short of modern standards. As a result, we have georeferenced them heavily. Your task would be to decide between different lines you see in the landscape and trace them.

Other tasks are likely to arise in the course of the project, including literature searches at SML.

Time commitment

We expect our RAs to put in at least 10 hours a week on average, though more may be assigned if the RA wants to put in more hours. For students on a gap year, we are happy for them to work with us full time.


- Ability to read or recognize Japanese or Chinese characters is an advantage. Advanced Japanese would be an even greater advantage. But neither are required.

- Attention to detail.

- Interest in digital humanities, history, and/or historical geography.

- Basic knowledge of Japanese history would be a plus.

We will teach you how to work with the specific functions in ArcGIS Pro.

Down the line, other tasks may come up within the project, but the two mentioned here will take hundreds of hours, likely spread over a several RAs.


We won’t be able to offer co-authorship on our project. But we hope that you will learn skills and facts worth learning when you work with us.

The pay would be $15 hour.


To apply, please send us a resume, a list of courses you have taken at Yale, and a brief statement of the nature of your interest and your relevant background (around 150-300 words). Please email these materials to

We will begin reviewing applications on Dec. 17th.