Cut, and then stitched
This project holds 124 boundary acts: 96 subdivisions and 20 consolidations. Sorted by date they tell one story with a turning point. The district cut ground for a hundred and fifty years; from 1970 it has only stitched.
District context rather than family history. It is here because every farm on this site is a piece of it.
The deeds and diagrams gathered here record 124 boundary acts, one parcel becoming another. Ninety-six of them are subdivisions, twenty are consolidations, and the rest are remainders and resurveys.
Sorted by date, they tell one story, and it has a turning point.
For a hundred and fifty years the district only cut
| decade | subdivisions | consolidations |
|---|---|---|
| 1850s | 4 | |
| 1870s | 1 | |
| 1880s | 2 | |
| 1900s | 4 | |
| 1910s | 8 | |
| 1920s | 7 | 2 |
| 1930s | 18 | 1 |
| 1950s | 6 | |
| 1970s | 2 | 4 |
| 1980s | 2 | |
| 1990s | 1 | 3 |
| 2000s | 2 | |
| 2010s | 4 |
The 1930s are the peak, and by a distance: eighteen subdivisions in ten years, more than the previous fifty put together.
Then it stops. There are six subdivisions in the 1950s, two in the 1970s, and one in 1990, and that last one is not really a cut at all, it is a piece falling out of the Farm 215 consolidation as it was assembled. From 2003 onward, every dated boundary act here is a consolidation.
The sign flips in 1970: consolidation overtakes cutting
Take the run from 1970 to today and count both ways: fifteen consolidations against three subdivisions. Before 1970 there are eight consolidations against ninety-three subdivisions.
That is the viable Karoo unit growing rather than a change in fashion. A farm that could carry a family in 1930 could not by 1980, and the arithmetic that had governed the district for a century, divide the ground among the heirs, reversed into the arithmetic that governs it now: put it back together, or leave.
What was being cut in the 1930s, and between whom
The 1930s are also the decade with the most changes of hands anywhere in these papers: fourteen ownership starts, against thirteen in the 1840s, which was the decade of the grants themselves.
But look at what the instruments are. Five Vrolykheid portions to one Marincowitz in 1935–36. Two Marincowitzes taking farms on the same day, 16 May 1934. A four-parcel cluster among siblings in December 1937. Two estate partitions in 1939. Only two instruments in the whole decade took ground out of a family.
So the busiest decade in the district's paper is not a decade of buying and selling. Drought and Depression did not move this ground onto a market; they made families formalise what they already held, write down, in deeds, the arrangement that had been an understanding.
The counter-example is on Botterkraal, where it did not work: six burials between 1927 and 1942, a flock walked home on foot from Colesberg, and a forced sale in 1943. That household has its own page, Six graves on Botterkraal.
The knife followed the first funeral
Of the farms granted on 15 April 1841, the order in which they were first cut is the order in which their grantees died.
Cornelis Floris Johannes Meyer of Spreeufontein died on 21 July 1848, and his ownership of it ends on that same day. Spreeufontein is the first of the grant farms to be subdivided, in 1850.
Middelwater is not cut until 1883. Vrolikheid not until 1921. And Rietkuil was never split at all while Pieter Johannes Swanepoel held it, he died in 1885, and the farm went whole.
Rondabel breaks the row, and usefully. Its first cut is dated to the day: 17 July 1872, diagram 922 of 1872, while its 1841 grantee was still alive - but he had let the farm go four years earlier, and the men who divided it were two living partners ending a partnership, not an estate settling. That division is told on the farm's own page.
A grant farm stayed whole while the man it was granted to was alive on it. Rondabel is the test of the wording: its grantee outlived the first cut by ten years, alive, and not on it. That is not a rule anybody wrote down; it is what the dates do.
The funeral was the land market
Which points at the larger pattern behind all of it.
Twenty ownership spans end on a recorded date with the holder's own death also dated. Seven of the twenty end within a year of that death, and three end on the day itself:
| 21 July 1848 | Cornelis Floris Johannes Meyer | Spreeufontein |
| 26 March 1885 | Pieter Johannes Swanepoel | Rietkuil |
| 13 May 1890 | Gerolm Marincowitz | Vrolikheid |
Land here had a mortality clock rather than a market. Ground moved when somebody died, and the transfers that happen while a man is alive are nearly all deliberate handovers, a retirement, or sons taking title in the months before a father's death.
Which makes the exceptions worth reading. A transfer that is not near a death is saying something: a plan, or trouble.
Every figure here is counted from the deeds and diagrams on this site and can be recounted. The decade table covers only boundary acts that carry a date; a few undated ones are excluded. It is one family's papers across one district, not a survey of the Karoo. The pattern is real in what is here, and what is here is not everything.
What this story is about
Research notes
[6 Aug 2026] Discovery items LAND-02, LAND-03, LAND-04 and SOC-01 written as one page, because they are one argument: how the ground was cut, between whom, in what order, and when the cutting stopped.
EVERY FIGURE RE-COUNTED BEFORE WRITING, not taken from the discovery report. Confirmed: 96/20/6/1/1 by relation; subdivisions peak at 18 in the 1930s; from 1970 it is 15 consolidations against 3 subdivisions, and from 2003 every dated act is a consolidation; 14 ownership starts in the 1930s against 13 in the 1840s; the first-cut order of the grant farms is Spreeufontein 1850, Middelwater 1883, Rondabel 1910, Varsche Fontein 1913, Vrolikheid 1921; and 7 of 20 datable spans end within a year of the holder's death with 3 on the day itself.
ONE CORRECTION TO THE DISCOVERY REPORT: it said six spans end within a year; the query returns seven. The three same-day cases it named are right.
The 1990 subdivision is not a counter-example to the turning point - it is a piece falling out of the Farm 215 consolidation as that farm was assembled, and the page says so rather than letting the table imply otherwise.
Closing caveat is load-bearing per 10: this is one family's papers over one district and the undated links are excluded, so the pattern is real IN THIS RECORD and the record is not everything.
[12 Aug 2026] VOICE REWRITE to STORY-VOICE.md, both languages. en 1/5/2 -> 0/0/1, af 7/5/2 -> 0/0/1 (narration/bold-sentence/punchline). The one punchline left in each is 'Sorted by date, they tell one story, and it has a turning point', which is a setup line for the section that follows rather than a manufactured close, and it stays.
A WRONG PERSON LINK, AND THE DRIFT CHECK FOUND IT BY ACCIDENT. The page linked Cornelis Floris Johannes MeyerM, the 1841 Spreeufontein grantee who died on 21 July 1848, to PER-000017 - TWICE in the prose and once in the frontmatter. PER-000017 is HENDRIK LODEWYK MALHERBE OOSTHUIZEN, 1909-1979, Frans's grandfather. A reader clicking that name in a sentence about an 1848 death landed on a man born in 1909. The right row is PER-000034, Cornelis Floris Johannes MeyerM, born 15 July 1791 and died 21 July 1848, which matches the story's own date exactly. The likely origin of the slip is OWN-000017, the Spreeufontein ownership row whose to_edtf is that same 21 July 1848 - an own_id transposed into a per_id. Corrected in both languages and in the frontmatter people list. NOTE HOW IT SURFACED: the new drift check flagged STO-0024 because PER-000017 had gained an 11 Aug note about a photograph. That note is irrelevant to this story; opening the record because the check named it is what showed the link goes to the wrong man entirely. The check cannot see a wrong link, only a changed record.
Narration removed: 'This project holds 124 recorded boundary acts', 'his ownership span in this model ends', 'Twenty ownership spans in this model end on a date', 'every dated boundary act in this record', and the closing note's 'counted from parcel_links and ownership', which named two CSV tables in a story about a district. Also removed the method archaeology 'That is a test this site can apply to any span it holds, and it was arrived at by putting two date columns beside each other'; the test itself is kept. KEPT DELIBERATELY: 'It is one family's papers across one district, not a survey of the Karoo. The pattern is real in what is here, and what is here is not everything' - AI-TELLTALES.md names this as the site's own voice about the limits of its evidence.
ONE MANUFACTURED CLOSING LINE CUT: 'The same ground, walked twice in opposite directions.' The paragraph above it already ends on the fact, 'put it back together, or leave', and the aphorism only restated it.
CHECKED AND CORRECT: PER-000083 Pieter Johannes SwanepoelL d. 26 March 1885 and PER-000020 Ger(H)olm Marincowitz d. 13 May 1890 both match the table exactly. verify_af passing.
The research behind this page, kept separate from it: dated working notes, open questions, and what is not yet settled.