We are in the process of checking the arithmetic of the Bills of Mortality, both its internal consistency as well as the accuracy of our work, and are making our Jupyter notebooks of our analysis public. The notebooks take into account transcription errors, printing mistakes, illegible data, or duplicate data to capture a comprehensive analysis of the data.

Checking for duplicate transcriptions: This notebook analyzes duplicate transcriptions in the Bills of Mortality dataset to estimate the overall accuracy of our transcribed data. It does not differentiate between data that was mistranscribed and data that was printed differently in two separate copies of the same bill of mortality; both are considered “inaccurate” data.

This visualization shows the data quality patterns for Bills of Mortality records by parish and week. You can examine either illegible records (difficult to read or transcribe) or missing records (incomplete data) to understand temporal and geographic patterns in data quality issues.

This visualization illustrates the seasonality of causes of death across the year, allowing you to analyze patterns and compare different causes. You can select a single cause to view its seasonal distribution, or compare two causes side by side. Dashed lines indicate areas where there are gaps in the data, and represent an interpolation between existing data.

Interactive choropleth map analyzing geographical patterns of suicides, drownings, and killings in London parishes, with filtering options by cause and year range based on text analysis research.

Multiple sparkline visualization showing temporal patterns of parish deaths with data normalization options (log10, normalized) and filtering by burial type (burials, plague, both) for improved visibility of smaller count variations.

Interactive choropleth map of London showing burial and plague death patterns across parishes, with controls to filter by year range and count type for geographical mortality analysis.

Heat map visualization showing causes of death over time from annual general bills, with darker red shades indicating higher death counts for each cause by year, updating regularly as new transcriptions are added.

Interactive calendar visualization showing weekly causes of death data for any selected year, allowing users to explore temporal patterns in mortality.

Histogram visualization displaying weekly mortality data by specific causes of death, with interactive controls to filter by year and individual causes.

Interactive word cloud visualization displaying causes of death from the Plague Bills, with text size representing frequency of occurrence and year range filtering capabilities.

Heat map visualization showing causes of death over time from weekly bills, with darker red shades indicating higher death counts for each cause by year, updating regularly as new transcriptions are added.

Dashboard visualization showing transcription progress with bar charts indicating total weeks versus transcribed weeks by year, updating regularly as new bills are added to the database.

We are in the process of building out a data API to support the data work we’re undertaking with the transcription of the plague bills. We anticipate hundreds of thousands of rows of data by the end of our transcription process, and we wanted an easy and efficient way to work with that data.

As part of our work in data-driven historical research at RRCHNM, we are building a data API to store and access data from databases. Following the process of transforming the DataScribe transcriptions into tidy data, the resulting data is uploaded to a PostgreSQL database where we can take advantage of relational connections among the different datasets we’ve compiled.

One of the ways we are using the transcribed bills of mortality is in data visualization and mapping, in an effort to ask new questions and revisit old ones.

At the Southern History Association’s annual meeting in Baltimore, we presented preliminary work on data visualization and the data API. An interactive notebook on this early work is available on Observable for perusal (note, the page may take a moment to load the 100,000+ records). You may also make a copy of this notebook to your own Observable account to edit.