Skip to main content

Build a research log

Historical GIS becomes difficult to audit when search results, PDFs, screenshots, notes and digitised points are separated from one another.

A useful research log keeps three things connected: what you searched, what the source actually says, and what geographic interpretation followed.

1. Search log​

Use this when research involves repeated archive, database or document searches.

search_id
date
source_system
query
filters
results_notes
useful_record_ids
next_search

This helps avoid repeating the same search months later and makes failed searches useful evidence rather than forgotten effort.

2. Evidence table​

Use one row for each source item or relevant passage.

evidence_id
source_type
repository
record_id
title
source_date
page_or_plan
name_as_source
source_wording
spatial_clue
matched_name
match_source
match_source_id
location_status
confidence_note
file_or_url
checked_date
next_action

Keep name_as_source and source_wording as close to the original as practical. Put modern or normalised names in separate fields.

Give any GIS feature a stable ID and link it back to one or several evidence rows.

A simple structure is:

place_id
display_name
location_type
location_status
geometry_source
primary_evidence_id
notes

If several evidence records support one feature, use a related table rather than squeezing several IDs into one text field.

Location status​

Use words that describe the research state rather than pretending certainty can be reduced to a number.

Useful values include:

  • unlocated
  • candidate
  • approximate
  • supported
  • conflicting
  • rejected

Define the terms once for the project and use them consistently.

Confidence should describe the evidence​

Prefer notes such as:

  • exact surveyed plan control
  • Gazetteer match supported by nearby named awa and block
  • approximate locality from documentary description
  • candidate match based on name only
  • conflicting locations across two plans

A generic 1–5 score is less useful unless the project defines what every number means.

Keep evidence and interpretation apart​

A good record distinguishes:

EvidenceInterpretation
source labels a kāinga beside a named awalikely corresponds to the locality shown on a later plan
track is visible on 1963 aerial photographmay represent an historical access route
Gazetteer contains a similar official namepossible modern match, not proof of identity

This makes disagreement easier to manage because another researcher can challenge the interpretation without losing the source evidence.

No geometry is a valid result​

Do not invent coordinates to complete a table.

A historical reference can remain useful when it is:

  • unlocated
  • associated with a broad locality
  • linked to several candidate places
  • represented only in the research table

The research log should contain both mapped and unmapped records so GIS does not silently discard the uncertain evidence.

A practical project pattern​

01_source/        PDFs, plans, imagery, source exports
02_working/ research tables, georeferenced copies, derived GIS
03_project/ QGIS project and notes
04_outputs/ maps, tables and published extracts

Keep original source files unchanged where practical. Record important processing using the common Provenance and source checking pattern.

Use it with the wider workflow​

Use these structures directly in a spreadsheet, database or GeoPackage. There is deliberately no generic placeholder download.

Last reviewed: 4 September 2026