Unknown Tobler W. Snow’s cholera map coordinate system (1992)

Question

The digitization of Snow map by Waldo Tobler dates from 1992 !

The original files can be found at https://web.archive.org/web/20060615025551/http://ncgia.ucsb.edu/Publications/Software/cholera/.

Here are the coordinates of the pumps located in London's Soho district.

8.6512012 17.8915997
10.9847803 18.5178509
13.3781900 17.3945408
14.8798304 17.8099194
8.6947680 14.9054699
8.8644161 12.7535400
12.5713596 11.7271700
10.6609697 7.4286470
13.5214596 7.9582500
16.4348907 9.2521296
18.9143906 9.7378187
16.0051098 5.0468378
8.9994402 5.1010232

• I used the following website to try to find a matching SCR without success. https://app.dogeo.fr/Projection/#/coords-to-points

• If the scr is unknown, what's the best way to automatically reassign coordinates to points (their location being known) ?

Any idea ?

Answer 1

The coordinates are probably not in any commonly-defined coordinate system, but something specific to that data set.

Its possible the (0,0) point is at the bottom corner of the original map, or the location of the pump, or some other well-known point that the distances were measured from.

The documentation from the archive.org link says:

"Scale of the source map is approx. 1:2000 Coordinate units are meters."

If the coordinates are metres, then if you can find out where the system origin (0,0) point is in EPSG:27700 (UK National Grid) coordinates, then adding them to that will get you very close to the locations in UK grid coords, and even closer if you can ascertain any difference between UK grid North and the data coordinate North, correcting with a rotation.

These aspects of the data are often not well-specified in datasets from around that time because it wasn't thought that people would be wanting to have these datasets georeferenced - GIS wasn't as easy in the early 90s - and the only application was to do some simple analysis on the point pattern, which is independent of coordinate system (on this scale). Sometimes even true geographic coordinates were modified to fit data sets into unit squares so that mathematical precision wasn't lost when subtracting two large and similar numbers.

FWIW, there's plenty of other digital versions of the Snow data out there, many possible derived from the same source you have here, and that might help you get the coordinates of the origin (0,0) - possibly those sources may also say their original source and how they've georeferenced them.