A first dataset, step by step

This walkthrough builds a small dataset from one source: a London customs account of 1480 to 1485 that mentions Bristol and Deptford Strand. It uses five of the eight sheets. The finished tables are the customs example, which you can download as CSV files and open beside this page.

Start from the temPlato workbook. Only the columns you fill in are shown below; leave the others empty.

1. List your places

One row per place, with an identifier you make up and a label to list it under. The label is only for display: the names the place has in your sources go in the names sheet.

place_id	label	country_codes
bristol	Bristol	GB
deptford-strand	Deptford Strand	GB

2. List your sources

One row per source. The date is when the source itself was made, as the catalogue gives it; from and to are the years it spans.

source_id	title	citation	uri	date	from	to	licence
tna-e122-19-10	TNA E 122/19/10	The National Archives, Exchequer: King’s Remembrancer, E 122/19/10	https://discovery.nationalarchives.gov.uk/details/r/C5846832	1480-1485	1480	1485
geonames	GeoNames	GeoNames geographical database	https://www.geonames.org/	undated			https://creativecommons.org/licenses/by/4.0/

The second source, GeoNames, is there because the coordinates in step 4 come from it, not from the customs account. Every row says which source it comes from, so a location taken from a modern database is never mistaken for one found in a fifteenth-century document.

3. Record the names

One row for each name a source gives, spelt exactly as the source spells it. place_id and source_id point at the rows you made in steps 1 and 2.

place_id	name	language	script	name_type	date	from	to	source_id	certainty	notes
bristol	Bristowe	enm	Latn		1480-1485	1480	1485	tna-e122-19-10	0.95	Clear manuscript hand.
deptford-strand	Deptford Strand	en	Latn	toponym;odonym	1481	1481	1481	tna-e122-19-10	0.7	Manuscript partially damaged; reading probable but not certain.

The first row says: the customs account, for 1480 to 1485, calls Bristol “Bristowe”. The second says how sure the compiler is of a damaged reading (certainty 0.7) and why (notes). Deptford Strand names both a place and a street, hence two name types.

4. Record locations

Latitude and longitude in decimal degrees, as most web maps give them. The geometry_role says what the point marks: here a point standing in for the whole town, rather than its outline or one particular building.

place_id	latitude	longitude	geometry_role	date	source_id
bristol	51.45523	-2.59665	RepresentativePoint	undated	geonames

GeoNames does not date its coordinates, so the date is undated.

5. Record what kind of place it is

The account treats Bristol as a port. The type can be written in your own words, with the matching concept from a published vocabulary (here the Getty Art & Architecture Thesaurus) if you know it.

place_id	type_label	type_uri	date	from	to	source_id
bristol	port	http://vocab.getty.edu/aat/300000745	1480-1485	1480	1485	tna-e122-19-10

6. Say which records elsewhere are the same place

If you know that one of your places is already in GeoNames, Wikidata or the World Historical Gazetteer, say so in identities. This is what lets your evidence join everyone else’s.

place_id	same_as	match_type	certainty	basis
bristol	https://www.geonames.org/2654675/	exactMatch	0.95	Same name and position; the only plausible match.

What happens to your rows

Each row of names, locations and types becomes one attestation: a statement that this source, for this date, gives this name, location or type for this place. Your four evidence rows become four attestations about two places, each keeping its source, its date and your certainty. Nothing you entered is merged or overwritten, and a user of the combined data can always see who said what, and when.

For a richer example, with manuscripts copied later, “ibid.” references and editorial headwords, see the place-name survey example. Every column is described in the sheet reference.