Disclaimer: I work at the British Geological Survey
I made an internal enquiry on this issue, to see if we had any further information on the Kenya maps, and this answer is a summary of the information received.
First off, it's important to note that these maps were hand drawn in the Kenya office and would have used the Kenya topography maps as their base. Most of the topo maps were drawn in the 60s and 70s by the Directorate of Overseas Surveys using photogrammetry techniques so should have been accurate. They would be using UTM36 and UTM37 as the projection with the datum as Arc 1960, which would give the following projection systems:
- Arc 1960/UTM zone 36N ~ Kenya - north of equator and west of 36°E; Uganda - north of equator and east of 30°E.
EPSG::21096
http://www.opengis.net/def/crs/EPSG/0/21096
http://www.epsg-registry.org/export.htm?gml=urn:ogc:def:crs:EPSG::21096
- Arc 1960/UTM zone 36S ~ Kenya - south of equator and west of 36°E; Tanzania - 30°E to 36°E; Uganda - south of equator and east of 30°E.
EPSG::21036
http://www.opengis.net/def/crs/EPSG/0/21036
http://www.epsg-registry.org/export.htm?gml=urn:ogc:def:crs:EPSG::21036
- Arc 1960/UTM zone 37N ~ Kenya - north of equator and east of 36°E.
EPSG::21097
http://www.opengis.net/def/crs/EPSG/0/21097
http://www.epsg-registry.org/export.htm?gml=urn:ogc:def:crs:EPSG::21097
However
- To draw the outline of the maps and add the UTM grids correctly the cartographers would have had to extract the XY grid locations for each sheet from a surveyors handbook, if they had one (BGS were using a really old American one in Zimbabwe and Zambia). If they didn't then they might have just overlaid their plastics over the old paper maps to draw the grids. This could easily cause errors in the grid line locations. BGS have found this sort of shift in other African maps of the same era. It may be that they didn't even use Arc 1960 and used WGS72 or something else.
- Even if they had the handbook for the grid locations they might not have drawn the grid lines correctly just because they were hand drawing the maps.
Advice
The advice therefore is to georeference the scanned maps into your MXD using the UTM grid from the map but then checking the features on the geological map such as rivers and roads match with any correctly located satellite imagery or other correctly referenced data that you already have.
If the grid lines and the topo features on the geology maps don't register correctly with other data sets then you should georeference the geology maps using features on the map and not trust the grids lines.
If you are adding GPS or other WGS84 data to your project in ESRI there are 4 transformations for the Arc 1960 datum (see http://georepository.com/search/by-name/query/arc+1960+/crs_page/1/#crs) for using with WGS84. The one to use for Kenya is the Arc_1960_To_WGS_1984_2 one as this covers Kenya only.
GEOGTRAN["Arc_1960_To_WGS_1984_2",GEOGCS["GCS_Arc_1960",DATUM["D_Arc_1960",SPHEROID["Clarke_1880_RGS",6378249.145,293.465]],PRIMEM["Greenwich",0.0],UNIT["Degree",0.0174532925199433]],GEOGCS["GCS_WGS_1984",DATUM["D_WGS_1984",SPHEROID["WGS_1984",6378137.0,298.257223563]],PRIMEM["Greenwich",0.0],UNIT["Degree",0.0174532925199433]],METHOD["Geocentric_Translation"],PARAMETER["X_Axis_Translation",-157.0],PARAMETER["Y_Axis_Translation",-2.0],PARAMETER["Z_Axis_Translation",-299.0]]
Best Answer
UTM is surely not the right projection. UTM 37 is centered on 39°E, and has a false Easting of 500000 metres to that. South of the equator, UTM 37S is measuring from the south pole to avoid negative values.
Apart from UTM, there was a Cassini-Soldner projection (possibly in feet instead of metres), and a 5 degree wide "East Africa War" transverse mercator CRS used in earlier times.
Using a Cassini-Soldner projection centered on 37°E/0°N with ARC 1960 datum, I get for your point in WGS84:
38.046282° East 0.059779° South (decimal degrees)
located about 43km east of Nkubu town.
Based on the documents linked in Convert Cassini-Soldner Projection coordinates from Kenya to UTM I suggest to use the following proj.4 parameters for a custom CRS:
For the central meridian 37° (valid 36°-38°E)
For the central meridian 39° (valid 38°-40°E)
Ellipsoid parameters are taken from Clifford Mugniers article, Datum shift parameters were computed from the coordinate tables in the document by Gordon Okumu Wayumba