19 September 2026
A birth date this site had wrong, and the parents behind it
Cornelia Carter's page on this site said she was born on 25 April 1848 in Mississippi. That came from her own death certificate, filed in Austin in 1937 by her son. It is wrong, and it has been replaced: she was born about 1853 in Arkansas. The reason to trust the second answer over the first is who was doing the telling. A death certificate is filled in by a survivor, at the worst possible moment, about something that happened eighty-nine years earlier. A census taker writes down what the household says while the person is standing there. Three of them did: she is seven years old and born in Arkansas in 1860, eighteen and born in Arkansas in 1870, twenty-seven and born in Arkansas in 1880. Those are three separate men in three separate decades, and they agree with each other and not with the certificate. The date is deliberately just a year. Two records do give April - the 1900 census gives the month, the certificate gives the twenty-fifth - but they give 1856 and 1848 for the year, so a date reading 'April 1853' would be a month borrowed from one record welded to a year neither of them states. April is written into her notes instead, where it can be explained. She also has parents on this site for the first time: Cyrus H. Carter and Martha Carter, who kept house in Pontotoc County, Mississippi in 1850 and in Sebastian County, Arkansas in 1860 before the family reached Travis County, Texas. The two badges on that link do not say the same thing, on purpose. Her mother is marked proven by a record that states it: the 1900 census lists Martha Carter in the household and calls her the head's mother-in-law. Her father is marked proven by argument, which means no document we have says 'Cyrus was her father' - the households do, one census at a time, and the reasoning is written out on her page for anyone who wants to disagree with it.
13 September 2026
The case files, reshaped for whoever carries the line
The case files have come off the top menu and off the front page. They are not gone — every one is still linked from its family room and from the page of each person in it, which is where someone tracing that line will actually arrive. What changed is their shape. Each file now opens with the claim you may have met in a published pedigree or an online tree, tells what happened, lays the records out in the order the argument runs, gives the verdict, and ends with what to change in your own tree if you carry the line. The Susan Jones file was rewritten against the tree as it now stands: her father was born before 1800, not about 1817; her mother was Keturah, not Selena; and both of those come from courthouse books read page by page.
12 September 2026
Eleven people we are not descended from
For years this tree ran the Crouch line back from a Pennsylvania farmer named Benedict Crouch, up through an Elijah Crouch, to Nathan Crouch of Cecil County, Maryland, and from him to the Crouches who sat on grand juries on Maryland's Eastern Shore in the 1690s. Today both of the joints that carried it there were cut, and eleven people left the family. The paper that did it is a distribution. When a man dies without a will his estate is divided by a filed accounting, and that accounting has to name every person entitled to a share — including any child who died before him, whose portion passes to that child's own children, by name. It is a closed list, and it is closed by law rather than by how careful the clerk was feeling. Nathan Crouch's was filed in 1838. It names seven children, divides the residue among them, and the seven shares add up to the whole estate. There is no Elijah on it. He was not Nathan's son. That left a second question, and it went the same way. Benedict had been hung from that Elijah — but the family he was hanging from turned out to be three families merged into one: four separate clusters of birth dates under a single set of parents, and eight of its nine children with no source at all behind them. A family that is holding three families is not evidence for any of them. So that link came out too, and Benedict Crouch now sits at the top of his own name with nothing above him. That is the honest place for this branch to end. He was born about 1800 in or near Greene County, Pennsylvania, and we do not know who his mother and father were. The part worth saying out loud is where the deciding document was. Not in a Maryland courthouse, and it needed no trip and no order: it was inside a collection this project had already searched — one our own search had reported, confidently, did not contain it. The search was broken, not the archive. We were within a day of scheduling a visit to a state archive to go and read a document that was already on this desk. The Somerset County Crouches are still there, and the 1735 estate account that gave Anne Toadvine a father and four named siblings is still a fine piece of paper. It simply belongs to another family. Somebody descends from those people. It is not us.
11 September 2026
His own hand, a second time
We have had John Temple's signature once: on the will he made on 19 May 1812, being sick and weak in body, the day before he died on the Ohio. Tonight there is a second one, seven months earlier and in a steady hand. It is on a bill penal — a promise to pay, with a penalty attached if you do not. On demand I promise to pay or cause to be paid unto Robert Tunstall his heirs or assigns the full and just sum of One hundred dollars, and if he did not, two hundred. Given under my hand and Seal this third day of October 1811. Underneath, Jno: Temple, and beside it the wax. A neighbour named William P. Matthews watched him sign. He never paid it. The paper is in the file because Tunstall sued the estate, and the lawsuit turns out to be four years older than we thought and in a different court altogether. It also tells us something ordinary and human: he was buying on credit in the autumn of 1811, the way a man running a store on a river does. Reading the file corrected us twice more. A name we could only half-read in the county clerk's copy — John Stephen-something, a man administering the estate alongside the widow — is written out in full on three separate originals: John Stephenson. And two figures we had published from a machine's reading of a photograph were wrong. A commission was charged on $2,180.81, not on $180.51. A receipt for fifty dollars was signed not by someone called Latance but by a Latane — one of Mary's own family, three hundred miles from home, settling her husband's accounts. Nothing we had published was stated as fact; both figures were flagged as unverified until somebody opened the images. Somebody has now opened the images, so they are corrected here, and the old readings are kept beside the new ones with a note saying which was wrong. A correction that hides what it corrects is not much of a correction.
11 September 2026
Her name was Letty
Later the same day we found the other copy. The inventory we had read was the clerk's — he copied it into the county will book in August 1812. The paper the appraisers themselves wrote, folded and handed into court that summer, has been sitting in the Wood County loose court papers, filmed sixty-six years ago and free to anyone with a FamilySearch account. It is still folded in the photograph, still endorsed on the outside in the clerk's hand: Appraisment, Est. Capt John Temple, Decd, Augt 1812. Reading it changed two things. The first is small: there is no mysterious abbreviation in the heading after all, only Est. for Estate, and no hidden rank in it. The second is not small. On the clerk's copy the first of the four enslaved people is written in a way that could be read two ways, and we recorded it both ways and refused to choose, because guessing at a name is a kind of erasure. On the original the hand is clearer. The letter is the same one that begins Large in Half of Large Keel boat, eight lines below, and it is not the letter that begins Smoothing in 2 Smoothing Irons. Her name was Letty. She was appraised at three hundred dollars together with her child Nancy, on a line between the candle moulds and the stone churn, by men whose own names the document does not record. We still do not know what happened to her, or to Nancy, or to the second Nancy and her child Tom on the line below. The bill of sale that would say is the next thing to look for.
11 September 2026
What John Temple was doing on the Ohio River
He died on 20 May 1812 in Wood County, on the Ohio, three hundred miles of mountains from his wife and four children in Essex County. He was forty-one. His youngest son had been born six weeks earlier and never met him. We have had his will for a while now, signed the day before he died, sick and weak in body, asking that his property be held by two local friends until a brother or brother-in-law from Essex could come and take charge of it. What the will never said was what he was doing out there in the first place. The inventory his appraisers took seven weeks later answers it. They went through his effects on 6 July 1812 and wrote at the head of the page: the estate of Captain John Temple. What they found was not a household. It was a store. Nine hundred and forty-four pounds of coffee, and six hundred and sixty-eight pounds more of it green. Pepper, allspice, ginger, nutmegs, indigo by the ounce, tea at a dollar fifty a pound. Salt by the bushel and again in barrels. Nails by the pound. Eight hoes, a chest of tools, empty barrels, jars and jugs at forty cents each. Against the bulk goods, over and over down the margin, the appraisers wrote the same phrase: quantity not ascertained. They could not weigh it all. Two entries settle it past argument. Three funnels, weights and scales; and a set of measures. Nobody keeps scales and a set of measures to feed a family. He was selling. He was carrying, too: a half share in a large keel boat, and a skiff. Parkersburg sits where the Little Kanawha comes into the Ohio, and in 1812 the Ohio was the road west. He took deer skins in trade. The captain's title the clerk gave him turns out to have a coat to go with it, the most valuable piece of clothing he owned. And he held no land there at all. He had gone out to trade, not to stay. The same pages carry something harder, and it belongs here rather than in a footnote. A settlement of this estate credited eight hundred and forty-seven dollars from the sale of enslaved people and named none of them. The inventory names them, two pages earlier, written in between the candle moulds and a stone churn: a woman and her child Nancy, and a woman named Nancy and her child Tom. Four people, appraised at three hundred dollars a pair. The first woman's name begins with a letter this clerk writes ambiguously, so it is recorded unresolved rather than guessed at, because a guess would invent a person who never existed. Where they went afterwards is written down nowhere we have found. The document that would say is the bill of sale, and it is not in the will book.
10 September 2026
The 1850 census settles a wife, and unsettles two dates
The 1850 census settled a question this tree had answered wrongly for years. Washington Rose's wife was Sarah, alive that October at 50, so the 1843 death is gone. The 11 July 1816 Putnam County marriage the tree carried belongs to a different bride, Elizabeth McGehee, so the couple's marriage now reads simply before 1830. A son, William, born about 1841, joins the family from that same census page.
6 September 2026
Two mothers, proved by a column on somebody else's line
Both of these questions had sat open for the same reason, and it was not a shortage of records. A census asks how each person in a house is related to the head of it. So when a child is written down as the head's daughter, that names her father and says nothing whatever about the woman standing next to him. The mother was left resting on who happened to be in the room, which is not proof of anything. The answer was already in the same two censuses, a line or two up. The 1900 census asks two questions of the parents that it never asks of a child: how long the marriage has lasted, and — of married women only — how many children she has borne and how many are still living. Lila Plunkett's line says twenty years married, four children borne, four living — and there are exactly four children in that house, Alice among them. Four borne, four living, four present: the household is the whole of her family, so Alice is hers. Her parents were named together again in a Little Rock paper the morning after her 1914 wedding. William Benedict Crouch's case turns on the other column. His parents' line says forty-seven years married, which puts the wedding in 1853, three years before he was born, and the marriage ran on unbroken from there. There was no other wife for him to have belonged to. Neither answer is written on the child's own line in the census. Both are on a parent's, a few rows away in the same house — and that is the entire reason these two sat unanswered as long as they did. There are more of them to go back for. Worth being plain about what this is: no record here says outright, in words, who either mother was. The case is built from two records that only make sense together. That is why the mark on these two now reads proven by argument rather than proven on a record, and why the note underneath each one ends by naming what would undo it — for both, a marriage record showing the father married earlier than we think.
5 September 2026
The check mark learns the vocabulary
A fortnight ago the mark beside a parent said one of two things: proven, or nothing at all. The research had a six-way vocabulary for how well a link is proved; the mark did not, so it flattened all of it. That gap was not harmless. Thirteen parents were carrying a check mark reading proven while the research note lower down the same page said the question had been looked at and what is in hand settles nothing — the page contradicted itself, and nothing here could see it, because the contradiction never moved the underlying flag. Three more had been examined and deliberately declined, and showed no mark at all, which made a decision look identical to an unopened question. The mark now carries the distinction: proven on a record, proven by argument, proven before this vocabulary existed and not yet re-read under it, asserted and under review, or declined. Each one links to a new page that says what it means, and a check that runs before anything is published now refuses to ship a page whose mark and whose note disagree.
5 September 2026
The wing catches up
This wing is rebuilt from the master tree after every research pass — and for twelve days it was not. The tree moved and the published copy did not, so anyone reading here in late August was seeing August's conclusions while the work had gone on without them. The gap is closed: every page is now built from the 5 September tree, and twenty-eight parentage conclusions that had already been earned are finally visible. The interesting part is why nothing caught it. The checks that run before publishing compared the number of proven links on each side, and that number had not changed — the research had been rewriting the reasoning behind links already marked proven, which is invisible to a count. A check that compares the notes themselves now runs before anything ships.
5 September 2026
The research notes now say how good they are
The notes on a person's page are kept here verbatim, and many of them now open with a line naming what kind of proof stands behind that parent. Direct means a record read for this project states the relationship outright. Argument means no single record states it, but two or more independent records, set against each other in writing with the conflicts dealt with, leave no room for another answer. Informant means one uncorroborated statement by somebody who need not have been there. Underdetermined means the question has been looked at, what is in hand reaches neither bar, and the note names the cheapest document that would settle it. The review is working outward from the closest generations and is nowhere near finished — so a page with no such line has not been looked at yet, which is a different thing from having been looked at and found wanting. The check mark on the parents line is older than this vocabulary and still says only proven or not; read the note if the distinction matters to you.
4 September 2026
One line in a 1933 typescript
A marriage bond abstracted from the Clark County, Kentucky, court records and typed up in 1933 reads: Quisenberry, Stephen C., and Cloe J. Bush, daughter Ambrose Bush (consent); surety, Colby Bush. Dated 7 October 1844. Read on the filmed page rather than trusted from an index, it does two things at once — it puts the marriage three years before the birth of the daughter this tree descends from, and, because a bride under twenty-one needed her father's consent, it names her father. What took the rest of the day was establishing which Ambrose Bush: the same county holds two other men of that name, and both are ruled out by their own dates in the same volume. The date corrected an older note here that had it as 10 October, taken years ago from a memorial site this project no longer cites.
3 September 2026
A Virginia will, and the son who is not in it
The will of Henry Foulkes senior of Nottoway County, Virginia — written in April 1808 and proved that September — was read end to end on the manuscript, all four pages. It names eleven children. Ten of them were already in this tree. The eleventh, a daughter called Nancy, was not, and now has a page of her own. And Edward Bass Fowlkes, whom this tree had carried as a son of that family, is not named in it — in any clause, in any spelling. He has not been quietly removed: no other father has been found for him, and the honest state of things is that he is still shown where he has always been shown, with no proof mark on his parents and a note on his page explaining what the will says. This is the ordinary shape of the work. A document you go looking for to confirm something can just as easily take it away, and the tree is more useful when it admits that than when it tidies it.
27 August 2026
Seven children, named in one clause
William S. Derrick died in Washington in 1852. His will sat in a box in the District of Columbia's archives, findable through a published index and orderable on a form; it was requested in late August and arrived three days later. In a single clause it names his wife and then all seven of his children — Lucretia Hart, Anna Louisa, Henry Clay, Josephine, Estelle, Clarence and Georgine Montgomery — in the order this tree already had them. One document, thirteen parentage marks: a mother proved for the daughter this family descends from, and both parents proved for her six brothers and sisters. Nothing about the shape of the tree changed. What changed is that it is now the record saying so, instead of a family historian writing in 1887.
22 August 2026
The other side of the family
This wing carries two ancestries — one for each side of the family — kept apart and counted separately, because they share no one and adding them together would mean nothing. Until late August, all of the proving had happened on one of them: on 21 August the second side carried no proven parentage at all. It now carries twenty-four. The evidence there is of a completely different kind. Most of these people died in Texas between the 1950s and the 2000s, inside the years the state issued death certificates, and a Texas certificate has a line for the father and a line for the mother's maiden name. Where the first side's hardest generations sit in Virginia and on the frontier before 1850, and a record naming somebody's parents very often simply never existed, this side had its answers waiting in a filing cabinet. The contrast is a fact about the records, not about the families. One link here had been written off in this project's own notes as blocked until 2031, waiting on a restricted record; it turned out to need a census page instead. The 1910 census records the father as single, the children begin arriving after that, and the 1920 census names the mother in the household with all four daughters — which leaves no room for anyone else.
21 August 2026
A home for the download
The downloadable tree now has its own page — what the GEDCOM file holds, what is deliberately kept out of it, and the honest caveat about importing conclusions — linked from the landing page and every cousin door. The download itself is unchanged: the same dated snapshot, rebuilt from the master tree after every research pass.
21 August 2026
The Temples, told whole — and four names corrected
A hundred images of the family's own papers arrived from the Wiltshire archive: four wills of Peter Temple in his own hand, the deeds of his son Samuel's 1722 marriage, and the receipt box his widowed daughter-in-law kept for twenty years. They resolved a puzzle the tree had carried for weeks, corrected four names a London clerk miscopied in 1738, and are now told as one seven-generation story — from a chancel burial on the River Wylye to a fatherless boy on the Rappahannock — linked from the Temple room, with every uncertain joint marked in the text.
19 August 2026
The photo albums arrive
Eight shared Google Photos albums are now linked from the family rooms, the cousin doors, and three person pages — scanned photographs for both sides of the family, an album each for Thomas Lewis Latané Temple Jr and III, and one for Ada Mae Johnson Catchings. The photos stay on Google; the site carries the pointers.
19 August 2026
Six doors for cousins
A doorway for every cousin. Share my grandparents and half this wing is your direct ancestry; share great-grandparents and a quarter is. Each door lights up your shared wedges of the fan, counts the ancestors we have in common, and hands you your rooms — start at Which door is yours?, linked from the landing page.
19 August 2026
The August drafts ship
Three stories finished in early August are finally live: the two fathers the Revolution took from the Carpenter line (Valley Forge and Blue Licks), Finding Ann Waring, and the honest story of the hanged rebel Anthony Arnold and the English father our tree had to un-invent. Every subject's person page now links their story.
19 August 2026
Three new stories from the beat
The tree-journalist agent's first leads, written and published: Anastasia Catching's 1807 will, signed with her mark; Abner Fowlkes, home from the war with one year to live; and an honest answer to the Mayflower question, graded link by link. All three are linked from their family rooms.
18 August 2026
The rooms, the files, and the machine
Eight family rooms — the story, the people, the records shelf and the honest grades for each of the eight directions. Five case files from the research's hardest junctions. And the moment machine: pick any date and meet the living cohort.
18 August 2026
Every name becomes a page
2,333 person pages, one for every named deceased person in the published tree — vitals, parents with proof badges, record transcripts, research notes kept verbatim, and family links to walk the tree by. Plus the places index, and click-through from the fan and the atlas map.
18 August 2026
The wing opens
The landing, the eight directions, the atlas and the deep-time poster, the people index, and the tree itself as a free download.
4 August 2026
Seven stories on the blog
The whole-tree overview and six family posts published in the blog's History category — the reading room on the landing page links them all.