Everything found or corrected in this archive, newest first. Where a record turned out to disagree with what we thought, that is written down too.
Discoveries
Fischbach, and an award nobody knew about
Six hundred frames of declassified Army general orders were downloaded and read, and two of them change what we know about Andrew's grandfather's battalion. On 11 February 1945 a wire crew of Company A, 34th Signal Construction Battalion, was putting in communications installations near Fischbach in Luxembourg when it came under heavy shellfire and a man was hit by shrapnel. Two of them, Staff Sergeant Frank Yelenchic and Technician Fourth Grade Chris Stako, left the ditch they had taken cover in and carried him to safety, and Third Army decorated them both. John Fitzgerald was the battalion's medical NCO; a man wounded by shrapnel in his own unit is exactly the work he was there for. The second document corrects something this archive said only hours earlier. It had judged a reported unit commendation to be a confusion with a similarly named company that served in Italy. It was not: General Orders 139 of Patton's Third Army, dated 14 June 1945, awards the Meritorious Service Unit Plaque to the 34th Signal Construction Battalion for service from 20 October to 26 December 1944 — a period that ends ten days into the Battle of the Bulge. It is an award he earned and that does not appear on his discharge.
Pressing on Andrew's grandfather's battalion turned up its own paperwork, declassified in June 2023 and digitised free at the National Archives in St Louis. The General Orders of the 34th Signal Construction Battalion are dull administrative sheets and they are the first primary documents this archive has about his unit. They give it a post number, APO 403, and a commander — Major, later Lieutenant Colonel, Walter H. Wood. They show it had at least a Company D. And they show it was shot at: a man wounded in France on 1 September 1944, and on 14 March 1945 two more, Sergeant Robert Schroeder of Spokane and Technician Fifth Grade Carlos Thompson of Harrisonburg, Louisiana, both hit by enemy action at Rosport in Luxembourg, a village on the Sauer at the German border. He is not named in these particular orders, which award Purple Hearts, and he was not wounded. But this is his battalion in its own words, and the fiche runs to nearly two hundred frames of which only a sample has been read.
His grandfather's war, from a folder in the family
Andrew's grandfather John Joseph Fitzgerald had an empty page in this archive — a birth, a death, a marriage and nothing between them. His discharge papers have been photographed and they fill it. He was a stock clerk of 542 Bergen Avenue, Jersey City, twenty-nine years old, five foot seven, when he was inducted in March 1943. He crossed to Europe on 11 February 1944 as a medical NCO with the 34th Signal Construction Battalion, and his discharge lists five campaigns: Normandy, Northern France, the Rhineland, the Ardennes, Central Europe. He was not wounded. Then, Germany finished, he was sent the other way — he sailed for the Pacific on 17 July 1945 and landed on 23 August, nine days after Japan surrendered, which is why a European record also carries the Asiatic Pacific medal. He came home in December and was discharged a corporal at Fort Dix on seventy-four points. The folder also holds Truman's letter of thanks. And the discharge certificate carries a Hudson County stamp of 16 March 1981 — he had died on the 14th, and someone took his papers to the Register's office within two days.
Gina's line has moved back a generation, from a free record that took ten minutes once the right door was found. Michael Mayer, her three-times-great-grandfather, was carried here as two unverified dates off a family tree and a marriage in Essex County. His New York City death certificate turned up whole: he was a butcher, he died at Fordham Hospital on 30 July 1902 of chronic nephritis, aged forty-seven, he had been living in a tenement at 2135 Prospect Avenue, and he was buried at Woodlawn two days later. And in two narrow columns near the middle of the form, it names his father and mother — Daniel, and Emma, both born in Germany. It also quietly contradicts the family tree, which said he was born in New York; the certificate says Germany, and that he had been twenty-seven years in the country. Both readings are kept, because the twenty-seven years sits awkwardly against a marriage in 1875.
Pushed hard on whether the Salvatore Spatarella of the Suffolk newspapers belongs to Gina's family, after the reason first given for doubting it turned out to be wrong. The push paid twice over: he and Marie are now proved husband and wife three separate ways, her family name was Ardi, and a man who had been nothing but a name on a shared grave has an occupation, a town and a life. Then the record that actually answers the question turned out to have been in this archive all along. The 1950 census of the Brooklyn household, read from the sheet rather than the index, holds five people — James, Amelia, Angelina at nineteen, Esther at sixteen and Bobby at seven — with the continuation box unticked and the closing rule drawn. A son born in 1939 would have been ten. He is not there, and on the newspapers' own reckoning of his age he would not be there either. He is not a son of this house. Whether he is a cousin is a different question and nobody has evidence for that either.
Two people have sat in this archive for months as nothing but a shared grave at Pinelawn — Salvatore A. Spatarella, dead at thirty-seven in 1977, and Marie Spatarella, who outlived him by forty-one years. Their pages said carefully that a shared grave is strong evidence of a marriage and no evidence at all of anything else. Long Island newspapers now settle it three ways: a Suffolk tax roll listing a lot held by 'Spatarella Salvatore & Marie'; her mother's obituary naming her among the surviving daughters; and a 1965 boating item that calls a Gerald Noest his brother-in-law, which he could only be through her. Her family name was Ardi. He was a refuse hauler of Commack and president of the county cartmen's association, and after he died the firm passed to her. Whether he belongs to this family is still open — but the reason previously given for saying he did not, that he lived in Commack while these Spatarellas were Brooklyn people, turns out to be wrong: Gina's father grew up in Commack.
Gina's grandfather has been recorded here as deceased on the family's word alone, with no date — and that missing date was the whole reason the line could go no further, because a Howard Jones cannot be searched for without one. His father's death notice, printed in a Glen Cove paper in January 1996 and read this week, calls him 'the late Howard G. Jr.' It does not give a date. It does something almost as useful: it documents the death instead of assuming it, and puts it before 5 January 1996, when he would have been about sixty-one. The same nine lines give his father an exact death date, a burial in the churchyard at All Saints, Great Neck — the first burial place recorded anywhere on that branch — and a second marriage nobody in the family had mentioned.
The record was here and unfindable at the same time
A search for newspaper stories about the family turned up very little, and then turned up why. Catherine Meaney's marriage notice from the Jersey City News of June 1889 has been in this archive for months — somebody once read it off the page image by eye. Searching that very page for the word Meaney returns nothing at all. The scanner read her name as Meauev, married as niarried, and Father Egan as F.gan. The notice was present and unfindable at once, which is almost certainly why an earlier search for every spelling of Leahey came back empty against a true count of 135. The search tool now looks for the manglings as well as the name. It also turned out that the New Jersey filter had never worked: the site accepted it, returned exactly the same number of hits, and handed back newspapers from Washington and South Dakota. Every geographical newspaper search this project has run was in fact a national one.
Andrew's direct line has moved for the first time this week, twice, and both women came out of a Bayonne parish register that had been sitting undownloaded. His great-grandfather James D. Leahey had a father recorded and no mother; his great-grandmother Barbara Brennan had the same. Both were christened at St. Mary Star of the Sea in 1886, and both entries name the mother. James D. was baptised Denis, on 3 April 1886, son of John Leahey and Margaret Mahoney — and the sponsors were Denis and Anna Leahey, his own grandfather and aunt, which is what identifies the family. He was very likely named for the grandfather standing over him, which would explain the D he used all his life. Barbara's mother was Catherine Lavelle, named at both her baptism and her sister Catherine's two years later, where the sponsor was a Peter Lavelle — the only trace of her own family anywhere in this archive. And a third person appears: Catherine Brennan, born 6 October 1888 and baptised the next day, whose christening this archive has quoted for months without ever giving her a page of her own.
Andrew's great-grandmother Mary Cooney is one of sixteen people where his direct line simply stops — no parents, not even their names. Her death had been carried on the family tree's word alone: 29 August 1949 at the Jersey City Medical Center, never checked against anything. New Jersey's death index confirms the month and the state, and gives a certificate number, 29315. A 1949 death certificate names the dead woman's father and mother, so that number is a generation waiting to be bought. Finding her needed one correction of habit: a death index files a married woman under the name she died with, not the one she was born with. The first search ran under Cooney and turned up a different Mary Cooney who died that January — right surname, right year, wrong woman, and it looked for a moment as though the family's date was wrong.
This archive held three women recorded only as 'Mrs. Louis Smith', 'Mrs. Harry Smith' and 'Mrs. Paul Aldbeck' — their husbands' names, from the family's death notices, with no given names and no dates at all. Two of them have their own names back. St. Mary Star of the Sea in Bayonne annotated its baptism register with the marriages that followed, decades later and in the same church. So the entry for a baby christened in 1887 carries, in its margin, the wedding of 1911 that would one day reduce her to her husband's name. Mrs. Louis Smith is Caroline Meaney, born 17 October 1887. Mrs. Harry Smith is Joanna Meaney, born 28 December 1889 — and Harry was Henry. Their mother, known here from a headstone as Catherine Roland, appears in the register as 'Cath. Rohling' and 'Cath. Rohlan', the clerk's two tries at the same name. The third sister is not in these books, and probably never could have been: the register starts about 1884 and she was likely born before it.
Catherine Condon and David Fitzgerald married in New Jersey in 1908, volume 9 page 563 — fourteen people in this archive hang below that couple, and their marriage had been carried for months marked 'identity unproved'. The entry was nearly missed in a way worth recording. Searching the index's transcribed text for 'Catherine' returns nothing at all: the row is plainly legible on the scan, and the words CATHERINE and FITZGERALD appear nowhere in that page's text layer. Three printed lines were simply never transcribed. Nothing about the volume looks wrong from the inside — sixty-five thousand rows read, twenty-one other Condons found, alphabetical order intact on either side of the gap. It was caught only because the search tool refuses to accept a zero when the surname is present and tells you to go and look at the page. The certificate itself would name both sets of parents, which is what this line actually needs.
Barbara Brennan's 1916 marriage was found last week by downloading a 20MB file and reading 401 pages of it by hand. It turns out there are 503 such volumes — New Jersey's birth, marriage and death indexes from 1901 onward, free on the Internet Archive, and now searchable from a single command. On its first afternoon it gave Marian Murphy an exact death date where the archive had none at all (22 May 1977), and independently confirmed two dates that had rested only on Find a Grave. It also produced a puzzle: James Francis Leahey, buried in his mother's grave, is not in New Jersey's 1983 death index at all, though his brother George is on the same page six lines above where he would fall. Either he died out of state and was brought home, or the memorial's date is wrong. The best part of the tool is the part that refuses to answer: it knows which years were never released, and when a search for a full name comes back empty it re-checks the surname alone and blocks the result if the surname is there, because on these scans a blank often means the OCR could not read the column.
One search of Holy Cross by cemetery rather than by name returned eleven Leahey burials, and the plot numbers did the rest. Barbara Brennan lies in Blk-64 Sec-A Gr-718 with two of her sons: John Joseph in 2A and James Francis in 2B — a son this archive did not hold at all. Her son George lies in a second family grave with his wife, a daughter and an infant who died in 1933. The best of it is the wife. This archive knew her only as 'a Vera', from the family's death notices: no dates, no sources, suppressed as possibly living. The cemetery calls her Julia Veronica Leahey, born at Manhattan on 1 September 1914 and dead at Bayonne in January 1965, and names the same daughter the notices did. Vera is the short form of Veronica. No string comparison would ever have joined those two records; it took reading a list of graves. Her granddaughter Barbara, born 1935, was named for the Barbara in the other grave, who was then alive a few streets away.
Andrew's great-grandmother Barbara Brennan was widowed in 1914 and remarried a Murphy, and for months that man was simply 'a Murphy' — which is also why no deed search was ever going to find the Belmar house where Andrew's father spent his summers. He was William Murphy, and they married in New Jersey in 1916. The record was free the whole time. The search had been running under Leahey, the name she was using; the New Jersey bride index files a woman under the name she was born with, so she is indexed as Brennan. She is the only Barbara Brennan in the entire five-year index. The marriage file number is 25547, and that certificate would give William Murphy a father and mother, an age, a birthplace and an address — he has none of those yet.
Reading four Cavan register images to settle whether Patrick Farrelly could write turned up something that touches the whole archive. Every Irish civil register image here is a certified copy the registrar made from his own book — the clause is printed at the foot of each page, and on one Mayo page of 1894 ten different women make ten identical crosses in one uniform hand. The mark annotations are copied faithfully, so who could write and who could not is unaffected. But the writing itself belongs to the registrar, and this archive cannot show what any of these people's handwriting looked like. Nine cropped images and their section heading have been recorded honestly, and a paragraph on Margaret Needham's page that claimed she wrote her own name — contradicting the rest of her own page, which has said for months that she made a mark — has been corrected. Patrick Farrelly himself is now neither on the marked list nor the signed one: he made a cross in 1880 and none is recorded in 1879, 1881 or 1884.
The question was whether an Andrew Leahy remembered from Pine Beach could be a relation. His family is now identified precisely, and the honest answer is that they are not yet connected, and that the most attractive theory about how they connect is dead. Andrew's grandfather was William A. Leahy, who died in December 2008 aged seventy one. He was born in Jersey City about 1937, raised in Bayonne, and moved to Pine Beach around 1977. He spent thirty one years in the New Jersey State Police and retired as a lieutenant of the Criminal Investigation Unit, having also worked as an investigator for the state Board of Education and the United States Marshal's Office. He served in the Marine Corps in Korea and belonged to the Celtic Club of Bayonne. His wife Gail, who died at Toms River in 2019, was born at Dover in 1939 and lived in Bayonne and Pine Beach before Toms River. Andrew is their grandson, not their son. The obvious candidate for a link was John Joseph Leahy Senior, the one man in this archive who takes a branch out of Bayonne and down to Ocean County, dying at Island Heights. If William had been his son everything would have fitted. He was not. John's sons were Jack, Robert and Walter, all three named on Find a Grave and two of them cut into his headstone. William's brothers were Daniel and Michael. There is no overlap, and that theory can be set aside. What survives is a circumstantial fit that is hard to ignore. Born in Jersey City, raised in Bayonne, and then moved to the exact corner of the Jersey Shore that this family, and no other Leahy line here, had already moved to. A Bayonne Irish Catholic Leahy in law enforcement, in an archive that already contained a John Leahy known for years only as the Boulevard police one. But Leahy, Leahey and Lahey were common names in Bayonne, and this project has already had to undo one identification built on nothing more than a surname and a place. So it stays a lead. The record that would settle it is the 1940 census, where William would have been about three years old in a Jersey City or Bayonne house with his parents and his brothers Daniel and Michael. That sheet names his father, and his father's name is the entire question. If it turns out to be a son or nephew of Malachy Leahy, then the two Andrews are third cousins.
Elizabeth Leahy, born at Bayonne in 1906, was one of Malachy Leahy's daughters. She married Charles Ryan, and she died in 1984 at Bridgeton, right down at the far end of New Jersey. Her monument at Holy Name Cemetery in Jersey City has now been read, and it is a considerable thing: an upright granite stone with RYAN cut across it, a carved panel of two haloed figures standing either side of a child beneath a descending dove, an IHS cross with a rosary hanging down each side pillar, and a band of flowers across the top. Three people are named on it. Charles M, 1898 to 1972. Elizabeth M, 1906 to 1984. And Michael C, 1928 to 1990, who was their son and who was not in this archive at all until the stone was read. That is worth pausing on. A photograph of a headstone, taken by a volunteer and put on the internet, has just added a person to this family who had otherwise vanished entirely. Nothing else we hold mentions him. It also draws a quiet contrast. Elizabeth's brother John took his family down to the Jersey Shore and lies under a flat slab at Toms River. Elizabeth stayed with the old ground, block 11 of Holy Name in Jersey City, the same cemetery her grandparents Dennis Leahey and Ann Galvin were buried in eighty years before her. Both stones are now on the site with the photographers credited.
A stone at Toms River, and a man who went home to be buried
Two Find a Grave memorials have filled in the Ocean County branch of this family, the one that left the Bayonne waterfront for the Jersey Shore. The stone at Saint Joseph's in Toms River is a single flat slab with one word across the top, LEAHY, and three people under it. John J Sr, 1905 to 1988. Estelle M, 1906 to 1978. Walter T, 1937 to 1989, who died a few weeks after the year turned on his father's death. An American flag stands at the head of it. The photograph is now on his page, credited to the volunteer who took it in 2019. The eldest son is the more interesting story. John Joseph Leahy Jr, known as Jack, was born at Bayonne on 20 July 1933 and died at Island Heights on 15 March 1999. He was a veteran, of an age for Korea, and he married Josephine DeCrescenzi in 1956. But he is not on the stone at Toms River with his father and his brother. He was taken back to Holy Name Cemetery in Jersey City and buried in block 6, section B, grave 103, in the same ground the family has used since Dennis Leahey and Ann Galvin were laid there in the 1900s. Two generations at the shore and he still went home to be buried. His children are not listed anywhere, which is precisely what you would expect, because they are alive. Jack married in 1956, so his sons would have been born through the late fifties and sixties and grown up in Ocean County, a few minutes from Pine Beach. That is as close as records can get to the question that started this: no other branch of this family went to the shore at all.
A Leahy branch at the Jersey Shore, and how the family reads itself now
Somebody asked whether an Andrew Leahy they knew growing up in Pine Beach, New Jersey could be related. The archive turns out to have a clear answer to the shape of that question, even if it cannot yet name the man. John Joseph Leahy Sr was born in Bayonne in January 1905, the son of Malachy Leahy, and he died at Island Heights in Ocean County in December 1988. Island Heights, Toms River and Pine Beach are neighbours. He married Estelle Birch in 1932 and they had three sons, Jack, Robert and Walter, all of them Ocean County men, and he lies with his wife and two of those sons in one grave at Saint Joseph's in Toms River, section C, row 25, grave 39. That is the moment this family leaves the Hudson County waterfront for the shore, and it is the only branch in the whole archive that went there. Malachy was a brother of John Leahey born 1861, and it is from that John that the direct line of this archive descends. So the two branches part in the 1870s in Bayonne and meet again, if they meet, as third cousins. Any Leahy who grew up around Pine Beach in the 1960s or 70s is far more likely to belong to this branch than to be a coincidence of surname. A search of Find a Grave for an Andrew Leahy anywhere in New Jersey returns nobody at all, which is exactly what you would expect for a man still living. Two housekeeping matters. That page used to live at an address ending boulevard-police, which was a note somebody made while telling three John Leahys apart and was never a name. It now reads john-joseph-leahy-1905, and the old address still works and always will. And a third of the people here, a hundred and eighteen of them, opened with the sentence a member of the Leahey family recorded in this archive, which tells a reader nothing. Most of those records were not empty at all: they held parents, a marriage, a birthplace, a trade, and the sentence had simply never been written. The site now writes it from the records it holds, says so plainly at the end, and leaves the properly written biographies alone.
Two birth years corrected, a mother's mark, and a man who waited until fifty-three for the farm
A pass over the thinnest records in the archive, starting with the ones that the most relatives depend on. Two of the Ventry Fitzgeralds in County Kerry had birth years that came from a family tree and had never been checked against anything. Both were wrong. John was recorded here as born in 1901. The civil register has him born on 5 October 1900 at Cahirtrant, and registered on 6 December, which is probably where the extra year came from. His sister Mary was recorded as 1903. She was born on 7 December 1902 and registered the following January. Both are now sourced to the register page itself, and both pages name their father John Fitzgerald, a farmer of Cahirtrant, and their mother Bridget Fitzgerald, whose maiden name was also Fitzgerald. Bridget registered both births herself, and both times she made her mark instead of signing. Her husband had done the same registering their son Patrick in 1895. Neither parent in that house could write their name. Getting there required a discipline worth writing down. The index will tell you the mother's maiden surname for births after 1900, which sounds decisive, and is not. There were two Fitzgerald births in Dingle in those years with a Fitzgerald mother, and only one of them was ours. The other belonged to an entirely different family. Nothing but opening the register image and reading the townland would have separated them, and attaching the wrong child would have been very easy. The censuses then closed a question rather than opening one. The 1901 and 1911 returns for Caheratrant both list the same three children and no others, so there is no missing Ventry sibling to go looking for. They also caught something you would never learn from a date. In 1901 John was forty years old, married, with two children, and the census still records his occupation as farmer's son, because his father Maurice held the land. Only in 1911, at fifty-three, is John entered as farmer and Maurice, then seventy-eight, as retired farmer. He waited most of his life for it.
One map, every place, and five places that were never on it
There is now a single map of the whole archive, and clicking any pin tells you who was there. It lives on the Places page and it builds itself: every place in the archive that has coordinates and at least one person attached gets a pin, and the pin's popup lists those people by name with links to their pages. Ninety one pins at the moment, spread across the eastern United States, Ireland, England, Italy, France and Australia, with cemeteries marked in red. Add a person to a place tomorrow and the pin updates itself with no work at all. Two things were quietly wrong and are now fixed. The map used to vanish entirely on a phone, which meant the one page in the archive that is genuinely about geography showed nothing at all to anyone reading on a handset. And the pins used to say only the name of the place, which is the one thing a reader can already see from the list above it. Both are corrected. The more embarrassing problem was invisible. This archive keeps its place table in two files, and the site generator only reads one of them. Five places added earlier today, including Naples where Carmine Cacace was born, Calvary Cemetery, Old St Raymond's where the Farley baby is buried, Great Neck and Sea Cliff, went into the file the generator does not read. They were correct, they were saved, and they simply never appeared. Pinelawn had the same problem in a different form: it had a pin's worth of coordinates in one file and none in the other, so six burials sat unmapped. Everything is now synchronised, and the checking tool has grown a section that compares the two files and complains, so the next time somebody edits the wrong one they will be told within a second instead of never.
Until today this archive had five hundred and thirty seven people and exactly one photograph of a face, tucked into a sidebar. The photographs are now a proper part of a person's page, and the first three are up. The one the family has always called Dennis is here: an elderly couple standing outdoors with two small children holding on to them, the man in a dark suit with a white moustache, the woman in a pale dress. The dress and the print put it in the 1930s, which fits Dennis C. Leahey, born 1867, and rules out his father the Captain, who died in 1907. The identification comes from the family and not from anything written on the print, and the page says so. There is a way to settle it that nobody has done yet: the newspaper printed a portrait captioned DENNIS C. LEAHEY when he died, and the two faces can simply be compared. The second is a snapshot on a wooden porch of a young sailor in United States Navy dress uniform standing beside an old man in shirtsleeves and braces. Neither has been identified. The uniform was worn in that pattern from the 1910s to the 1940s so it does not even date the picture. It is published as unidentified because someone may recognise the porch. The third is the best object in the collection. It is a metal badge reading S. O. CO. OF N. J., FIRE DEPT., 44, BAYONNE WORKS, and above it the round photographic identity badge that went with it, with the holder's face and the same number. That is the works fire brigade of the Standard Oil refinery at Bayonne, where this family worked for three generations: George Hoffman was a stillman there in 1900, his sons were labourers there, and Patrick Gorman retired from the same company in 1938. The badge is ours. Which man wore it is not recorded, and the young face in the photograph is nobody we can name. Every photograph now carries a line above its caption saying plainly whether it is identified by a record, identified only by the family, or not identified at all. A face nobody has proved is a face nobody has proved, and that should be visible without reading the small print. Person pages have also started cross referencing themselves. Where two people are recorded at the same specific address, each page now names the other and links to them. It is only six pages so far, but it grows on its own as addresses are added, and it is how a household or a lodger becomes visible instead of sitting in two records that never meet.
A tool that reads the archive against itself, and what it caught in the first minute
The archive has grown to the point where the expensive part is no longer finding a record. It is noticing that two records already sitting here bear on each other. So there is now a tool that reads the whole corpus and reports where that is true, using no internet at all, in about a second. Running it for the first time it found four things worth having. Vincent Leahey was recorded as a son of Dennis Leahey and Ann Galvin, and had been carried as a candidate for one of their six unidentified children. He cannot be. He was born around March 1912 and Dennis died in April 1907, five years earlier, when Ann was already in her sixties at least. Being buried in a family plot shows family, not parentage. He was almost certainly a grandchild, one of the sons' children, and the link has been removed. He died on 8 October 1918, which is the exact week the influenza pandemic's second wave was tearing through the New Jersey ports, and a six year old dying in Bayonne that week is very likely a flu death, though the stone gives only a date. The tool also noticed that three people share Holy Name plot F–EA–112 without any of their pages mentioning the others. It noticed that Mary Cooney of Derrycriff shares an address, 264 West 155th Street, with Andrew Farley and his daughter, which is the bridge between the girl born in Mayo and the woman who married him. And, least comfortably, it caught a contradiction introduced here earlier the same day: one page called Anne Cooney the Annie Farley of her brother's army file while another page had her married to Michael Cawley. Both cannot be true. The Cawley records win, and Anne's page has been corrected. That last one is the argument for the tool. A person working through five hundred records over months will contradict themselves, and the contradiction will sit there looking like a fact.
The Cawleys found in Manhattan, and a page layout bug that hid a link
Anne Cooney of Derrycriff had a family in this archive that consisted of four names and nothing else: a husband, Michael Cawley, and sons Andrew, Marty and Peter, with no dates, no places and no records behind any of them. An earlier sweep had looked for them and found nothing at all. The reason is now obvious. That sweep searched the Irish civil registers, and the Cawleys were in Manhattan. A single search of American records against both parents' names returns four of their children at once. Michael Joseph, born 5 March 1911 in New York, died 5 August 1921 aged ten. Edward, born 21 March 1916, died 17 March 1917, four days short of his first birthday. Martin, our Marty, born 1 May 1920, died 28 January 1999. And Margaret, born 18 November 1921, who later married a Nelson and died in 1996. Two of the six known children died as children. It also settles a question that has been open here for a long time. Because Anne was married to Michael Cawley and raising his family in Manhattan through the 1910s and 1920s, she cannot also be the Mrs A. J. Farley named in Sergeant Peter Cooney's Army file. That leaves her sister Mary, which is exactly what the 1920 census and Helen Bridget's death certificate say. The Army wrote one sister's name onto another's married surname. There is a poignant coincidence in the dates. Mary's newborn Helen died on 21 June 1921 and was buried at St Raymond's. Six weeks later Anne's ten year old Michael Joseph died, on 5 August, and was buried on the 8th. Two sisters from the same Mayo townland, both living in Manhattan, burying a child each in the same summer. On the site itself, a layout bug has been fixed that had been quietly damaging every page. A rule intended to make links easy to tap on a phone was applying a forty eight pixel minimum to every link in the body text, which turned each one into a solid block that cannot break across a line. On Peter Cooney's page that pushed a link right underneath the sidebar and left a six hundred pixel hole in the middle of the article. The navigation was also being cut off mid word on wide screens. Both are corrected.
Two days old, and the certificate says where they took her
Helen Bridget Farley was born in Manhattan on 19 June 1921 and died on the 21st, two days old. Until today all this archive knew was that she had existed and died. Her death certificate has now been read, and it answers the question her family has been asking. She is buried at St Raymond's Cemetery in the Bronx. The certificate is New York City number 15717 of 1921, and field seventeen, place of burial, says St Raymond. She was buried on 22 June, the day after she died and the same day the certificate was filed. The cause of death is given as congenital atelectasis, which means her lungs never fully opened. She died in a tenement in West 15th Street. Her father is entered as Andrew J. Farley, born Ireland, and her mother as Mary Cooney, born Ireland. What the certificate does not give is the grave. She has no memorial of her own anywhere: a search of every Farley at Old St Raymond's returns forty one memorials and not one of them is her. So she is either in an unmarked grave or, more likely for a two day old child in 1921, in somebody else's. The cemetery office holds an interment book that would give the section, the grave, and the name of whoever owned the plot. That last name is the one worth having, because it would tell us which family took her in. One earlier guess has to be withdrawn. Her father Andrew was buried at Holy Cross in North Arlington in 1969 in a grave where somebody already lay beneath him, and that space looked like a good candidate for her. It is not. She was in the Bronx from the beginning. The person under her father is now almost certainly her mother Mary, who died in 1949 and whose burial has never been recorded either.
Ruth Noga, aged twelve, living as somebody's niece
Ruth Cacace was born Ruth Noga, and until this week we did not even know that much. Now we know something about her childhood, and it is not a comfortable thing. The 1920 census returns exactly one Ruth Noga in the whole of New York State, and she is ours: twelve years old, born in New York, living in Brooklyn. But the column recording her relationship to the head of the household does not say daughter. It says niece. At twelve she was being raised by an aunt or an uncle rather than by her parents, and seven years later she married Carmine Cacace at seventeen. Something had happened to John Noga and Florence Meyer before 1920, and we do not yet know what it was. Death, separation, or simply not being able to keep her. The census does record both her parents as born in New York, which tells us the Noga and Meyer families had arrived a generation earlier than we assumed. Her ancestry on that side is probably three or four generations deep in Brooklyn, not one. We also tested a candidate for her father and threw it out, which is worth describing because the temptation was real. A John Noga of Brooklyn, born around 1885 in the United States to Charles Noga and Agnes Kluska, both of them born in Poland, died in September 1940. Right name, right borough, American born exactly as the census requires, with a Polish immigrant generation sitting neatly behind him. It would have given us a whole new branch in one move. His death record says he was single, and that he worked as a longshoreman. A man who never married did not father Ruth. So he is written down as rejected rather than quietly forgotten, because otherwise somebody finds him again in six months and attaches him, and a coincidence of name and place becomes an ancestry. That is exactly how this archive got Peter Cooney wrong, and how it got Carmine's birthplace wrong. This time it was caught before anything went into the tree.
A salesman in plastics, and a column that contradicts the one beside it
Pressing further on the Jones line has turned up the 1950 census, and with it a man in more focus. Howard Graham Jones was forty five, still at Great Neck, and working as a salesman in the plastics industry, which in 1950 was about as modern a trade as existed. The enumerator spelled it Plasties and his job Saleman, and the index has faithfully preserved both. His later life comes from Social Security records. He filed his claim on 29 March 1969, three weeks before his sixty fifth birthday, by which time he had moved along the north shore to Sea Cliff. He died in January 1996 at ninety one. The 1950 page also confirms the correction we made yesterday. Helen is recorded there as forty, which agrees exactly with her age of thirty in 1940 and puts her birth around 1910, twenty years later than the 1930 this archive had been carrying. Two independent censuses now say the same thing. But the same page contains something that does not add up, and rather than quietly choose a side we have written both down. The column recording Howard's marital status says Divorced. Helen is enumerated on the line directly beneath his, in the same household, and the index reads her as his wife. Those two facts sit an inch apart on the same sheet and cannot both be straightforwardly true. It may be a slip in one column. It may be that they divorced and went on sharing a house, which happened more often than tidy family trees suggest. The original sheet needs looking at before anyone decides. What we still cannot reach is the generation above him. Howard Graham Jones senior and Anna Bateman remain nothing but names passed down in a family tree, with no record behind either. Neither census names them, the Social Security file has no space for parents, and a search for a Jones father with a Bateman mother returned 381 American records without our couple among them. It will take his birth certificate, or his marriage to Helen, and that marriage would also tell us whether her name really was Brown.
A fifteen year old crossing in wartime, and a great grandmother born twenty years too late
Two branches on Gina's side moved today. The Jones line had eight people in this archive and not a single source between them. Everything came from a family tree and nothing had ever been checked. It now has a household. The 1940 census finds Howard Graham Jones at Great Neck, out on Long Island, aged thirty five and born in New York, with his wife Helen, aged thirty, and their three children. The census also notes they had been living in the same place in 1935. There is a fourth adult in the house, a woman of twenty one named Anne Bart, whose relationship the index does not give and who was most likely a servant or a boarder. That record also caught a plain error of ours. Helen was recorded here as born in 1930, which cannot be true, because her eldest son was born in 1934. The census gives her age as thirty, so she was born about 1910. We had her twenty years too late. Somebody had probably typed a marriage year into the birth field long ago and nobody had looked since. Two things about her stay unresolved rather than papered over. We call her Helen E. Brown; the census calls her Helen B. Jones, and the middle initial does not match. And the surname Brown appears in no record at all, only in the family tree. On the Cacace side, the 1930 census turned up under the misspelling Carace, which is why every earlier search missed it, and it answers when Carmine came over. He arrived in 1916, aged fifteen. That is worth sitting with. Italy had been at war since May 1915 and the Atlantic crossing was under submarine threat. Whatever put a fifteen year old boy from Naples on a ship to Brooklyn in 1916, it was not an ordinary passage, and we do not yet know what it was. The same page shows the family in 1930: Carmine thirty, Ruth twenty one and born in New York, and three small children, the youngest of whom, Frederick, is entered as nought days old. The enumerator arrived within a day of his birth.
A few hours ago this archive concluded that the Cacace side could go no further without buying a document, because Carmine Cacace was born in New York and his immigrant parents could not be picked out of the fifty or so Cacace households in Brooklyn. That conclusion was wrong, and wrong in an instructive way. Carmine was not born in New York. He was born at Naples in February 1901. The reason we thought otherwise is that an earlier pass read the birthplace column of the 1950 census as N.Y. and everything was built on top of that one hard-to-read word. Five records say Italy: two naturalisation entries, his draft card, and the 1940 and 1950 census indexes. And the decisive point is simply that he was naturalised at all. A man born in New York does not petition to become a citizen. This is the second time in two days that a single doubtful reading has been treated as settled and had a structure built on it. Once the premise was fixed, one record answered everything. Carmine married at Brooklyn on 28 February 1927, and the marriage record names his parents: Nicola Cacace and Caterina Picone. That is the Italian generation, named. The same line gives the bride as Ruth Noga, aged seventeen, which supplies Ruth's maiden surname, the one thing this archive had listed as the single most useful outstanding question on her side of the family. Two people called John Noga and Florence Meyer also appear on the record, very probably her parents, though the index does not say so. If they are, then Ruth was not Italian at all, since Noga is a Polish name and Meyer German, and this was a mixed marriage in 1927 Brooklyn. There is a quiet confirmation in the children's names. Carmine and Ruth called two of theirs Nicholas and Catherine, which is Nicola and Caterina, the grandparents, exactly as Italian families named. That agrees with the marriage record without depending on it. Carmine married in February 1927 and became an American citizen in October of the same year, certificate number 142652. That certificate has a petition file behind it, and petition files give the exact town, the ship and the date of arrival. Naples is a city of half a million. Narrow it to a comune and the Italian civil registers are free online, which would reach Nicola and Caterina's own marriage.
Islandeady, typed onto an American index by a clerk who had never heard of it
Sergeant Peter Cooney's identification is now proved, and by a source none of us had looked at. The United States Veterans Administration kept a master index of servicemen between 1917 and 1940. Its card for Peter Cooney gives his birth as 4 September 1890, his death as 14 October 1918, and his residence place as Islndeady Castlebar, Ireland. Islndeady is Islandeady. It is the parish and the civil registration district that contains Derrycriff, the townland where Peter Coony the farmer and Margaret Needham had their son in October 1890, and where the father made his cross on the birth register because he could not write his own name. Some clerk in a United States government office, who had certainly never heard of the place and was working from what somebody said out loud, typed it down as it sounded. That single misspelling is the strongest evidence in the whole question. It is completely independent of everything else we had. It does not come from the Graves Registration card, or from the regimental roster, or from the Army correspondence naming his three sisters. Four separate lines of evidence now point the same way, and they were produced by different offices for different purposes. Which means Father Duffy's remark that the sergeant had been out with the regiment in ninety eight is simply a mistake, and can be set down as one. He was writing from memory about hundreds of dead men, and he attached an older namesake's service history to the wrong Cooney. There was another in the regiment, Michael of Company C, who turns up throughout the same book. One small human thing remains. Three records give three birthdays, all in 1890: the Irish register says 15 October, his gravestone memorial says 22 September, the veterans' index says 4 September. People who emigrated often gave a birthday they had never once seen written down. He knew the year. The day was whatever he had been told.
Two townlands that are probably one, and a third cross from Patrick Condon
The 1868 register of Margaret Condon's birth has been read, and it does two things. It gives Patrick Condon a third documented cross. He made his mark at his marriage in 1865, again here in June 1868, and again registering Catherine in 1874. Thirteen years and three records, and he could not write on any of them. The entry is unusually explicit about it: where other registers write his name around a small X, this one reads simply, The Mark of Patrick Condon, Father, Mohera. The second thing is a correction to our own map. This archive has been carrying two separate County Cork townlands for this family, Maghna and Moherna, placed about nine kilometres apart. Both were entered with approximate coordinates, meaning nobody had ever verified either. The 1868 register plainly reads Mohera, in the parish of Castlelyons, both as the place of birth and as Patrick's dwelling. His 1865 marriage gives his residence as Maghera. Those are almost certainly one place with three spellings, which matters because the family's story here has been partly a story of movement, tracing them from Maghna to Moherna to Rathcarrig to Flemingstown. If the first two are the same townland then that first move never happened and only the later ones are real. It has not been merged. Both place records now carry the alternative spellings and a note explaining the problem, and the question stays open, because resolving a conflict by quietly overwriting one side of it is how this archive has gone wrong before. Verifying the townland needs the Irish placename databases, and both of them refused automated access today. That one is for a human with a browser.
The Cacace side of the family had been built out of a single page of the 1950 census and left there. It now has its burials. Carmine Cacace and his wife Ruth are at Pinelawn Memorial Park in East Farmingdale, out on Long Island, and they are in the same grave: section 60, block 3, range 30, plot D, grave 114. Carmine was buried on 24 May 1978, three days after he died. Ruth was laid in beside him on 7 September 2004, twenty six years later. The same pattern repeats twice more on the Spatarella side. Frank Spatarella and Caroline, whose maiden name the memorial gives as Zollo, share one grave and died nine months apart in the same year. Salvatore Spatarella died in 1977 and Marie was buried with him in 2018, forty one years afterward. Three couples, three graves, and in every case one of them waited a long time. Their dates until now rested on compiled profiles; they are now carried by memorial records with plot numbers. The more useful result is knowing exactly where this line stops. Carmine was born in New York in February 1901, which means his parents were the ones who came from Italy, and they are the next real target. Two things now rule out the easy routes. They are not at Pinelawn, because every one of the eleven Cacace burials there begins with Carmine's own birth year or later. And the surname will not find them, because a sweep of the 1950 census turned up around fifty separate Cacace households in Brooklyn alone. It is an ordinary Neapolitan name. Choosing a likely family out of fifty would be guessing, and this archive has already had to undo one bad merge made exactly that way. So no amount of free searching bridges this gap. It needs one document that names a parent, and the best of them is the marriage record of Carmine and Ruth, because it would name both sets of parents and also give us Ruth's maiden name, which we still do not know.
The card was on the disk all along: three weeks in the open, and a wrong number
Having put Sergeant Peter Cooney's identification back together, the obvious next step was his Graves Registration card. It turned out to be sitting in our own document folder, cited but never actually looked at. It corrects something we had wrong and adds a good deal we did not have. First the correction. This archive listed his Army serial number as 49525. That is not his serial number. The card's first column is headed File No. and carries 49525; the column beside it, headed Army Serial No., reads 91675. An earlier transcription had swapped the two, and every note since had repeated it. His serial number is 91675, and that is the number that opens his service record and burial file at the National Archives. Then the part that stays with you. Peter was killed on 14 October 1918, and the card gives his date of burial as 6 November. He lay where he fell for twenty three days. The ground around Landres-et-Saint-Georges was still being fought over and nobody could reach him. When they finally did, he went into an isolated grave numbered 4, on commune land, fixed by a map reference rather than a place name. The card notes that the marker was a cross, and that his identification tags were both buried with his body and fastened to the marker, so that whoever came next would know. Somebody did come. In June 1919 he was moved to Romagne-sous-Montfaucon, and in September 1921 to his permanent grave at Meuse-Argonne. His family were not told about that last move until 7 November 1923, five years and three weeks after he died. The card also records that a photograph was taken of his grave, number D-33650, and a sketch made of the plot. Neither has been obtained yet. And it shows his sister Annie moving house between the Army's letters, from 37 West 50th Street to 13 East 63rd, which gives two addresses to look for her in.
Sergeant Peter Cooney of the Fighting 69th is ours after all — a correction
This archive had it wrong, and the correction matters enough to say plainly. Peter Cooney, born at Derrycriff in County Mayo in October 1890 to a father who could not write his own name, emigrated to New York and was killed in action in France on 14 October 1918 as a sergeant in Company M of the 165th Infantry, the old Fighting 69th. He was two days short of his twenty eighth birthday. Until today this page said that identification was chronologically impossible. The reason was a single clause in Father Francis Duffy's memoir of the regiment, where the chaplain listed the sergeants he had lost and wrote that Sergeant Peter Cooney of M Company had been out with the regiment in ninety eight. Peter was eight years old in 1898. Earlier research took that one line as decisive and stripped his death, his burial and his service out of the record as somebody else's. That was the wrong order of evidence. The Army's own paperwork says otherwise, and says it repeatedly. His Graves Registration card names his next of kin as Mrs Annie Farley, sister. The regimental roster gives his address as 37 West 50th Street in New York, which is where his sister Annie lived. Army correspondence names three sisters: Bridget Horan of Jersey City, Nellie Cooney, and Annie Farley. Peter and Margaret's daughters at Derrycriff were Bridget, who married a Horan, Ellen, who would have been called Nellie, and Anne, who married a Farley. His sister Bridget wrote to the Army about him in 1929. His memorial gives his birth year as 1890. The correspondence says his mother died in Ireland around 1921, and Margaret Needham died at Derrycreeve in March 1922. Three sisters by name, a shared street, the right birth year and a mother dying in the right place at the right time. Duffy was writing from memory about hundreds of men, mid sentence, in a paragraph that runs through a dozen names in a row. The regiment had another Cooney, Michael of Company C, who turns up all through the same book. A chaplain misremembering which Cooney was the old soldier is ordinary. What is not ordinary is letting that outrank a War Department form. The details we had set aside come back with him. A shell fragment struck his left side in the advance toward Landres-et-Saint-Georges. He asked for a stretcher and died about fifteen minutes later in a shell hole, before the bearers could reach him. He was buried three times: first in an isolated grave on commune land where he fell, then moved in June 1919 to Romagne-sous-Montfaucon, and finally in September 1921, nearly three years after his death, to grave 19, row 37, block E at Meuse-Argonne, where he is now. His serial number was 49525.
Who signed and who made a cross, and what it actually tracked
Four register pages have now been read closely, from four different districts, and together they say something the individual entries could not. In Islandeady in Mayo, thirty nine informants out of forty made a cross. But in Mitchelstown, in Ventry and in Galbally the split is close to even. So the mark is not one thing everywhere. In Mayo it was simply what everybody did. In Cork, Kerry and Limerick it separated people inside the same parish, and on those pages it separates them by trade with almost no exceptions. Take the Galbally page of May 1873. Four people signed their own names: Roger Hennessy, a farmer. John Fitzgerald, a farmer. Mary Walsh, married to the relieving officer. Timothy Fogarty, a farmer. Six made crosses, and every one of them belonged to a labourer's household, except the shoemaker's wife. That is the whole division. A man with land had reason to learn a signature and somewhere to practise one. A labourer's wife had neither. It is worth saying plainly because these crosses now appear on a good many pages here, and they are not a remark about anybody's mind. They are a record of what work did to people. The same page adds John Fitzgerald's fourth documented signature, and gives his daughter Ellen a properly sourced birth at Cullane South on 2 May 1873, where before she had a date and nothing behind it. Her mother appears there as Maggie, the only time anyone wrote her name informally. And a caution worth keeping: there are two John Fitzgeralds in this archive. The Cork one signed his name four times. The Kerry one, registering his son Patrick at Ventry in 1895, made a cross. Both were farmers, so this does not prove they are different men. It is simply one more thing that would have to be coincidence if they were the same.
The Cooneys' marriage line, and a caution about reading what you expect
Peter Cooney and Margaret Needham were married in the Roman Catholic chapel at Islandeady on 26 February 1884. Their marriage register has now been read at the signature line, and Peter's cross is plain: the clerk wrote Peter Cooney and left a gap, and in the gap is the X, with the word his above it and mark below. That pushes his illiteracy back to his wedding day, matching the crosses he made registering Bridget in 1886 and Peter in 1890. Margaret's line is worth being careful about. It sits directly beneath his and carries the clerk's mark annotation in exactly the same position, which is the form used when someone could not sign. But the cross itself is faint in the copy that survives, and at this resolution it cannot be called plainly visible. So it is recorded as consistent with the clear cross she made in 1894, not as separate proof of its own. That distinction matters more than it might seem. It would be very easy to look at that line, knowing already that Margaret could not write in 1894, and see the cross one expects to see. This project has made that mistake before, reading an age off a register as the number the arithmetic wanted. The honest version is that the annotation is there and the stroke is not clearly legible, and both of those things get written down.
The best thing a civil register holds is not a date. It is the moment someone had to put pen to paper, and what happened next. Three more of those have now been cut out of the page and added to the people they belong to. Margaret Needham's cross, made at Derrycriff on 21 August 1894 when she registered her daughter Ellen. The registrar wrote out Margaret Cooney and she made the X between the two words, and underneath it says mother, and under that Derrycriff. Then the marriage line from Castlelyons on 7 September 1865, which carries both Patrick Condon and Mary Ryan at once. The clerk wrote Patrick Condon on the upper line and Mary Ryan on the lower, and in the gap in each name there is a cross, with the word Mark written small beneath it. Two people beginning a marriage, neither able to sign for it. Mary's own cross has been cut separately so it sits on her page too. These are not signatures in the sense of handwriting we could identify. The letters around them belong to a clerk. But the cross itself is theirs. Someone handed each of them a pen, and that stroke is the mark they made with it, and it is very nearly the only physical trace any of them left. Four such images were already on the site. There are now seven.
Married with two crosses, and a labourer named at last
Patrick Condon and Mary Ryan were married in the Roman Catholic chapel at Castlelyons on 7 September 1865, by Father Daniel McSwiney. Patrick was twenty five, a bachelor and a labourer living at Maghera. Mary was twenty two and still resident with her parents at Coole. Where the register asks the bride and groom to sign, both of them made a cross instead. That pushes what we know about them back nine years. We already knew Patrick could not write in 1874 and Mary could not in 1876 and 1878. Now we can see that neither of them could write on the day they married either, which means there was never a point in their documented lives when either could sign a name. Their two witnesses that day, Edmond Smythe and Anne Wynne, both wrote theirs out in full. The register also gives something to a man who until now was only a name on a page. Mary's father James Ryan had no dates, no occupation and no place attached to him at all. The marriage line records him as a labourer, and Mary marrying out of her parents' house at Coole puts him there in September 1865. He is also not marked deceased, unlike Patrick's father John Condon on the same line, so we know he was alive that day. It is not much, but it is the difference between a name and a person.
A shopkeeper signs, a labourer's wife makes a cross, two rows apart
Two more registers that had been sitting on disk, cited but never read, from Mitchelstown in County Cork. They show that Mary Ryan could not write. She registered her son James in May 1876 and her daughter Mary in July 1878, and both times she made her mark. Her husband Patrick had done the same registering Catherine in 1874, so neither parent in that house could sign a name. The registers also confirm Patrick as a labourer three times over, which had rested on a single entry before. But the reason to publish this is what surrounds those marks. In Mayo, where the Cooneys lived, we found that thirty-nine of forty informants made a cross and only one man in the whole stretch signed. Mitchelstown is not like that at all. On the page holding Mary's entry, five informants made marks and five signed their names. Half that page could write. So in Cork the cross does not describe a whole community the way it does in Mayo. It separates people inside one, and on this page it separates them by trade: the tailor signs, the shopkeeper signs, the labourers' wives make their marks. The clearest version of it is two rows apart in the same book, in the same week, before the same registrar. Entry 436 is Redmond Condon, a shopkeeper of Mitchelstown, who signed his name in a clear and practised hand. Entry 437 is Mary Condon of Flemingstown, a labourer's wife, who made a cross. The same surname and the same town. What differed was not the family they came from but the work they did.
Four birth registers from Derrycriff in County Mayo had been downloaded and cited but never actually read. Reading them answers something the indexes never do. When Peter Cooney went in to register his daughter Bridget in March 1886, the registrar wrote his name for him and Peter made a cross between the words. He did the same when he registered his son Peter in 1890 — four years apart, the same mark. In August 1894 his wife Margaret went in to register their daughter Ellen, and she made a cross too. Neither parent in that house could write their own name. A third cross belongs to Bridget Needham, who was present at Anne's birth in 1888. She shares Margaret's maiden surname and she was in the room, which makes her very likely Margaret's mother or sister, though the register does not say so and we should not pretend it does. What makes these crosses worth publishing is the company they keep. On the 1886 page every one of the ten informants made a mark. On the 1888 page, all ten. On the 1894 page, all ten. Across about forty entries from this district between 1886 and 1894, exactly one man signed his own name in a flowing hand: John Kirkpatrick of Cloghar, registering his son in 1890. So the cross is not a note about what Peter and Margaret lacked. In that townland, in those years, the cross was ordinary and the signature was the exception. It is one of the few marks either of them left on paper, and it is theirs — the pen was in their hand.
A search run to the end, and the harbour crossing dated
Ann Galvin told two censuses she had borne eighteen children. Twelve are named. The route that looked most promising for the other six was New Jersey marriage records, which name both parents of the couple marrying — and it did work, giving up three Bayonne weddings and one exact Syracuse birth. It has now been followed to its end, and the end is close by. Across the whole United States only eleven indexed records name both a Leahey father and a Galvin mother, and four of them are this family's; the rest belong to unrelated families in Minnesota and Boston. A sweep of 34,488 Bayonne marriages added nothing. The 1870 census, which would have caught children who died before 1880, cannot be reached either: no matching household appears in the index at all. Recording a route as exhausted is worth as much as recording a discovery, because it stops the same ground being walked twice. What the searching did settle is when the family crossed the harbour. The 1880 census lists all eleven children then at home with a birthplace each, and read in order the column says New York through Mary Ellen in 1875 and New Jersey from Peter in 1876. The move belongs to those two years. The six who are missing were most likely infants: the known births leave gaps at about 1862, 1865, 1869, 1871, 1873, 1877 and 1881, and children who died between censuses rarely entered any index that bothered to name their parents. Finding them now means the cemetery office, the parish register, and pages turned one at a time.
Syracuse confirmed by a record, and three sons proved
Until now the family's Syracuse years rested entirely on obituaries. James Leahey's Social Security application gives his birth outright: 27 February 1864 at Syracuse, New York, with his parents named as Dennis Leahey and Anna Galvin. That is the first exact date for any of the Syracuse-born children and the first record, rather than a newspaper, to put the family there. The same search turned up three Bayonne marriages, each naming the same parents: Matthew to Catherine Meaney on 10 June 1889, James to Mary Mullins on 7 April 1890, and Dennis to Anna Meaney on 7 June 1891. Their descent from Dennis and Ann had rested on obituaries; now it rests on records.
A possible Syracuse household, and why we have not claimed it
The 1865 New York State Census is now searchable, and every Lahey, Leahy and Lehy household in Onondaga County has been listed. One is tempting: in Ward 04 of Syracuse lived Dennis and Anna with children Catharine, John and James — three of the right names, in the right city, in the right decade, and James Leahey's 1947 obituary does say he was born in Syracuse. It has not been claimed, for two reasons. Dennis's age is ten years out against the 1860 census, and the couple's age order inverts. And Catharine at nine would have been born about 1856, three years before Dennis and Ann married. Checking the original page against the index would settle it. Matching forenames in an Irish ward is exactly the kind of reasoning that has misled this archive before.
A likely death for Maurice Fitzgerald, and a hint of his wife
Five men called Maurice Fitzgerald died in the right districts at a plausible age. Reading the registers, one stands out: a farmer who died at Ballylanders on 10 July 1883, aged 64, of heart disease. Ballylanders is this family's own place - his supposed son John was raising children at Cullane, Ballylanders in those years - and the occupation and age both fit. It is not proof. An Irish death register never names parents, so nothing in it states the relationship. But the informant is interesting in its own right: Margaret Fitzgerald, wife of the deceased, who signed with a cross. If this is the right man, that is John Fitzgerald's mother, a woman this archive did not know existed.
A small thing in the registers turns out to be useful. When a father had already died, the clerk wrote deceased under his name — as he did for Thomas Needham in 1884 and for a neighbour on the very page that records John Fitzgerald's marriage. Maurice Fitzgerald and Richard Fraher have no such note in 1870, and neither do Andrew Farrelly and Patrick Boyle in 1879. So all four were alive on those days, which gives four men who had no dates at all a point on the calendar, and puts their deaths inside the period covered by civil registration. Searching for them has so far found a shortlist for Maurice and controlled blanks for the other three.
Peter Cooney of Derrycriff could not write. He registered his daughter Mary's birth in 1884 and his son Peter's in 1890, and both times the registrar wrote his name around the cross he made. His mark is now on his page, cut from the 1890 register.
Where a register survives, person pages show the actual signature or mark, cropped from the original document. John Fitzgerald wrote his own name when he registered his son David's birth in 1875 - you can read John FitzGerald, Father, Cullane in his hand. His wife Margaret could not write, and on their 1870 marriage register the clerk wrote her name around the cross she made: Mary her x mark Fraher. It is the closest thing to meeting them.
Born 11 December 1884 at Derrycriff, Islandeady, County Mayo, to Peter Cooney, a farmer who registered the birth with a cross, and Margaret Needham. She is one of the candidates for the Mary Cooney who married Andrew Farley in New York, though the 1920 census names his wife only as Mary, aged 34, born Ireland - close to this birth but not an exact match, and with no parents named. It stays a possible identification rather than a proved one.
Person pages now show whether someone could sign their own name. It is one of the few things a civil register records about a person rather than about an event, and on the same page you can see the difference between neighbours. John Fitzgerald of Cullane signed his own name three times - at his marriage in 1870 and when registering his sons Richard and David - while most fathers on those same pages signed with a cross. His wife Margaret Fraher made her mark. So did Patrick Condon, Patrick Farrelly and Peter Cooney, each registering a child's birth.
A correction: the Fitzgeralds and Condons were neighbours, not distant families
Richard Fitzgerald's 1872 birth register shows that Cullane, Ballylanders sits in the Galbally registrar's district, County Limerick, inside the Mitchelstown Poor Law Union. Flemingstown, where Catherine Condon was born, is in the same union. An earlier note here described the two families as living twenty miles apart in different counties, which overstated it and made their marriage look odd. They were within one union, and the marriage was an ordinary local one. Galbally had never been searched before this month, because it was missing from the district lists we were using.
David Fitzgerald's birth at Cullane, and a small human detail
His birth register has been read: born 28 March 1875 at Cullane, Ballylanders, to John Fitzgerald, a farmer of Cullane, and Margaret Fitzgerald formerly Fraher. John registered the birth himself and signed his own name. That is worth noticing: most fathers on the same page of the register signed with a cross, and at his own marriage five years earlier John signed while his wife Margaret made her mark. David is John Joseph Fitzgerald's father, so this is the direct line.
Sergeant Peter Cooney: killed in action, and his serial number recovered
His Graves Registration card records the cause of death in two letters: k/a, killed in action, on 14 October 1918 with Company M of the 165th Infantry. The Army set down nothing more specific. The card also gives his serial number, 49525 - which earlier research had listed as unobtainable because the American Battle Monuments Commission site would not yield it. It was on disk in this archive all along. That number opens his individual service record. He was first buried on 6 November 1918 in Grave 4, an isolated grave, in the commune of Landres-et-Saint-Georges.
Margaret Needham's father was already dead when she married in February 1884. His death has now been searched for properly and is not there. Adult Needham deaths in the Westport district between 1864 and 1884 return twenty-five results and not one is a Thomas, so this is a controlled absence rather than a failed search. The likeliest explanation is that he died before 1 January 1864, when Irish civil registration of deaths began - in which case no civil record exists at all, and only a parish burial register or a gravestone could reach him.
Peter Cooney was born at Derrycriff, one day short of the anniversary of his death
His birth register has been read: born 15 October 1890 at Derrycriff, Islandeady, County Mayo, to Peter Cooney, a farmer who registered the birth himself by mark, and Margaret Needham. He was killed in the Meuse-Argonne on 14 October 1918 - a day before his twenty-eighth birthday. The archive had carried 16 October as his birth date; both are now recorded. It also sharpens an old problem: Father Duffy's regimental history says Sergeant Peter Cooney was out with the regiment in 1898, which no man born in 1890 could have been.
Two more generations on the Cooney and Needham lines
The marriage register for Peter Cooney and Margaret Needham, 26 February 1884 at the Roman Catholic Chapel of Islandeady, names Peter's father as John Cooney, labourer, and Margaret's father as Thomas Needham, landholder, already deceased. It also settles a place that had caused confusion: the chapel and registrar's district are Islandeady in County Mayo, not Island Eddy in County Galway.
Two more generations on the Cooney and Needham lines
The marriage register for Peter Cooney and Margaret Needham, 26 February 1884 at the Roman Catholic Chapel of Islandeady, names Peter's father as John Cooney, labourer, and Margaret's father as Thomas Needham, landholder, already deceased. It also settles a place that had caused confusion: the chapel and registrar's district are Islandeady in County Mayo, not Island Eddy in County Galway.
Catherine Condon's birth settles a question about the direct line
It had been suggested that Catherine Condon might come from the Condons of Cullane in County Limerick, the same townland as the Fitzgeralds she married into, rather than from the Flemingstown family in Cork. Her own birth register answers it: born 19 April 1874 at Flemingstown, daughter of Patrick Condon, a labourer, and Mary Ryan. Her father registered the birth himself, by mark. The Cullane Condons are a different family. Catherine is John Joseph Fitzgerald's mother, so this sits on the direct line.
The civil marriage register for John Fitzgerald and Margaret Fraher, married at the Roman Catholic Chapel of Kildorrery on 25 February 1870, names the groom's father as Maurice FitzGerald, farmer, and the bride's father as Richard Fraher, farmer. Maurice had been the most troublesome name in this archive: one earlier attempt to confirm him relied on a grandson being christened Morice, which is not evidence, and another moved the family to Cork on a guess and was disproved. This register states his name and occupation outright. It still gives him no dates, no townland of his own and no wife.
The family remembered that Andrew Farrelly's brother Bernard died in an accident. The civil death register confirms it: he died at Clara, County Cavan on 8 October 1915, aged 34, a bachelor and a farmer's son. The certified cause reads “Probably injury to spine caused by fall from cart; 24 hours; no medical attendant”. His brother Patrick was present at the death.
Andrew Farrelly's parents, and two more generations
Patrick Farrelly of Clara married Margaret Boyle of Gartbratten on 8 May 1879 at Miltown chapel. The register names Patrick's father as Andrew Farrelly, farmer, and Margaret's father as Patrick Boyle, farmer — two generations that were previously unknown, and the Boyle line had none at all. Patrick died at Clara on 10 June 1927 aged 70 and Margaret on 9 April 1928 aged 70, each reported by one of their children who was present at the death.
Andrew Farrelly's baptism, and the family townland
Found in the Kildallen parish register: “Andrew | Farrelly | 4 | July | Patt Farrelly / a Mary Boyle | Clara”. This finally supplies the townland, Clara, which was illegible on his civil birth register. The baptism is dated six days before the civil birth date of 10 July, which is an ordinary pattern — births were often reported to the registrar with a later date to stay inside the registration window.
Both census households were recovered. In 1901: Patrick 47 a farmer, Marget 46, Bernard 19 a farmer's son, Ellen 16 a dressmaker, Terence 13 and Patrick 10 at school. By 1911 Ellen has left and the three sons are all labourers. Andrew is in neither census — he would have been twenty in 1901, so he had already emigrated, which is the first evidence of when he left Ireland.
Malachy married Catherine Agnes Tobin in 1898. Their daughter Helen was born in January 1899, Catherine died in February 1901 aged 27, and Helen died in June 1903 aged four. Holy Name plot F-E-33 holds five of them in sequence: Catherine, Helen, an infant Mary, his second wife Margaret Murphy, and Malachy himself. Tobin is a surname new to this archive.
“Corker” was a trade, not a place
Dennis Leahey's undertaker's register entry reads “Aged 70y, born Ireland, Corker”, which looked as though it contradicted the Templemore baptism by placing him in County Cork. Reading the other rows on the same page showed the notes column runs age, birthplace, occupation — and the Census Bureau's occupational index lists “Corker, boat or ship”. He was a boat caulker. The Tipperary origin stands.
Every change made to the archive, taken automatically from the research record.
Leah Adelaine Stillwell: her death certificate is not online, and a Harry rejected
Measured NYC Municipal Archives coverage for Manhattan rather than
assuming it: 39,479 births digitised for 1889, and no deaths at all
for 1889, 1900 or 1910 -- deaths begin somewhere before 1920. Her
certificate 3335 must be ordered. The site answers "No Results" when
there is no digitised image, not when there is no record, so its
negatives are worthless without a coverage check; recorded in TOOLING.
Put two dates side by side that the archive had never compared: her
son Harry's tree-supplied birth of January 1889 against her death on
30 January 1889. Searched the complete Manhattan 1889 birth index --
39 Mayer births, no Harry -- and examined the one Harry Meyer of that
year, certificate 1603. Father Hermann Meyer, a German saloon keeper;
mother Annie Kropf. Rejected on the face of the document.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Sweep inventory: 444 corroborated on a second table, and a Fitzgerald refused
Frame 00442 is an independent lift table giving the battalion the same
444 as the troop list on 00379, so the strength is corroborated rather
than read twice.
Controls: a loose sweep for manglings found only the 273rd and 253rd
Signal Construction Companies and the 436th Signal Construction
Battalion. And the only Fitzgerald in a thousand frames is Capt
Francis R. Fitzgerald, Asst Adjutant General (Personnel) on the I
Corps staff -- wrong rank, branch and given name. Nothing attached.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Four hundred and forty-four: the battalion counted, and the Pacific filled in
Swept all 1,000 frames of the I Corps reel in fifteen minutes with six
parallel workers, splitting every page into columns before OCR and
keeping only the text. Five frames name the battalion.
I Corps's troop list for the occupation of Japan gives it outright:
34th Sig L Cons Bn, 444, Luzon. That closes the strength question and
disagrees with Europe -- Company B alone was 249 enlisted in August
1944 -- because the battalion shrank. A unit's size is a date, not a
fact, and the archive now holds two dated measurements.
Also: he staged from Luzon; his battalion is named in the signal annex
for Operation BLACKLIST, on a page allotting naval gunfire to coloured
beaches six weeks after the surrender; and it was released from I
Corps on 1 October 1945 because Japan's own wire lines were intact
enough to use.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Drafted the NARA reproduction request for the battalion's unit history
Full citation chain (RG 338, Unit Histories, Entry UD 37042, box 5658,
NAID 1220770), a ready-to-send email, and the independent-researcher
route with the questions to ask.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
The battalion's own Historical Reports found at College Park, RG 338 box 5658
Searching the unit's later name -- Signal LIGHT Construction Battalion,
learned from the I Corps report in Japan -- returns eight records that
the old name never did. Seven are Pacific operations reels; the eighth
is the battalion's own narrative history. Unrestricted, paper only.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Japan: his battalion under I Corps, and marked for turnabout shipping
The least promising of Codex's 27 catalogue hits -- I Corps, a Pacific
formation -- turns out to be the one that fills the emptiest line on
his discharge. I Corps's Signal Section report puts the 34th Signal
Light Construction Battalion in the Osaka-Kyoto-Wakayama area and
records, by 1 October 1945, that it would revert to Kobe Base for
"turnabout shipping". He sailed for the United States on 19 November.
Recoverable only by splitting the two-column typescript down the
middle; the OCR interleaves the columns into nonsense otherwise.
Also confirms sideways that the unit's 1945 name was Signal *Light*
Construction Battalion, the T/O & E 11-25 retitling of February 1944
found separately while chasing its strength.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Did Wolfe live: the order says probably, and two searches that cannot say no
The Purple Heart order is dated six weeks after the wound and carries
no posthumous language, which is real evidence he was alive. ABMC does
not hold a Master Sergeant Robert L. Wolfe of Ohio -- but ABMC covers
only men left buried overseas, so that is support, not proof.
Recorded two uncontrolled-negative traps: AAD's free-text search does
not index serial numbers (Fitzgerald's own returns 0 of 8.7 million),
and ABMC's directory returns zero for any multi-word query. Also the
serial-prefix convention, controlled against Fitzgerald's own number.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Every occasion the battalion had a man wounded, against a family story
Told he once held a man's intestines in with his jacket, on a
minefield. Not found, and the entry says so. Set out the three
documented woundings instead: M/Sgt Robert L. Wolfe of Youngstown in
France on 1 September 1944 (name recovered by eye from GO 2, lost in
OCR); an unnamed man hit by shrapnel at Fischbach on 11 February 1945
and carried out of the open; and Schroeder and Thompson of Company D
at Rosport on 14 March 1945 -- two men, one company, one day, on the
Sauer opposite the Siegfried Line.
Rosport fits the minefield; Fischbach fits the scene; neither is
proof. And a medic treats whoever is in front of him, so the casualty
may belong to another unit entirely and be unfindable in these orders.
Coverage wall measured: 809 reels of morning reports for December
1944, zero for January 1945. Both 1945 incidents are past it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Withdraw the "already an NCO" inference; the battalion was at Fort Ord
His discharge gives "Highest grade held CPL" with no date, and MOS 673
"Medical NCO" is a job code, not a rank -- so the claim that he was
already an NCO in December 1944 was an inference stated as fact.
Corrected on his page and in finding 145.
The morning reports record grade changes outright (Co A, 3 Jan 1944:
"39375647 Miller Tec 4 -- Rd to Pvt"), so his promotion is findable.
Looking for it turned up January 1944 at Fort Ord, California --
released from Fourth Army to III Corps on 15 January, at sea for the
ETO by 11 February -- and no medical detachment reporting that month
at all.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
The Bulge, day by day: thirty-one morning reports mapped
Reel 465, frames 00423-00453 -- one form for each day of December
1944, SECRET when written. Jarville by Nancy 1-6 Dec; 1.5 miles NW of
the Cite de Valmont 7-21 Dec, including 16 December; Niederkorn,
Luxembourg 22-31 Dec, arriving at 1630 after five hours on the road on
the day Third Army turned north for Bastogne.
The places validate each other: Company B at Differdange P-663043 on
25 Dec against the detachment at Niederkorn P-665057 is a 1.4 km grid
difference, which is the true distance -- so the grid is metric to
100 m. Travel times match the distances at convoy speed.
30 December names four of his thirteen, promoted to pfc; one of them,
Mitchell Siegel 36357174, is also on the August company roster.
Rendered as a numbered journey and map on his page.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
TOOLING: NARA search-within cannot produce a negative; reels bisect on unit number
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Fifteen men: the medical detachment's own morning reports
Bisected Reel 572 on its filing order -- ascending by unit number,
branch ignored -- and found the 34th Signal Construction Battalion at
frames 540-695, between the 34th and 35th Evacuation Hospitals. Twenty
-eight frames fetched instead of four thousand.
Medical Detachment, 26 August 1944: 2 officers, 13 enlisted, signed
Fred Levy, Capt., MC; a ten-hour road move that day. He is one of
fifteen, and unnamed, which is what "no change" means.
Company B, 4 August 1944: 5 officers and 249 enlisted. The circulating
"about five hundred" for the battalion is too small.
Also controlled the catalogue search two ways and established that
NARA's own search-within does not reach the text its global index
holds -- a negative from search-within is worth nothing.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Bind documents to people by citation, and show them where they are cited
doc_matches_person() counted a two-word name overlap as identity, so
eight Irish civil registers naming a father "John Fitzgerald" attached
to John Joseph Fitzgerald of Jersey City, born 1913 -- and the Ventry
1889 marriage to fifteen pages. 126 of 195 documents already carried a
citation binding them to specific people; the generator never used it.
Order of authority is now citation, explicit id, unambiguous name.
Search targets and rejected candidates attach to nobody. Attachments
across the site: 586 -> 301.
The same citation says where on the page a document belongs, so 71
pages now render records inside the timeline event that cites them,
captioned by the event rather than by boilerplate.
27 real records attach nowhere and are left that way rather than
guessed; most name people this archive does not hold.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Five Signal Corps photographs of the 34th Signal Construction Battalion
Frames 0492-0496 of the 111-SCX shelflist carry five consecutive caption
cards for the battalion on the Moselle in November 1944, photographer
Grant. All five re-fetched and re-OCR'd independently before recording;
frame 0497 is the 33rd, so the run is exactly five. He is not named and
no card names any individual.
Also: the strength question recorded as open rather than answered, since
T/O & E 11-25 is not online free and the circulating figure is uncited.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Third Army GO 139: Fischbach, Company A, and a unit award - correcting myself
Codex downloaded and validated 594 fiche frames, then hit a Vision framework
failure in its sandbox and fell back to three Tesseract modes; I re-read the same
frames with Apple Vision here, which works fine outside that sandbox. Between the
two passes every frame of FICHE_1446, FICHE_87 and FICHE_89 has now been read.
CORRECTION. Hours earlier this archive judged a reported Meritorious Unit
Commendation for the 34th Signal Construction Battalion to be "probably a
conflation" with the 34th Signal COMPANY, which served in Italy. That was wrong.
General Orders 139, Headquarters Third United States Army, 14 June 1945, awards
the MERITORIOUS SERVICE UNIT PLAQUE to the 34th Signal Construction Battalion for
service from 20 October to 26 December 1944 - a period ending ten days into the
Bulge. The note is withdrawn on his page and replaced with the primary citation.
That order also settles the Third Army question by issuing from that
headquarters, and it is an award he almost certainly earned and which does NOT
appear on his discharge - a separation form of December 1945 would not carry a
unit award made that June. Recorded as the correction worth requesting.
NEW, and the best picture yet of what the battalion did: on 11 February 1945 a
Company A wire crew building communications installations near FISCHBACH,
LUXEMBOURG was shelled and a man hit by shrapnel. S/Sgt Frank E. Yelenchic
(33404030) and T/4 Chris Stako (33404012) left cover and carried him in; both
decorated for heroic achievement. Fitzgerald was the battalion's medical NCO.
CONTROLLED NEGATIVE: no John J. Fitzgerald and no ASN 32775493 in any of the 594
frames. The only Fitzgerald is 1st Lt Donald Fitzgerald, Coast Artillery,
Soldier's Medal for a fire aboard a mined ship. The sweep correctly finds the
battalion at FICHE_1446 frames 116-120 and FICHE_89 frame 34, so the reading
works. Other fiche in the series remain unopened - the catalogue returned 27.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Found his battalion's own General Orders at NARA St. Louis, declassified 2023
Pressing further produced the first PRIMARY documents about the 34th Signal
Construction Battalion - its own General Orders, RG 64, Army General Orders,
FICHE_1446 frames 116-120, National Archives at St. Louis, declassified under
E.O. 10501 in June 2023 and digitised free.
They establish:
- APO 403
- commanded by Major, later Lt Col, WALTER H. WOOD, with 1st Lt Russell B. Wood
as adjutant and 1st Lt Cornelius C. Sullivan as assistant adjutant
- at least a Company D
- and that it took fire: a man wounded in France on 1 September 1944; and on
14 March 1945 Sgt Robert D. Schroeder (39450468) of Spokane and T/5 Carlos L.
Thompson (38336348) of Harrisonburg, Louisiana, both Company D, wounded by
enemy action at ROSPORT, LUXEMBOURG - a village on the Sauer at the German
border.
John Fitzgerald is not named in these particular orders, which award Purple
Hearts; he was not wounded. Recorded with the route to finding him by name: the
fiche is 199 frames and only a sample was read, Good Conduct Medal orders are
what would name him, and two further fiche that matched the same search
(FICHE_87, FICHE_89) have never been opened. Image URL pattern and the four-digit
frame numbering are recorded so the next pass does not have to rediscover them.
This is also far better evidence than the unit narrative circulating on the web,
which three separate attempts have failed to verify.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
What his medals actually were, and a unit award that is probably another unit's
Asked whether he was decorated. Recorded plainly on his page.
He held no valour or achievement decoration - no Bronze Star, no Silver Star, no
Commendation Medal, and no Purple Heart (block 34: wounds received in action,
NONE). Of the six entries in his decorations block, one is not an award at all
(GO 33 WD 45 as amended is the order authorising his campaign credits) and three
are presence medals: the Victory Medal, the American Campaign Medal, and the
Asiatic-Pacific Campaign Medal from the voyage that arrived after the surrender.
Two are worth more. The EAME medal is the one carrying his five campaign stars -
credits rather than decorations, but they are what the ribbon bar says. And the
Good Conduct Medal is the only item that had to be earned and recommended: it
could be withheld or forfeited for disciplinary trouble.
Also flagged: web sources assert a Meritorious Unit Commendation for the 34th
Signal Construction Battalion. Probably a conflation - the same searches return a
9 December 1944 meritorious-service plaque for San Pietro, Cassino, Anzio, Rome
and Leghorn, which is ITALY, where his battalion never served. Those are the
honours of the 34th Signal COMPANY, a different unit with a near-identical name.
The Army page cited has the slug 34SIGCO and 403s. Recorded as unverified.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Not D-Day: what the Normandy star does and does not mean
Asked whether he was at Omaha Beach. His own discharge answers it, and the answer
is no.
The Normandy campaign credit runs 6 June to 24 July 1944 - seven weeks, not one
day. The Army separated the assault from the build-up with a device: "Normandy
with arrowhead" meant the landing of 6-7 June; Normandy without one meant landing
on 8 June or later, not with the assault force.
His decorations line carries NO arrowhead - only the campaign authority, the
Victory Medal, the EAME, American and Asiatic-Pacific service medals, and the
Good Conduct Medal. And his service-abroad line has him sailing for the ETO on
11 February 1944 and arriving on the 22nd, which is England, where the battalion
staged. Signal construction troops were not assault troops.
Recorded with the caveat that he may well have crossed Omaha later - the beaches
carried the build-up for months - but that is a guess, and a different thing from
6 June.
Also flagged on his page as the family story to guard against: five stars opening
with Normandy invite the D-Day conclusion. The genuinely unusual part of his war
is the other end of it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Pressed on the 34th Signal Construction Battalion: no unit book, but morning reports
Three results, two of them negatives worth having.
NO PUBLISHED UNIT HISTORY EXISTS. Checked against Controvich's Checklist of
Signal Corps Unit Histories and Personal Memoirs (Feb 2021), the standard
bibliography. The checklist prints the wartime designation in brackets under each
modern battalion number - [31st Signal Construction Bn], [35th Signal
Construction Bn] - followed by its publications. The 31st, 35th, 38th and 41st
all have wartime titles. Under the 34th there is no bracket and no wartime title,
only a 1955 Mohringen booklet about the postwar corps battalion. Controlled: the
same pass finds five other Signal Construction Battalions in the same sequence.
MORNING REPORTS ARE THE ROUTE TO HIS COMPANY. A National Archives Catalog search
for the battalion returns 27 results, mostly rolls of WWII morning reports, each
carrying about a thousand digitised images free to view. Morning reports list men
by company day by day and are a separate series from the personnel files burned
in 1973. The rolls are titled by date and roll number, not unit, so the work is
finding which roll covers the battalion.
TWO REFERENCE WORKS SET ASIDE, NEITHER A NEGATIVE. The ETO Order of Battle on the
Internet Archive is the Divisions volume - no signal construction battalion
appears in it at all, so his absence means nothing. A second ETO scan has
unusable OCR.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
GO 114 names his battalion in the Ardennes credit list, and two negatives
War Department General Orders No. 114, 7 December 1945, "Units Entitled to Battle
Credits", lists the 34th Signal Construction Battalion by name under ARDENNES -
an independent primary confirmation of one of the five campaigns on his
discharge. The order defines the combat zone (forward of Euskirchen-Eupen-Liege,
east bank of the Meuse, thence along the Franco-Belgian border and the southern
border of Luxembourg) and the window (16 Dec 1944 to 25 Jan 1945), and states the
credit is for personnel present for duty with a listed unit inside them.
Recorded as gaps: the discharge names the battalion but not his company (the
Separation Qualification Record and NPRC morning reports would); and his serial
is absent from NARA's enlistment database, which NARA itself says is incomplete.
Harold Crisci was researched at the owner's request as a family friend, not kin.
His enlistment record is reported to the owner and no person was created here.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Family-held originals: John Joseph Fitzgerald's war, and two draft cards
Andrew photographed original documents and they are transcribed onto the pages.
JOHN JOSEPH FITZGERALD (1913-1981), Andrew's grandfather, had an EMPTY summary -
a birth, a death, a marriage and nothing between. His discharge folder fills it:
- stock clerk of 542 Bergen Avenue, Jersey City; inducted 16 March 1943 aged 29,
five foot seven, 158 lbs, blue eyes; Local Board 21, Hudson County
- medical NCO, 34th Signal Construction Battalion; to the ETO 11 February 1944
- five campaigns - Normandy, Northern France, Rhineland, Ardennes, Central Europe
- not wounded
- then sent the other way: sailed for the Central Pacific 17 July 1945, arrived
23 August, nine days AFTER Japan surrendered, which is why a European record
carries the Asiatic Pacific Service Medal; home 2 December
- discharged a corporal at Fort Dix 14 December 1945 on an ASR score of 74;
two years, eight months, twenty-nine days
- the folder also holds Truman's printed letter of thanks
- and the certificate is stamped by the Hudson County Register on 16 MARCH 1981.
He died on the 14th. Someone took his papers in within two days.
HOWARD GRAHAM JONES, Gina's great-great-grandfather - 1942 draft card. Confirms
his birth to the day (18 April 1904), independently names Helen B. Jones as his
wife, and gives 15 Willow Lane, Great Neck with a telephone, plus his employer.
It also says his birthplace was PARIS, NEW YORK, against the New York City on his
page. Both kept; there is a real town of Paris in Oneida County, and if that is
it, the 1905 state census would be the first fixed point on the Jones line before
Great Neck.
HAROLD CRISCI, b. 27 May 1919 Paterson, mother Mrs Angelina Crisci - NOT recorded
anywhere. Crisci appears nowhere in this archive and no connection is known, so
nobody was created. Flagged in RESUME_HERE to ask the owner.
Also flagged: none of the four Fitzgerald documents has been scanned flat, and
the Separation Qualification Record is too blurred to read - it is the one that
describes his duties in his own words.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Antenati is down, and the free Italian route may not exist at all
Kept pushing after the Salvatore question closed. Two things:
The Ardi trail ends. "Stephen Ardi" and "Gloria Noest" return exactly one page in
the whole NYS corpus - the obituary already read - so Marie's mother's given name
and married surname are not recoverable there. And since Salvatore is no longer a
candidate son of James Gennaro, the Ardis are probably not Gina's family at all,
which makes them the wrong thing to chase.
Redirected to what IS her direct line: James Gennaro Spatarella, her
great-grandfather, born Frignano (Caserta) 17 May 1901. The line stops dead at
him - no father, no mother - and his Italian birth act would name both.
His page said the route was the free Portale Antenati. Checked: the portal is
DOWN site-wide, returning "Error establishing a Redis connection" on every URL,
and the legacy dl.antenati.san.beniculturali.it mirror is retired. Nothing could
be confirmed. Worse, catalogue descriptions indicate the Archivio di Stato di
Caserta's online holdings run 1808-1865, which would not reach 1901 - so the
claim on his page that "Frignano's records are on the Antenati portal, which is
free" may simply be wrong. Corrected to say so, flagged to verify when the site
returns, and the alternatives recorded: a written request to the Comune di
Frignano for an estratto dell'atto di nascita, or his full 1929 EDNY
naturalisation petition at NARA New York, which the index summarises but does not
reproduce.
Logged in BLOCKED_AND_PAID.md and promoted in RESUME_HERE as the biggest single
gap on Gina's side.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Not that Salvatore: the 1950 census answers it, and the answer is no
Pushed on the Spatarella question after the geography objection was withdrawn.
The push paid before it closed:
- Salvatore and Marie are established as husband and wife three ways (joint
Suffolk tax lot 1977; her mother's obituary naming her among the daughters; and
the 1965 item calling Gerald Noest his brother-in-law, which works only through
her). Marie's family name was ARDI.
- A man who was nothing but a shared grave now has an occupation, a town, two
firms, an association presidency and a documented life.
Then the record that answers the parentage question turned out to be one the
archive already held, uncited for this purpose: the 1950 census, Brooklyn ED
24-1062 sheet 73, read from the sheet rather than the index. The household is
complete - James 48, Amelia 37, Angelina 19, Esther 16, Bobby 7 - with the
closing rule drawn under line 29 and the HOUSEHOLD CONTINUED box unticked.
A son born 4 June 1939 would have been ten. He is not there. Frank at 21 and
Nicholas at 18 are also absent, which is what grown sons do and a boy does not.
And the newspapers' own ages (32 in 1965, 37 in 1973, implying 1933-36) put him
at 14-17, who would also be at home. On every reading of his age he is missing.
Salvatore A. Spatarella is not a son of James Gennaro Spatarella. That kills the
specific hypothesis and nothing more - a cousin or nephew is still possible and
equally unevidenced. He stays unattached, and his page says so.
The sequence is the lesson: the first exclusion was half-luck and half-wrong
reasoning; being told the geography was wrong forced a proper investigation; and
the proper investigation reached the same verdict for a completely different and
much better reason. The first no was luck. The second is evidence.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Salvatore and Marie were married; and a wrong exclusion withdrawn
Finding 137 refused to attach the Salvatore Spatarella of the Suffolk papers to
the Salvatore A. Spatarella in this archive. Refusing was right; half the
reasoning was wrong. One leg was the ages. The other was that he lived at Commack
"and this family's Spatarellas were in Brooklyn" - and the owner then said Gina's
father moved to Commack with his family as a boy. The place objection is
withdrawn and finding 137 corrected on its own page.
Pushing instead of excluding answered a question open since these two were
created. Their pages said a shared grave is strong for a marriage and no evidence
of anything else. They WERE husband and wife, three ways:
- Suffolk delinquent tax lists, Oct-Nov 1977: a lot held by "Spatarella Salvatore
& Marie", seven months after he died.
- Her mother's obituary, Smithtown News 13 Jan 1977: "two daughters, Marie
Spatarella of St. James and Gloria Noest of Kings Park; a son, Joseph of
Bayside; a brother, Stephen Ardi of Jackson Heights; three sisters, Carmella
Ardi" - which also gives Marie her family name, ARDI.
- The Long-Islander, 4 Nov 1965: "Sal Spatarella, 32, and his brother-in-law,
Gerald Noest" - who is his brother-in-law only through Marie, whose sister
Gloria was Mrs Noest. The 1965 item and the 1977 obituary explain each other.
The same chain identifies the man in the grave: secretary-treasurer of Gallagher
Sanitation and All American Refuse Removal (1973), president of the Suffolk
County Cartmen's Association (1970), of Greenfield Drive Commack; died 28 March
1977; and in August 1977 All American is reported as owned by Marie Spatarella of
St. James. Recorded factually with citations, including the July 1973 guilty
plea, because it is contemporaneous public record and is the only thing that
places this man anywhere.
STILL NOT ESTABLISHED and not assumed: that he belongs to Gina's family. A 1939
birth fits James Gennaro Spatarella's sibling gap exactly and he lived where
Gina's father grew up; but James Gennaro and Amelia are buried in Queens, and the
papers give ages implying 1933 and 1936. Three points, three birth years.
Cheapest test named and not yet done: Marie's 2018 obituary.
New standing rule in AGENTS.md: surname + place is not evidence of a STRANGER
either. Before ruling someone out on geography, ask the family where they lived.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
"the late Howard G. Jr." — thread 3 half unblocked from a Glen Cove obituary
Gina's side had had nothing all session. Finding 136 named the one free source
that could reach it, nyshistoricnewspapers.org, and recorded it as blocked to
automation by Cloudflare. Andrew opened it in a browser; it paid in minutes.
Record-Pilot (Glen Cove), 18 January 1996, page 30:
Howard G. Jones, of Sea Cliff formerly of Great Neck, died on Jan. 5, 1996.
Husband of Mary B. and the late Helen B. Father of G. David, Janet and the
late Howard G. Jr. Stepfather of six daughters. Grandfather of 21. ...
Funeral Service at All Saints Episcopal Church, Great Neck. Interment
Churchyard.
RESUME_HERE stated thread 3 as: "No death date is known and that is the entire
blocker." Four words change that. "the late Howard G. Jr." is the first
documentary evidence of his death the archive has ever held - he was dead on
family testimony alone - and it bounds it before 5 January 1996, at about
sixty-one. The date is still unknown but a bounded search is a searchable one.
Also from the same nine lines:
- exact death date for his father, 5 January 1996 (the SSDI gives no day);
- the first burial place on the whole branch: the churchyard at All Saints
Episcopal, Great Neck - a small ground likely to hold Helen too;
- a second marriage nobody knew of, to a "Mary B.", with six stepdaughters. No
person created for her: an initial is not a surname and she may be living;
- a death bound for Helen E. Brown, who had no death recorded at all;
- independent confirmation of the sibling group held on testimony alone.
Source notes: it works in a browser only. It is Veridian, not Open ONI, so the
API shape guessed in finding 136 was wrong and /search/pages/results/ 404s; the
working URL forms are recorded. Coverage measured - Long Island strong, Brooklyn
thin and late (Kings is 56 of 961 Cacace hits), which is why this reached the
Joneses of Nassau and will not reach the Brooklyn Cacace, Spatarella and Masino.
Deliberate refusal recorded: "Spatarella" is dominated by Salvatore "Garbage
King" Spatarella of Commack, prosecuted in the early 1970s. Rare surname, same
given name as a man in this archive, right region - and the ages exclude him (32
in 1965 and 37 in 1973 against a birth of 1939), as does Suffolk against this
family's Brooklyn. Nothing attached, and the exclusion is written down so the
next person finds it made rather than the temptation.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Fitzgerald, Farley and Jones swept: no stories, and the reason measured
With the search finally working, the other three lines were swept. None produced
a story. Measuring why is the durable result.
Coverage audited by faceting the collection and counting which papers actually
return hits. Hudson County's entire holding is The Jersey City News (1889-1906)
plus two 1880s Jersey City weeklies - nothing after 1906. And there is NO Long
Island newspaper in the corpus at all; every "Great Neck" hit comes from a
Manhattan daily.
- Fitzgerald: unreachable twice over. Married 1908, died 1939 - all past the 1906
wall - and the surname returns 8,049 NJ hits so nothing can be isolated.
- Farley: the one line the corpus could reach, since the Manhattan dailies cover
Andrew Joseph Farrelly's 1920 household at 264 West 155th Street. Still nothing:
that address returns no hits 1915-1930, and "Farley" in NY in 1921 returns 643
dominated by Cardinal Farley and James A. Farley. The specific miss is Helen
Bridget Farley, who lived two days in June 1921 and got no notice.
- Jones: a gap, not a negative. "Howard Jones" + "Great Neck" returns exactly one
page, a sports column about Howard Jones the Iowa football coach.
nyshistoricnewspapers.org - free, proper Nassau/Suffolk coverage, same Open ONI
API - is behind a Cloudflare challenge and 403s any script. Recorded as blocked
per AGENTS.md rather than bypassed. It is the only free newspaper route to Gina's
side, and BLOCKED_AND_PAID.md now lists the browser searches worth doing.
Each of the four people carries its own controlled negative with the coverage
reason stated, so nobody re-runs it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
news.py was lying twice: a dead state filter and unsearchable OCR
Went looking for newspaper stories about people already here. Found few, and
found out why - two independent silent failures, neither visible from the output.
1. --state did nothing. loc.gov accepts location_state=new jersey, returns 200,
and drops it: "Leahey" gives 3,513 hits unfiltered and 3,513 filtered, topped
by Washington and Delaware papers. The faceted form fa=location_state: gives
225, all New Jersey. Every geographic search this project ran was national, so
every geographic newspaper negative on record is worthless.
2. The OCR mangles the names, proved against a clipping this archive has held for
months. The Jersey City News of 15 June 1889 carries "Miss Catherine Meaney
were married by the Rev. Father Andrew M. Egan"; loc.gov's text of that page
reads "Catherine Meauev were niarried by the Rev. Father Andrew M. F.gan". A
search of the page for "Meaney" returns ZERO. The archive has the notice only
because someone transcribed it by eye from the image. This is almost certainly
the real cause of the old false zero where every spelling of Leahey returned
nothing against a true count of 135.
--variants now searches the substitutions this print makes (n/u, y/v, m/ni,
e/c, l/i, h/b). It finds the 1889 page under "meauey" and "meauev". A plain
Leahey sweep of NJ 1885-1908 gives 89 hits; variants reach 16 more pages.
3. ocr.txt is 403, and so is the resource JSON with a browser UA - the reverse of
the Find a Grave rule. --read now fetches via the project UA and prints only
the passages naming the person.
Found and recorded: a same-name trap (Patrolman Dennis Leahey breaking up a
Hoboken mayoral meeting in 1899 is not our Bayonne captain); a Jersey City Meaney
circle of 1889 attached to nobody because surname plus place is not evidence.
Two hits read and rejected.
Also documented: the Bayonne Times is not in Chronicling America at all, so a
silent negative for a Bayonne family usually means the paper is absent.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
news.py: --state was silently ignored, so every geographic search was national
loc.gov accepts location_state=new jersey, returns HTTP 200, and drops it.
Verified: "Leahey" unfiltered returns 3,513 hits; with location_state= it still
returns 3,513, topped by Washington, Delaware and South Dakota papers. The
faceted form fa=location_state:new jersey returns 225, all New Jersey.
Nothing about the failure was visible in the output. Every --state search this
project has run was in fact national, so any "not in New Jersey papers"
conclusion or count from before today is worthless. Same shape as the location=
county trap on irishgenealogy.ie: a parameter the site accepts and drops.
Also documented: ocr.txt returns 403; fetch the resource JSON with ?fo=json using
the project User-Agent (a browser UA gets 403 here, the reverse of the Find a
Grave rule) and follow resource.fulltext_file. And the Bayonne Times is not in
Chronicling America at all, so a silent negative for a Bayonne family usually
means the paper is absent, not the story.
Same-name trap recorded: a Patrolman Dennis Leahey policing Hoboken in 1899 is
not this archive's Bayonne boatman.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Everett Virgil Mead verified: LTC, Payne Iowa to Salt Lake City, and his parents named
He was entirely tree-supplied - no sources at all - and I had repeated the
lieutenant colonel rank on that basis. A Find a Grave memorial now corroborates
the dates to the day, sharpens the birthplace from 'Fremont, Iowa' to Payne
township, gives his burial at Mount Olive, Hamburg, and styles him LTC with a
veteran flag. Still compiled, not a service record, and his page says so.
His parents are named for the first time - Frank Mead 1877-1952 and Anna Mead
1880-1936 - and deliberately NOT created as people. The Meads join this family
only through the marriage of Everett's son to Andrew's first cousin once removed,
so building upward here extends the family of a cousin's spouse. The names are
recorded as a doorway rather than acted on.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Ramsey v. Leahey: the second report, the vote, and eleven citing cases
The archive recorded only the final affirmance, 103 N.J.L. 501. There are two
published reports and the other is the one that matters: 102 N.J.L. 513, 134 A.
91, the opinion of Parker, J., 14 July 1926 - which is also cited more often.
Parker's holding: Ramsey lived aboard the barge, was on call around the clock,
and had to buy his own food ashore with his employer's knowledge. That he stopped
for two or three hours to visit a friend did not put him outside his employment.
"He must have food; it was not brought to him, so he must needs go and get it."
The Court of Errors and Appeals affirmed per curiam and the report prints the
vote: eleven for affirmance, none for reversal, the Chancellor and Chief Justice
among them.
Eleven later decisions cite it, 1927-1957, and it left the state - the Court of
Appeals of Texas in 1929 and the Supreme Court of Iowa in 1948.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Two great-great-grandmothers from the Bayonne baptism register
Finding 134 answered "has any direct ancestor been pushed back?" with no. This
answers it with two. Andrew's direct ancestors: 37 -> 39, both additions
2x-great-grandmothers, and the generated pages say so themselves.
Targets were chosen by asking which direct ancestors were born inside the
register's 1884-1894 window and were missing a parent. Nine qualified; six had
both parents already; two did not, and both were christened at St. Mary's.
MARGARET MAHONEY - entry 77, Denis Leahey b. 3 Apr 1886, "John / Marg't Mahoney",
sponsors Denis Leahey and Anna Leahey. John Leahey (b.1861 d.1887) had no wife
recorded and his only child had no mother. She is the only Mahoney in the archive.
The sponsors identify the family: the child's grandfather Dennis, alive at Bayonne
in 1886, and his daughter Anna b.1874 - and the boy carries the grandfather's
name. Recorded `probable`, not documented, with both inferential steps stated: that
this is our John Leahey, and that the child Denis is the man the archive calls
James D. Leahey. If that holds, the D stands for Denis. His birth also sharpens
from "about 1886" to 3 April 1886.
CATHERINE LAVELLE - named as mother at Barbara's 1886 baptism and Catherine's in
1888. The archive's only Catherine Lavelle carries a tree birth of May 1894, which
is impossible for a mother of 1886, so the maternal links had been rightly
detached - leaving the evidence homeless. The baptismal mother now has her own
page, following the existing catherine-meaney-mother pattern. Neither page merged
or overwritten; crosscheck twins flags them as possibly one person, which is the
honest state and what both pages say.
CATHERINE BRENNAN b. 6 Oct 1888, baptised the next day - a daughter whose baptism
this archive has cited for months without ever giving her a page. One-day baptism
is this register's mark of a child not expected to live; whether she survived is
now an open question.
Recorded as exhausted for the direct lines: every direct ancestor born 1884-1894
is checked. Mary Cooney was born in Ireland; Gina's three were born in New York
and already have both parents. This corpus cannot reach Gina's family at all.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
The frontier of both direct lines, and one advance on Andrew's
Asked whether any direct ancestor had actually been pushed back this session, the
computed answer was no. Four people gained a parent link and all four are
collateral - two half-siblings of Andrew's grandfather's generation, a great-uncle,
and an infant cousin. Direct ancestors were touched only with detail: a literacy
state, a withdrawn autograph claim, a spouse, a burial plot. Gina's side received
nothing at all.
So the frontier of both lines is now mapped, because nobody had drawn it:
- Andrew: 37 direct ancestors, 16 at the frontier, the closest being his
great-grandmother Mary Cooney.
- Gina: 48 direct ancestors, 25 at the frontier, and hers stop closer in - three
great-grandparents with no parents recorded at all, all dead within living
memory. The softest targets in the archive, never yet worked.
The one advance: Mary Cooney's death was carried on the family tree's word alone,
sourced:false, from a Geni export - "29 August 1949, Jersey City Medical Center",
never tested. The New Jersey death index confirms month and state and gives
certificate 29315. A 1949 certificate names father and mother including the
mother's maiden name, so that number is a generation on Andrew's own line.
Confirmed: month and year. NOT confirmed: the day, which has no column in the
index, and identity, which has no age or address to test.
The lesson is the exact inverse of finding 129: a death index files a married
woman under the name she DIED with, a bride index under the name she was BORN
with. Searching 1949 under Cooney finds a different Mary Cooney who died that
January - right surname, right year, wrong woman, and it reads like a disproof.
Both conventions now head the njvitals section of TOOLING.md.
Lead not followed: the New York State Death Index is also on archive.org, 60+
volumes. It is the only route to Gina's side, and it excludes New York City -
recorded with that caveat rather than built in haste.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Caroline and Joanna Meaney, recovered from their husbands' names
This archive held three women as "Mrs. Louis Smith", "Mrs. Harry Smith" and
"Mrs. Paul Aldbeck" - no given names, no dates, suppressed as possibly living
because nothing documented a death. Two now have their names back.
St. Mary Star of the Sea, Bayonne, annotated its baptisms with the marriages that
followed decades later in the same church. So the fact that reduced these women
to their husbands' names in a death notice is the fact that returns them:
265 | Caroline Meaney | Oct 17 1887 | Thomas / Cath. Rohling | Married to
Louis Smith Jan 22 - 1911 in St. Mary's, A.M. Egan
10 | Joanna Meaney | Dec 28 1889 | Thomas / Cath. Rohlan | Married to
Henry Smith. Jan 14 - 1913 at St. Mary's. A.M. Egan
Harry was Henry. Both are now documented with birth, baptism and marriage.
Three free corroborations: the mother, held from a headstone as Catherine Roland,
is "Cath. Rohling" and "Cath. Rohlan" in the register - the clerk's two tries at
the name, from an independent source. A.M. Egan officiated at both weddings and
also married their sister to Matthew Leahy in 1889. And a William Smith stood
sponsor at Joanna's 1890 baptism, twenty-three years before she married a Smith -
suggestive, not evidence, and no Smith was created on the strength of it.
The third sister: "Aldbeck" occurs nowhere in the 274-register, 4,248-entry
corpus. Recorded with its limit stated - the register begins about 1884 and she
was likely born before it, so this proves little.
Also here:
- 257 of the 274 St. Mary registers have never been cited. Surname counts for the
un-mined ones are logged in RESUME_HERE; match on parents, not surname.
- irish.py sweep --event marriage is unusable: it filters by implied age and
ignores the spouse's name, so every person returns thousands of national hits
("and 2621 more"). A real gap in the tool, recorded rather than worked around.
- Display bug fixed: 53 person pages were showing literal asterisks. rich() drew
bold only and plain() stripped bold only, so single-asterisk emphasis leaked
into pages, meta descriptions and search.json. Both now handle it, using
Markdown's rule that emphasis cannot open or close on whitespace so "5 * 3 * 2"
stays arithmetic. 53 -> 12, and all twelve remaining are OCR garbage inside
source transcriptions, which must not be touched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Spread njvitals across the other lines: a 1908 marriage, and a row missing from the OCR
Catherine Condon and David Fitzgerald, New Jersey 1908, volume 9 page 563 - the
biggest New Jersey link on crosscheck's unverified list, with 14 people below it,
carried for months as "[identity unproved]" with no source. Now cited precisely
enough to order the certificate, which names both sets of parents and would put a
father and mother behind David Fitzgerald while testing the identification.
The way it was nearly missed matters more than the find. Searching that volume's
text for "Catherine" returns zero: the strings CATHERINE and FITZGERALD occur
nowhere in page 457's text layer, though the row is plainly legible on the scan.
The CONDON baselines run at a steady 90px pitch and then jump 284px - three
printed lines were never transcribed. Nothing looks wrong from inside the volume:
65,336 rows parsed, 21 Condon rows, alphabetical order intact either side.
Only the automatic control from finding 131 caught it. Without that check this
session would have written down "[NEGATIVE, CONTROLLED] no Catherine Condon
marries a Fitzgerald in NJ 1904-1909" on a link carrying fourteen people. The
control now names this failure mode and prints the page images to read by eye.
Tool changes:
- row clustering now measures each page's line pitch instead of assuming 25px.
The 1946-1948 volumes were welding three brides into one unmatchable string.
Regression-tested: the Barbara Brennan row is still found, rows parsed in that
volume went 22,380 -> 26,589.
- `sweep` added: reads people.json, finds everyone deceased, NJ-connected and
missing a death (65 people across 19 surnames), works out which volumes could
hold each, and reads each book once for all surnames rather than once per
person. Prints its plan and refuses to exceed --max-volumes.
TOOLING.md now leads with the headline caveat: this corpus produces candidates,
never negatives.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
tools/njvitals.py: 503 free volumes of New Jersey vital records, searchable
Finding 129 found Barbara Brennan's 1916 marriage by hand - a 20MB OCR download
and 401 pages reconstructed from word coordinates. Right once, absurd to repeat.
The corpus is bigger than that find suggested: 503 volumes of New Jersey birth,
marriage and death indexes on the Internet Archive, 1901 onward, with an OCR
layer and word coordinates. Free. Larger than irishgenealogy.ie for this family.
First afternoon's results:
- marian-murphy gets an exact death date where the archive had none: 22 May 1977,
certificate 23618, middle initial C matching Find a Grave's Marian Catherine.
- george-leahey and vera-leahey move from compiled to documented - the state
index confirms 04/22/83 and 01/65, and Julia's middle initial V confirms that
the family's "Vera" was a Veronica.
- James Francis Leahey is NOT in the 1983 death index under Leahey, Leahy or
Lahey, controlled by reading the page image: his brother George sits six lines
above where James would fall, in an unbroken run from LEAF to LEALE. Either he
died out of state or the memorial's date is wrong. Recorded, not resolved.
Four traps are built into the tool rather than left in a document:
- bride volumes file a woman under her maiden name (this cost months);
- whole periods were never released - no death index 1904-1919 or 1930-1948, no
grooms 1904-1929 or 1936-1948, and 1980 has no L-Q at all - so the tool
reports a gap instead of a negative, and `coverage` prints every hole;
- the OCR letter-spaces the scans, so all matching is on stripped text;
- given-name columns are unreadable in some volumes (1970 A-G renders Ernest as
"iRNF.ST"), so an empty full-name search automatically re-runs on the surname
alone and REFUSES the negative if the surname is present.
That last one enforces the project's oldest rule - control every negative - in
code rather than in memory. Four false zeros have damaged this project before.
SSNs printed in the later volumes are deliberately not recorded anywhere.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Searching Holy Cross by cemetery: two family graves, and Vera turns out to be Veronica
Finding 129 left the James-versus-William Leahey conflict for a census. It did
not need one. AGENTS.md has always said searching Find a Grave by cemetery
surfaces people you do not know exist; one search returned eleven Leahey burials
and the plot numbers did the work.
Blk-64 Sec-A Gr-718 holds Barbara Brennan with two sons - John Joseph in 2A and
James Francis in 2B, a son this archive did not hold at all. A mother and two
sons in one grave is a good deal more than a memorial's relationship label.
Blk-62 Sec-A Gr-1207 holds her son George Edward Leahey, his wife, an infant who
died in 1933 and a daughter.
The wife is the find. This archive carried a sourceless, dateless "Vera Leahey",
suppressed as possibly living, known only because the family death notices said
George "married a Vera". The cemetery calls her Julia Veronica Leahey and names
the same daughter. Vera is the short form of Veronica. No string comparison in
this project would ever have joined those two records.
New: james-francis-leahey-1911, infant-leahey-1933, william-j-spirko-jr.
Dated for the first time: george-leahey, catherine-leahey.
Unsuppressed on a documented (compiled) death: vera-leahey -> Julia Veronica
Leahey, barbara-leahey -> Barbara E. Leahey, married name Spirko.
Deliberately not done: william-leahey is not merged, renamed or deleted, though
no William Leahey is buried at Holy Cross and James fills his birth-order slot;
the death notice recording William's marriage to a Bender would decide it. No
person created for the unlinked "Baby Leahey" of 1968. Nothing upgraded past
compiled - not one civil record was fetched for any of these people.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Barbara Brennan's second husband is named: William Murphy, New Jersey 1916
RESUME_HERE listed this as the blocker on the Belmar house: the deed "may be in
her second husband's name — and that man has never been identified."
He is William Murphy. New Jersey Marriage Index, Brides, 1915-1919, A-B:
"BRENNAN BARBARA | MURPHY WILLIAM | 25547 | 16". Free, on the Internet Archive,
with page images.
It hid because the bride index files a woman under her MAIDEN name. Every search
had run under Leahey, the name she had used since 1904. She is indexed as
Brennan. All 401 pages of the volume were scanned: exactly one row names both
Brennan and Barbara, and the scan was controlled against a name known to be
there. The year fits the hole exactly - widowed 1914, married 1916, children
1916/17 and 1917/18 - and the son carries his father's name.
The deed search can now run under William Murphy and Barbara Murphy by name.
Find a Grave, all compiled, added her plot (Holy Cross, Blk-64 Sec-A Gr-718 1A),
death years for the children, and two things left deliberately unresolved: a son
James Francis Leahey 1911-1983 this archive does not hold, against a William
Leahey it holds with no sources at all. Both recorded, neither acted on.
Negatives logged so they are not re-run: there is no grooms index for 1915-1919
at all; the Bayonne Times death index fills its wife column only 7% of the time
and so excludes nobody; and the 1918 directory on disk returns zero Brennans for
a whole city, which is a broken control rather than a result.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
A page can contradict itself: new crosscheck finds two more withdrawn claims
Yesterday's Margaret Needham error was found by eye, which is not a method. The
new `contradictions` check compares a page's literacy field against every
sentence on that page and reports where they disagree. It found two more, both
invisible to every existing check because the data was right and only the prose
was wrong:
- margaret-needham open_questions[1] still asserted she signed her own name, in
the paragraph directly beneath the one withdrawing that claim, and asked for
the 1884 image to be re-read — which the paragraph above it had done.
- peter-coony-1860 repeated the withdrawn claim about his bride. Finding 125
corrected her page and never touched his.
A correction that lands on one page does not land on the pages that repeat it.
Controlled: inverting a sentence on mary-ryan-1843 makes the check fire, so the
zero after fixing is a real zero. Retraction paragraphs and two-people-one-
sentence cases are suppressed deliberately.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
Nobody signed these pages: every civil register image here is the registrar's copy
Set out to settle whether Patrick Farrelly could write. Read all four records
naming him: a cross in 1880, and no mark in 1879, 1881 or 1884 — each time on a
page where the same registrar recorded other people's crosses. He is now
`disputed`, a third literacy state added to the generator, rather than filed
under an answer the evidence does not support.
Reading those pages closely turned up something that touches the whole archive.
Every irishgenealogy.ie image is a certified copy the registrar made from his own
book — the clause is printed at the foot of every page, verified across five
districts and four counties, 1865-1894. On the Islandeady page of August 1894 ten
different women each made a mark and all ten crosses are identical.
The literacy findings survive; the mark annotation is copied faithfully, which is
why the same-page controls work. The handwriting framing does not. "In their own
hand" is now "How the register recorded their name", and all nine captions say
whose hand it is.
Also fixed:
- Margaret Needham's page claimed she wrote her own name, contradicting her own
literacy record and finding 125, which have said "mark" for months.
- The 1889 birth informant is Mary Farmer of Glentygrigny, present at the birth,
not the father. Read from the image for the first time; no person created.
- patrick-farrelly's 1880 mark cited src-455 (Bernard 1881) instead of src-080.
- patrick-farley carried the same 1889 birth twice.
- A junk /tmp/probe_1884.png entry removed from signatures.json.
People changed: patrick-farrelly, patrick-farley, ellen-farrelly, margaret-boyle,
john-fitzgerald, margaret-fraher, margaret-needham, thomas-garvey.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qt6T2mbD8MqkfD3Uzfyndf
RESUME_HERE rewritten for the 10-11 August state
The old one was from 5 August and said 532 people; there are 680. Rewritten
with the current numbers, the six open threads ranked by value, the Codex
sandbox fix and its caveat, the three new crosscheck checks, and the two
corrections I made to my own work this session.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
John Leahey left the marriage about 1947: family testimony, and the census agrees
Andrew records that his father's father left his father's mother around the
time his father was born in 1947. That converts a line I flagged yesterday as
unreadable into a documented one: the 1950 census enters John as Son, 42, in
his mother's household at Bayonne, with no wife and no three-year-old beside
him.
Recorded on both their pages as family testimony with its date and source,
alongside the one thing the paper does add -- Catherine died in 1976 still
indexed as Catherine V. Leahey, his surname, twenty-nine years later. No
divorce has been searched for and none is claimed.
Written plainly and without embroidery on either page.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
The 1950 census of Bayonne: Belmar explained, and two corrections to this morning's work
Andrew's father, born 1947, stayed with Barbara at Belmar in the summers
before school -- so about 1947-1952, which the 1950 census falls inside. It
finds her at BAYONNE, head of her own household. So Belmar was a summer
house, and a census records usual residence: that is why searching Belmar for
her found nothing. The route is closed by evidence rather than by failure,
and the answer is in Monmouth County property records.
The same record untangles a knot I got wrong this morning. Her household
holds four grown children: John Lahey 42 and Catherine Chambers 45 from the
first marriage, and William Murphy 32 and Marion Murphy 33 from the second.
I had removed William Murphy's parent link today, reasoning he had married
into the family and been mistaken for a son in a death notice. He is her son,
entered as Son, aged 32. The reasoning failed because it assumed one
marriage: Barbara was widowed in 1914 and married again, and William was born
about 1918. Restored, with the father recorded as the unidentified Murphy.
Marion Murphy was carried as a Leahey daughter under a married name, with
James D. Leahey as her father. She was born about 1917 and he died on 1 June
1914. She is William's sister, not his wife -- both children of the second
marriage. Detached from James.
And Catherine Leahey, who had no dates and no married name, is Catherine
Chambers.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Barbara Brennan and the Belmar house: family recollection recorded, searches logged
Andrew recalls his father's father's mother had a house at Belmar. The
archive contained zero mentions of Belmar, so this starts from nothing.
Recorded the negatives with their exact scope so they are not repeated: the
1950 census of Belmar has no Barbara of her age, no Leahey at all, and its
Murphy households are all younger and born elsewhere; the 1940 census of
Belmar and South Belmar yields one Barbara of the right age, Barbara Kraemer
born 1882 in New York, who is not her.
Also recorded why a census was unlikely to answer it: it records usual
residence, and a summer house used by a Bayonne family would not appear --
and if bought in her second husband's name it would be titled Murphy, a man
this archive still has not identified.
Named the routes that would answer it: Monmouth County deed index, her 1961
death record or obituary, Belmar tax rolls, and identifying the Murphy.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Eight Hickey births from the free Queensland index; three wrong dates and a duplicate child
A controlled sweep of Queensland births 1860-1882 returned eight children of
John Hickey and Ellen Condon, every one naming both parents: William 1861,
Mary 1862, Margaret 1864, Patrick 1868, Johannah 1870, Catherine 1873,
Bridget 1875, Thomas James 1877. All eight parent links upgraded from
compiled to documented, and every Hickey child has now dropped off the
unverified list.
Three corrections fell out of it.
Mary Ann Hickey carried THREE birthdays: 27 October 1862 in her date field,
6 October 1862 in her own summary two inches below it, and 6 November 1862 in
the registration. Nobody had noticed the page disagreed with itself. The
registration wins. The index also calls her simply Mary.
Joanna Hickey and Johannah Hickey were one girl who died at nine, entered
twice a letter apart. crosscheck twins never caught it -- the spellings
differ and the vaguer record had no dates to share. Merged on the strength of
birth 1870/C/1659 and death 1880/C/927.
John Hickey b.1866 has no birth registration at all, alone among his eight
siblings. His place is not in doubt -- his 1955 death names Ellen Condon as
mother -- but the 4 August 1866 date rests on compilation only.
Negatives recorded: the marriage registration leaves both parties' parents
blank, so that route to Ellen's mother is closed; and nine Queensland John
Condon deaths yield no identifiable father for her, since the index gives no
ages.
And one labelled guess: a John Condon died in Queensland in 1864, son of
Patrick Condon and Mary Cashell. Our phantom is a Mary Cashin, wife of a John
Condon, mother of a Patrick. Same three names rearranged. Filed as
speculation about where the error came from, claimed as nothing.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Patrick Farrelly: a mark in 1880, none in 1881, both read from the images
Read Bernard's 1881 birth register (entry 495) and went back to Andrew's 1880
one to check a claim I had written into the archive earlier today from an
index-derived quote.
Both quotes turn out to be right, and the difference is in the records, not
in the transcription. July 1880: 'Patrick x Farrelly, his mark, Father,
Clarah', with the clerk writing his above the cross and mark below, exactly
as for four other fathers on that page. September 1881: 'Patt Farrelly,
Father, Clara' -- no cross, no annotation, again on a page where four other
men clearly have one.
His summary previously said flatly that he could not write. That now reads
more carefully, and the open question declines to claim he learned: clerks
were inconsistent, and an absent annotation is weaker evidence than a present
one. Ellen 1884 and Patrick 1889 are cited and unread; they would settle it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Queensland: the free index names the parents, and a 77-descendant link is now documented
The archive assumed the Queensland registry gave nothing away without paying.
It was wrong. The historical index is free and prints parents, and two of the
largest unproven joins fell out of it in one pass. Both records verified
independently in the browser, not taken on the agent's word.
1906/C/2799, John Hickey d. 18 July 1906: father William Hickey, mother May
Grady. That is exactly the pair the archive carried as an unproven compiled
guess with 77 people hanging off it. Upgraded to documented; it has dropped
off the unverified list entirely. The registry writes May where the archive
holds Mary -- both now attested.
1882/C/2628, Ellen Hickey d. 24 August 1882: father John Condon, mother a
dash. First record of any kind to name Ellen's father, so a real gain -- but
it does not make him our John Condon, and her parent link stays compiled.
And a fourth silence on Mary Cashin: the one record that had every reason to
name her -- her supposed daughter's death registration, taken from someone
who knew the father's name -- leaves the mother blank.
BLOCKED_AND_PAID updated with the CSRF/POST recipe for the index, the fact
that detail pages 202 to curl but load in a browser, what the images actually
cost, and that Trove is bot-blocked rather than searched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Cornelia Brokaw's plot found: Elmwood, Section E, Lot 33 -- and what it could answer
Her Find a Grave memorial gives the exact plot, plus a middle name the
archive did not have: Cornelia STRYKER Brokaw, her mother's maiden surname,
which independently matches the FamilySearch tree. Dates match this archive's
to the day; age at death 98.
That plot is the lever on a question this project has carried for a while --
whether three generations of women share one ground. Alice (buried 22 January
1906) and Leah Adelaine (1 February 1889) are recorded only as 'New
Brunswick' with no cemetery named.
Find a Grave cannot answer it and now says so on all three pages: no memorial
for Leah Mayer, none for Alice White in New Jersey, and not one Mayer of any
kind in Elmwood. The Stillwells and Brokaws who are there belong to other
households. The interment card for Section E Lot 33 would settle it in one
request to the cemetery office.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
The Brokaws: a Revolutionary ancestor in Washington's general orders, and why one died in Ohio
Followed the Brokaw/Stryker line through Find a Grave and the FamilySearch
tree. Four sources added, all marked compiled -- these are memorial
biographies and a user-editable tree, not records -- and two people added:
Caleb Brokaw (1746-1814) and Jane Van Nostrand (1748-1823), a generation the
archive did not have.
Caleb Brokaw, farmer and mason of Weston, served as a private in Captain
Peter Dumont Vroom's Company, Second Regiment Somerset County Militia. In
February 1779, with the army at Middlebrook, a complaint reached the
Commander-in-Chief against a Colonel Craige for 'beating and otherwise
ill-treating Caleb Brokaw, an Inhabitant of this State'. Washington convened
a Court of Inquiry, was dissatisfied with its evasion, and issued a general
order on officers' treatment of civilians. The text is quoted from
Fitzpatrick and is independently checkable at the Library of Congress.
His will names seven enslaved people, one of them by name: Deydan, left to
his widow among the furniture. The inventory values a man, woman and child at
, boys of fourteen and ten, and a girl of eight. Recorded plainly,
because it is what the record says and because Deydan has a name.
Abraham C. Brokaw's Ohio death is explained: drafted into the War of 1812,
discharged at Jersey City, paid .94 and Bounty Land Warrant 95.043 for
forty acres in Ohio. He died there in 1826; his widow stayed in New Jersey
and lies in Old Somerville Cemetery, Row 8, Plot 318.
And the archive's earlier 'no Vroom connection' is wrong: a Brokaw served in
a Vroom's company in the 1770s and a Vroom married his great-granddaughter
eighty years later, in the same county.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Howard Graham Jones Jr.: grave searched hard, not found; Richie Jones unflagged
Richie Jones marked deceased on family testimony (Andrew Leahey, 9 August
2026), same provenance note as his father. Sandra Lynn Jones stays living.
Ruth S Mayer was already recorded deceased, died 1983.
Grave search for Howard Jr. exhausted the free routes and all are recorded
with their exact queries: Find a Grave has no memorial for him, for Ruth
under either surname, or even for his father who died at Sea Cliff in 1996 --
this family is not on that site at all. No obituary found by web search. And
the archive holds no burial anywhere on the Jones side, so there is no known
family plot to search; the only grave on the branch is August Mayer's, at
Long Island National Cemetery, and he is Ruth's father.
The blocker is the missing death year.
Also fixed three stale 'Florence Selig' references that survived finding
117's correction on Ruth's and Joan's summaries and in August Mayer's open
question -- their mother is Florence Rose Seelig.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Howard Graham Jones Jr.: Find a Grave searched, no memorial found
Two queries recorded so they are not re-run. Howard/Graham/Jones with no year
returns 21 men born 1892 through 1951 and none born 1934; Howard Jones born
1934 +/-3 returns neighbours in 1932, 1933, 1935, 1936, 1937 and no match.
Stated plainly on the page that this proves nothing about where he is buried
-- Find a Grave is volunteer-contributed. The untried source is the SSDI,
which gave his father's death to the month, and it needs a FamilySearch login
that had expired.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Howard Graham Jones Jr. is dead: living flag removed on family testimony
He was held as a living person, so his page showed only his name and
relatives -- which is why the archive has nothing on his grave. Andrew
Leahey, who married into this family, confirmed his death on 9 August 2026.
The flag is removed on that word and the page says so plainly: this is family
testimony, not a record. No death date, no place and no burial are recorded,
and nothing has been searched for him yet because the living flag meant he
had no sources at all.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
The Cullane Condons are Limerick-born: a lead tested and closed same day
Fetched the 1901 and 1911 censuses of Cullane to test whether Condons in the
Fitzgeralds' own townland could explain David Fitzgerald's marriage to
Catherine Condon. The household is exactly the one from the death registers
-- Patrick as head with his unmarried sisters Nora and Margaret farming with
him, house 9 in 1901, house 7 in 1911 -- but all three are recorded born
County Limerick, in both censuses, and so is the second Cullane Condon
household. Catherine's people are Cork.
A shared surname in one townland is not a connection. The lead is downgraded
the same day it was raised, before anything was built on it, and the original
question stands: nothing yet explains how a Cullane Fitzgerald met a
Flemingstown Condon. New Jersey remains the likeliest answer.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Patrick Condon's death: 497 candidates, two read, both excluded -- and Condons at Cullane
A national sweep gives 497 Patrick Condon deaths in Ireland 1878-1930, 20 in
Mitchelstown. Only two fit a man born about 1838-40, and both register images
now read and both fail: the 1912 Ballinoe death is a COOPER where ours was a
labourer, and the 1915 Cullane death is a BACHELOR where ours was married
with eight children. His death remains unfound.
The second is kept for the entry beneath it. Two Condons were living at
Cullane, Ballylanders -- the Fitzgeralds' own townland: Patrick, bachelor
farmer, died there October 1915 aged 75, and Margaret, spinster and farmer's
daughter, died there February 1916 aged 58, consecutive entries reported by
the same neighbour. David Fitzgerald of Cullane married Catherine Condon of
Flemingstown in Jersey City, and how that happened has never had an
explanation. A lead, not a finding -- the 1901/1911 census of Cullane would
test it.
Also recorded: the 1915 death is filed under 1916, the registration year, so
a search bounded at 1915 misses it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
One corner radius, not six: 27 relics of the old 7px system normalised
The dead :root block declared --radius: 7px; the live one says 3px. Left
behind were 27 hardcoded border-radius values -- 6px eleven times, 4px six,
8px five, 5px three, plus a stray 7px and 3px -- so corners were visibly
inconsistent from one component to the next.
All 27 now use var(--radius). Deliberate shapes kept as they were: six 50%
circles, one 99px pill, five squared-off zeroes. 13 declarations already used
the variable; now 40 do.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Twenty-one dead CSS variables removed; computed values provably unchanged
site.css had two :root blocks. The second, 3,300 lines down, silently
overrode twenty-one declarations in the first -- the entire palette, --body,
--max, --measure, --radius and --shadow. Anyone editing the top of the file
to change a colour or the typeface would have seen no effect whatsoever, and
the declared Arial body font has never been what the site uses; it is Georgia.
Removed the dead declarations, kept the eight the first block uniquely
provides (color-scheme, --mono, --space-1..6), and left a comment saying
which block is live. Verified by computing the final value of every :root
variable before and after: zero changed.
Found by a Codex review pass, which flagged the duplicate roots but framed it
as tidiness rather than as dead code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Ellen Condon is not in the register that should hold her
The Castlelyons baptisms to a father named John Condon, 1830-1855, have been
fully enumerated: eleven children across five mothers, and no Ellen among
them. The nearest is a Norry to the Cronin household -- different name,
different mother.
Her page could previously only say that no record tied her to the Cork
Condons. It can now say that the register which should hold her does not.
With the reported Sixmilebridge origin in County Clare, the case for treating
her as a Castlelyons Condon is weaker than it has ever been. 77 people
descend from the link.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Only one Condon-Murphy marriage at Castlelyons in forty years
Enumerated every Condon-Murphy marriage in County Cork 1815-1855. Exactly one
is at Castlelyons: the March 1836 entry, two years before the candidate
Patrick's baptism there. Every other is a different parish.
So the parish does not obviously hold two couples of these names, and the
couple behind the candidate baptism is not one of a crowd in their own
parish. That is the strongest thing this identification has earned.
The 1833 Margaret still needs a home -- most likely a couple married
elsewhere who moved in, the Castletownroche pair of 1828 being the shape of
it, or a wrong year in one of the two index entries. Both need images.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
The Castlelyons marriage exists -- and it contradicts my own table
John Condon married Margt Murphy at Castlelyons in March 1836, two years
before the candidate Patrick was baptised there. Index-only, no image. That
is the record this archive went looking for.
It also breaks finding 126's table, which grouped Margaret 1833, Patrick 1838
and James 1842 as one household. Margaret 1833 predates the marriage by three
years and cannot be their child. That grouping was my inference from matching
parent names, not something any record asserts. Withdrawn.
A third John Condon married a Margaret Murphy at Castletownroche in 1828 and
baptised an Ellen there in 1829 -- not our Ellen, but proof the pairing is not
distinctive in north Cork. Which is the same trap that produced Mary Cashin,
and I walked into a version of it while writing the finding about avoiding it.
Nothing reattached.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
No civil birth for Maurice Fitzgerald, and the only candidate is excluded
Having moved his birth from the tree's 1874 to about December 1870 on the
strength of the Ballylanders baptism, the obvious test was the civil register.
A national sweep returns 49 Maurice Fitzgerald births in Ireland for 1868-74,
overwhelmingly in Kerry. Exactly one is in this family's part of Limerick, at
Kilmallock -- and its image reads James Fitzgerald and Alice Casey of Clash,
Rockhill. Excluded.
Both directions recorded: it removes the last support for 1874, which now
rests on nothing, and it leaves the 1870 baptism standing alone, which
under-registration in rural Ireland makes unsurprising rather than damning.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Codex's sweep, verified and applied: a sister's burial, a baptism before birth, and a four-year error
Passed two research jobs to Codex and checked everything before acting.
Fourteen more pasted quotes. I had fixed the five Cooney IMAGE sources and
missed their five INDEX twins (src-145..153), plus nine Fitzgerald sources
(src-050..059) all carrying one block of family dates. Quotes cleared and
labelled as not-their-own rather than left looking like transcriptions.
May Frances McGreevy was carrying her sister's burial. The 'Buried May 1897'
event quotes Sarah Ellen McGreevy's cemetery row -- died of burns at eight --
and Sarah Ellen is separately in the archive. The row was cited because
'Buried with Jack HOARE' is the only thing linking these McGreevys to a
Hoare, which is the entire basis for May's married surname. A collateral clue
had been rendered as her own burial. Removed; the clue survives as a clue.
Andrew Farrelly's baptism is rejected, not found. The Kildallen entry
predates his civil birth by six days and names Mary Boyle where his mother is
Margaret. Demoted to a failed candidate. The baptism of the man this American
branch descends from is still unlocated.
Maurice Fitzgerald was baptised four years before he was born, so the birth
year was wrong. 1874 was tree-supplied and unsourced; the baptism index gives
19 December 1870. Corrected -- and it makes him the eldest child, ten months
after his parents' marriage at Kildorrery in February 1870, ahead of Richard.
Also: islands are 33/131, not the 31/124 I quoted -- my figure predated
today's merges. Codex had it right.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
'Gentry, Dingle' is Ventry; place tables synced; a wrong docstring corrected
Thomas Garvey's birthplace came from the family tree as 'Gentry,Dingle',
which is not a place, so he had no map pin. He married in 1891 as a grocer of
Ventry, kept the shop at Ventry in 1911 with eleven people in four rooms, and
died at Ventry in 1914. A capital V read as a G. Mapped to Ventry, with the
birth still flagged unsourced -- this corrects a garbled string, it does not
document where he was born.
Place drift 7 -> 0: the seven places missing from the gazetteer all had
coordinates in people.json, so no pins were ever lost; places.json was simply
stale. Synced.
And crosscheck's own docstring was wrong: it said the generator reads
people.json['places'] only. It reads both -- data['places'] builds the place
pages and map, places.json is a separate coordinate gazetteer. The direction
of a drift matters and the check now says so.
Unmapped places 6 -> 5.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
PLAN: NLI reproduction restriction, and not every register has the date map
The microfilm's second frame forbids reproduction without NLI permission, so
transcribe and cite the URL rather than publishing images. Castlelyons
marriages (04994/05) has no page-to-date map and badly faded early pages --
28 images, Jan 1830 to Nov 1880 -- so check for the map before committing to
a browse.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Read the 1838 Castlelyons baptism from the original register
The National Library of Ireland parish register images turn out to be
entirely usable, and PLAN.md was wrong to treat them as impractical. The
method is documented: parish search by ?q=, the embedded page-to-date-range
JSON map that gives the exact page without bisecting, #page/N/mode/1up, and
the reader's own zoom.
With it: page 32 of Castlelyons microfilm 04994/04, May 1838 --
13 | Patrick | John Condon | Margaret Murphy | John McCarthy, Catherine Meade
The register confirms the FamilySearch index to the letter, which is worth
saying after a day in which that collection produced two impossible dates.
It also gives the sponsors, which no index carries and which are the first
names attached to this household from anything but a civil form.
It does not prove this is our Patrick; the age objection stands. But the
comparison is no longer a documented Mary Cashin against a hypothetical
Margaret Murphy. It is a Mary Cashin who appears in one derivative index
quote and nowhere else, against a Margaret Murphy written into the parish
register in 1838.
Images cannot be downloaded -- Cloudflare bot challenge on iserver.nli.ie.
Recorded, not circumvented.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
PLAN: the parish register collection is index-only, and step five cannot be run on it
'Then open the register image. Not optional.' is impossible for
Ireland, Catholic Parish Registers 1740-1900 -- every entry reads 'Image
Unavailable'. Recorded, along with the consequence: two of seven Condon
baptisms pulled by permalink today are chronologically impossible and neither
can be proved without going to the National Library of Ireland.
Treat the collection as a finding aid, not as evidence.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Permalinks for the Condon baptisms, and two of the seven are impossible
src-972 was a search URL carrying 51 people. Each baptism now has its own
record permalink, and re-running the query found a seventh child the archive
had never recorded.
Two indexed dates cannot be true. Hannah is indexed 10 March 1870, thirty-two
days before her sister Ellen's confirmed birth; Mary is indexed 12 June 1878,
a month before her own civil birth of 11 July. Both repair against the
family's own practice of baptising within nought to two days of birth --
Hannah to 1872, Mary to 12 July -- but neither is proved. John of Rathcormack
does not repair at all and is not added as a person.
Correction to a correction: this morning I recorded Hannah's 1870 as a
transcription slip made in this archive. The permalink shows FamilySearch
indexes it as 1870. The archive copied it faithfully and I blamed the wrong
party.
And the methodological point: this collection has NO IMAGES. Every entry says
'Image Unavailable'. PLAN.md's step five -- 'then open the register image, not
optional' -- cannot be followed here at all.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Two Ryans written across from a stone this archive photographed; a burial classifier reverted
Charles M. Ryan and his son Michael C. Ryan both had their burial stated in
prose in their own summaries and recorded nowhere as a fact -- no burial
field, no sources, nothing structured -- while the monument naming all three
Ryans sits photographed on this site. crossinform had been calling this
'spouse has a documented cemetery', which understated it: not a lead, but
documentation filed on the wrong page.
I tried to teach crossinform to split 'named on the relative's stone' from
'lead only' and reverted it. The classifier matched a mother's given name
inside a funeral-book row, and matched a sentence stating plainly that the
*spouse* was the one interred. A check that asserts a burial wrongly is worse
than one that only points, because it invites the uncritical write-back this
file exists to catch. The reasoning is left in the docstring so the next
person does not retry it blind.
Also reverted: extending the burial lead from spouses to parents and children
took the section from 105 rows to 288 and added no usable leads. Everyone has
a relative buried somewhere.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Five duplicates merged, a son-in-law detached, and a hybrid record removed
The relaxed twins check found ten pairs; worked all of them.
Merged (maiden vs married, or a spelling apart):
mary-virginia-hanley -> mary-virginia-horan
mary-beatrice-1892 -> mary-beatrice-hickey-1892
michael-mcgreevey-1900 -> michael-mcgreevy-1900
may-hoare-1893 -> may-frances-mcgreevy-hoare-1893
mary-o-connor-1878 -> mary-condon-1882
The last is the interesting one. It was not a duplicate of a person so much
as of a hypothesis: it asserted the Flemingstown birth, the Condon-Ryan
parents and the O'Connor marriage all at once -- the exact join both other
pages say is unproved and must not be made. It also gave Patrick O'Connor two
wives with his three children split between them. Merged into the wife, whom
the 1920 census and marriage index actually document, with the parent link
deliberately not carried over. The open question stands: 38 in 1920 implies
1882, and a Flemingstown birth would make her 41.
William Murphy was recorded as a son of James D. Leahey while his own summary
said he married into the family. Parents removed. His wife is very probably
Marian Murphy, a Leahey daughter under her married name, which would make the
'[unnamed] Murphy' stub her as well -- flagged on all three, not merged, since
the evidence is two surnames in one obituary.
Two remaining pairs marked NEGATIVE: the Hickey sisters, and the Seelig/Selig
families that finding 117 deliberately separated.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Invert the Mary Cashin search: five John Condon wives at Castlelyons, none of them her
Every previous attempt searched for Mary Cashin and could only confirm or
fail. Searching instead for Condon baptisms with a father named John, in the
parish where Patrick married and lived, returns five John Condon households
between 1830 and 1855 -- wives Margaret Murphy, Margaret Cronin, Johanna
Cotter, Mary Moynahan and Margaret Joyce. She is not among them. A third
negative, and the first that names alternatives rather than finding nothing.
Exactly one Patrick Condon was baptised at Castlelyons in the window:
13 May 1838, to John Condon and Margaret Murphy. Recorded as a candidate with
its age problem stated (25 at marriage implies 1839-40, not 1838) and the
naming pattern that supports it (eldest daughter Margaret, sons John and
James after the two grandfathers). Nothing reattached; 145 people still hang
from an unproved link, but it now has a falsifiable alternative.
Also: Mary Cashin's sole appearance in the archive is inside a derivative
index quote, while the register image of that same marriage has no column
for the groom's mother at all.
Found on the way: a second impossible pair in src-972 (John 10 Dec 1871 vs
Hannah 8 Mar 1872, 89 days); Mary Condon 1855 detached from a father who
would have been fifteen; and crosscheck twins relaxed to compare across
differing surnames, which is how maiden-vs-married duplicates hid. 1 pair
became 10.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
An impossible baptism year in the archive's most load-bearing source
Codex, consulted on the project, pointed at crosscheck's weak-citations list.
src-972 is on it -- and src-972 is the FamilySearch parish-register query
PLAN.md calls 'the single most valuable correction made to this file',
documenting six Condon baptisms on which fifty-one people rest.
Reading its own detail: 'Hannah, 10 March 1870' and 'Ellen, 11 April 1870',
to the same mother, thirty-two days apart. Hannah's civil birth is 8 March
1872 and a baptism two days after birth is ordinary, so the parish year was a
slip. Her page always carried the correct civil date, so nothing downstream
was wrong -- but the error sat unnoticed in the source for months.
New check: crosscheck.py siblings, for children of one mother born less than
250 days apart (same-day skipped as twins). It reports 0 on current data and
would NOT have caught this one, because the bad date lived in source prose
rather than a birth field -- verified by injecting the date, which surfaces
it at 32 days. It guards the class of error, not this instance.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
PLAN: the index-vs-image tally, nine entries in
Three wrong dates and three other substantive errors out of nine checked.
Image-read count 18 -> 32; index-only 36 -> 26.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Catherine Griffis upgraded from candidate to identified; writeback rule in PLAN
Her page said the October 1917 records were candidates because 'neither
record names a husband or a parent'. The Dempsey funeral book holds two rows
from the same house at 16 West 21st Street -- hers, and her daughter's a year
and a day later, which names Timothy Lynch and Catherine Griffin as parents.
Her own row names her parents, John Griffith and Anna Shannon, who are
deliberately not created as people on one derivative naming.
PLAN.md now documents crosscheck writeback and the rule behind it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Anna Griffis had two dates and no page
Her birth and death dates were in the data with no summary, no timeline and
no sources -- the Dempsey funeral row that produced both was only ever in
finding 18. Written back, including the two conflicts the book contains
about her: Joseph Walsh or John Walsh, England or Wales. Neither resolved.
The 26 July 1869 birth is flagged as what it is: 1945 minus a recorded age
of 75y 8m 28d, not a birth record.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Maurice Fitzgerald: four rejected deaths written onto his page
He had a name, a trade and nothing else -- no summary, no sources, no open
questions -- while finding 23 held four death images read and excluded on his
behalf. The negative work is the useful part and it was invisible: anyone
looking at the page would have started the same four searches over.
Now cited as four rejected candidates with the reason each fails. The two
farmers name sons William and Patrick, not John; the other two are labourers,
one dying in the union workhouse.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
crosscheck: a 'writeback' check for readings that never reached the data
Findings transcribe records in a pipe-delimited house style. This takes the
longest field of every such transcription and reports the ones that appear
nowhere in people.json -- the failure mode that produced Margaret Needham's
phantom signature, John Fitzgerald's unrecorded literacy and five Cooney
sources carrying one pasted block of dates.
22 of 128 register quotes currently fail it. Some are false positives (a
field present under different wording, or a negative control cited on
purpose); the docstring says so. It is a prompt to look, not a defect list.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
John Fitzgerald could write, and two labourers on his page could too
src-054 read from the image: David Fitzgerald, 28 March 1875, Cullane,
Galbally district, Mitchelstown union, Co. Limerick. His father signed his
own name -- the only Irish-born man in the direct line known to be literate,
against Maurice Fitzgerald, Patrick Condon, Patrick Farrelly and Peter Coony,
who all made marks.
Finding 94 had already established his signature and it had never reached
his page. The same reading also weakens that finding's thesis: on the 1875
Galbally page two of the three signers were labourers. Amended there rather
than left standing.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Thomas Coony born the sixteenth, not the fifteenth; Derrycriff set complete
src-148 read from the image -- the last of the five Derrycriff birth
registers standing on the index alone. Third index date found wrong out of
eight entries now checked. Peter Coony's mark appears a fourth time.
Also flagged: Bridget 16 Feb 1886 and Thomas 16 Feb 1887, same day in
consecutive years, from two independently read images.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Helen Cooney merged into Ellen Cooney: one daughter, not two
The crosscheck flagged ellen-cooney-corcoran ~ helen-cooney-1895 (same
parents, same 1894 birth) the moment the register reading landed. The archive
already held the documented Ellen, with a husband and three children; Helen
was the family tree's unsupported '22 May 1895, Island Eddy'.
The merge test the archive itself named -- two small girls a year apart in
the 1901 census of Derrycreeve -- fails: there is one. Bio folded, redirect
added, relation lists cleaned. The privacy guard caught the unfolded bio.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Peter Coony was a landholder on his wedding day; Patrick Farrelly made his mark
Two more cases of a reading that lived in a finding and never reached the
data. Finding 71 had Peter Coony, his father and Thomas Needham all as
labourers; the marriage image gives Peter as Landholder, his father John as
Labourer, Thomas Needham as Landholder. The rise from labourer happened
between the generations, not within Peter's marriage as the archive had it.
Finding 74 recorded 'Patrick Farrelly his mark, Father' on his son Andrew's
1880 birth register; his page never said he was illiterate. It does now.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2
Margaret Needham never signed anything
Reading Ellen Cooney's 1894 birth register ('Margaret Cooney, her x mark,
Mother') contradicted the claim on Margaret Needham's page that she signed
her 1884 marriage register. Reopening the marriage image settles it: both
parties made a mark, in the same form as the couples above and below.
The awkward part is that finding 92 reached this conclusion months ago from
these same images and was never written back to people.json -- along with
all four Derrycriff birth transcriptions, whose source records were still
carrying one pasted block of dates belonging to none of them. Backfilled.
Also: 'Helen Cooney, 22 May 1895, Island Eddy' is Ellen Cooney, born
1 August 1894 at Derrycriff.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XZMAwrGnz7tmeNQ84B2ah2