GENUKI : Place names database

Carmarthenshire

Pembrokeshire

Glamorgan

Contents

A Place Name Database

The database is as compiled by Dylan Moore with these exceptions;

Latitude and Longitude data has been omitted.
The x,y co-ordinates provided can be utilised to search on online maps such as streetmap.co.uk & old-maps .
The Ordnance Survey Get-a-map can also be searched using the OS grid ref obtainable from a streetmap.co.uk search result.
Chapman code abbreviations have been added to indicate which county a place is in.
Note that the county will be the ancient county as used on Genuki, e.g there are some border places actually situated in PEM but which are identified as being in CMN here simply to match Genuki's customary practice of having only one parish page even if part of a parish is in an adjoining county - the Lewis 1833 extract on each parish page explains this further.
The number of entries in the database for each county/parish/hamlet included has been added - see the County/Parish breakdown

Accompanying comments by Dylan Moore

This is a database of just over10,000 places in south-west Wales mentioned in the 1891 & 1901 censuses.
The area covered is a band stretching from St David's to the Gower Peninsula [embracing parts of the ancient counties of Pembrokeshire, Carmarthenshire and Glamorgan.]
The [original] database contains Place-name, Parish, National Grid Co-ordinates, Latitude and Longitude.
This has been developed in the course of study of population distributions in the Language Boundary zone.
In addition to some scattered odds and ends outside the central band, it includes about 80 civil parishes for which there is 100% coverage (i.e every census address is listed).

As regards the location of place-names, the 1:10560 County Series 1st Edition is the most comprehensive, and because in many places it was surveyed in or near to 1891, it's particularly good for understanding the 1891 census schedule layout. The 1:2500 map usually doesn't contain any more names, because the 10,560 was mostly just a photo-reduction of the 1:2500 base map. The latter only comes into its own in identifying individual houses in urban areas. A lot of names in the 1st edition were left out in the 2nd Edition (typically 1900-1907) but of course it's useful for places (e.g. the coalfield) where new building was happening. A proportion of places are not explicitly named - probably almost zero in parishes with scattered populations and largish farms, whereas it could be more than 70% in other areas. In the majority of these cases, if the enumerator is helpful and hasn't shuffled the schedules, un-named houses on the map can be interpolated from the sequence of schedules. In other cases I have resorted to later map editions, and in particular the modern OS digital maps often show obscure house names, although so many of these have changed in the last century that this process can be frustratingly unproductive.
In a significant number of cases, places have actually "moved". An old farm may be abandoned, and everything is moved (including the name) to another location. In some cases, I have to admit that attribution is no more than an informed guess. Keeping these guesses in the database is, I would say, justified because it saves the user from making the same, perhaps less well-informed guess. On occasions where I have had the opportunity to check such guesses on the ground, my accuracy is probably about 70%. And if it's not where I said, it's certainly not far off.
Of course, if anybody out there wants to pick holes in it, I'm delighted to update my records.

My work requires that addresses be uniquely identified so that occupants in any census can be matched with it, as can names on the OS maps. It's a database of places: not census schedules.
Spelling in the enumerators' books is anarchic in the extreme, so the database spelling necessarily doesn't always agree with these. I mainly (but not completely consistently) used the map spelling. I realize that the casual user might find problems with this, but there's no way round this, particularly with Welsh names. Danycoed might be Tanycoed or Dancoed, etc, and that only if the book entry is legible. Any looking-up process needs an element of trial and error.