Skip to content

Mining Census

Production, employment, and financial statistics from South African mining census publications.

Unit of Observation

Individual mine (for gold producer tables) or mineral type (for financial statistics). This is not magisterial-district-level data — mines are identified by name and mining region (OFS, Far West Rand, West Rand, East Rand, Klerksdorp).

District-level aggregation

17 magisterial districts contain NRC member mines (see teba_panel/nrc_member_mines.csv). A mine-to-district crosswalk for the full mining census has not yet been built.

Files

67 CSVs in mining_census/mining_completed_converted/ covering:

Content Years Files
Statistical tables (production by mine) 1964-1987 ~50
Census of Mining (comprehensive) 1978, 1981 2
Gold production and pricing Various Several
Coal analysis tables Various Several
Chamber of Mines membership Various Several

Gold Producer Panel

Output: mining_panel_1964_1987.csv (extended April 2026)

  • 696 rows covering 17 years (1964, 1966, 1969-1979, 1981, 1983, 1985, 1987)
  • 30-57 mines per year (declining as mines were consolidated or closed)
Variable Description
year Year
ref_no Mine reference number (WARNING: reassigned between years — match by mine_name)
mine_name Mine name
ore_milled_1000t Ore milled (thousand metric tons)
gold_kg Gold output (kg; pre-1970 converted from troy ounces)
working_revenue_r1000 Revenue (R thousand; pre-1970 converted from Rand)
working_costs_r1000 Costs (R thousand)
working_profit_r1000 Profit (R thousand)

Checks (April 2026)

  • National totals agree with the separately published Financial Statistics series in every overlapping year, 1966–1975 (a script comparison)
  • 1966: two ore-milled cells carry an extra digit (Grootvlei, Rand Leases) and the 1966 gold ounces are unreliable; these are proposed rows in corrections/mining_census.csv, not yet applied by a builder
  • "DR" prefix in CSV data means debit (negative value). Affects net_sundry_revenue column. Not a transcription error.
  • Mine reference numbers are reassigned between publications — cross-year analysis must match on mine name, not ref_no.
  • Boone's 17 mine-level DTAs (_boone_mine_data/) come from the same source spreadsheets as these CSVs, so they are not an independent check

Financial Statistics (extracted 2026-03-29)

5 markdown files with revenue, costs, profits, employment, and wages by mineral type:

File Coverage
Mining-FINANCIAL_STATISTICS_1966-1969_extracted.md 1966-1969
Mining-FINANCIAL_STATISTICS_1970_extracted.md 1966-1970
Mining-FINANCIAL_STATISTICS_1971_extracted.md 1971
Mining-FINANCIAL_STATISTICS_1972_extracted.md 1972
Mining-FINANCIAL_STATISTICS_1973_extracted.md 1973

Extraction method: Gemini Vision API. Note: 1974 is missing (publication gap between 1973 and 1975 editions).

Provenance

  • Source: Census of Mining / Statistics of Mining, Department of Mines / CSS
  • Digitization: Rao's Infocube (March–April 2013) for the statistical tables; the vendor flagged poor scans for 1975, 1978, 1983 and 1985 (46 pages). Gemini Vision API (2026-03-29) for the Financial Statistics extractions, not page-checked
  • Source PDFs: raw_archives/01_production_censuses/mining/source_pdfs/