Commons:British Library/Mechanical Curator collection/georeferencing campaign

Shortcut: COM:BL_MAPS Wikimedia and the British Library are trying to georeference over 50,000 old maps and plans found in the Mechanical Curator collection, a set of 1 million images extracted from digitised 19th century books.

This page explains why, and how volunteers with a little time can help. Once the georeferencing is in place, this should clear the way for the images to be uploaded to Wikimedia Commons en masse, with reasonably accurate automatic initial categorisation.

See this progress page for latest progress information, and links to images available to georeference.

Questions and answers edit

What is georeferencing? edit

The basic principle of the georeferencing is to find geographical points in common that can be found in both the digitised image and on a current map. With enough points, this makes it possible to view the old map laid on top of the new one.

 

Using the georeferencing application it is easy to add points, or move existing points around. For more detailed instructions, see the georeferencing status page, which also contains links to the images available to georeference, broken out into approximate subject areas according to the titles of the books they came from.

The image above shows georeferencing points added by User:Basvb for a plan of the city of Haarlem in the Netherlands as it was in 1750. Saving the points gave this map overlay, which can be faded up or down and zoomed in or out, allowing one to explore for example how the street layout has changed or stayed the same, and how the present zig-zag shape of the canal exactly reflects the water defences around the city wall as they were in the 18th century.

How can I browse images that have already been georeferenced? edit

In parallel to the georeferencing status page the corresponding georeferencing done page gives an up-to-date count and links to the images that have already been georeferenced -- click on any of the "georef done" templated links with the solid gold background to see images from a particular (book-level) subject area; then click through to the Flickr page for an individual image, and scroll down to find a link to see the image in Georeferencer, superimposed on a modern map.

 
Locations of the centres of the 3000 images georeferenced in the pilot phase of the project.

On the BL site there is an updated current version (with clickable links).

 
Barchart illustrating the OSM logarithmic size scale from 1 (whole world) to 5 (metropolitan France) to 12 (central Paris) to 18 (a cathedral just north of Paris), for the same 3000 map pilot images

Alternatively, by using the information gained from the georeferencing, it is possible to characterise the images much more precisely, now based on the image itself rather than just the title of the book. The map on the right shows the centre points for the first 3000 images. The BL site has an updated current larger zoomable version with clickable links.

The location and scale information from the georeferencing have also made it possible to add hierarchical descriptive Flickr tags to describe each georeferenced image, giving yet another way to search the georeferenced images. Images of a particular scale and location can be retrieved using URLs such as:

(where the names of the tags for continents, countries, states, regions, cities, etc. using have been derived from the names used for particular levels of addresses returned by OSM's Nominatim service).

What sorts of images are in the collection? edit

This British Library blog post discusses some of the images that were georeferenced in the pilot run last year.

The map and plan images came from 60,000 volumes that were scanned which came mostly from a part of the stacks that contained books in the areas of geography, geology, travel, ethnography, international history -- so everything from guidebooks to local histories to school atlases to manuals of navigation to popular accounts of exploration, giving a comparably wide range of maps and plans of the world as Victorians knew it, as well as history as they understood it. So there are all sorts of material, from plans of cathedrals and castles and battlefields, to town plans (and reproductions of earlier town plans), to atlases and encyclopedic works on geography, to maps of individual journeys of exploration.

About a quarter of the images of maps and plans relate specifically to the UK, the rest relate to countries and places from all over the world. Really the best thing is to have a browse, via the image links on either the georeferencing status page or the georeferencing done page.

The non-map images are also well worth a look, which can be browsed by book via the index pages linked from the above two pages. Books with (non-map) images that have been uploaded to Commons are identified by links in the lists on a green background. In all about 25,000 images have been uploaded to Commons so far, as of May 2015.

When will the map and plan images get uploaded to Commons? edit

The aim is to ultimately get images being uploaded to Commons within minutes of their being georeferenced. However, before starting bulk uploads, it's worth being able to make initial Commons categorisations as accurately as possible. The methods currently being used for the Flickr tagging are working, but there is still some refinement that would be desirable.

(The slides from this presentation about the project, given at GlamWiki 2015, cover some of the still to-do points, from slide 40 onwards).

However, it should be possible to start test uploads of image cases that are easiest for both identification and Commons categorisation -- in particular, old maps of whole countries -- very soon.

Further resources edit