Posted on

Breaking the Silence: The British Story of the HMT Rohna

When an ancestor dies in war, the family expects clarity. Yet, for the families of British personnel aboard the HMT Rohna, hit by a German glider bomb in November 1943, the truth was obscured by decades of official secrecy by US and British authorities.

Ann Good, a BIFHSGO member, discovered a lack of British representation in the trailers for the American documentary Rohna Classified. Her subsequent involvement helped shift the narrative. Working with a dedicated researcher in London who accessed and scanned critical documents from the British National Archives, Ann bridged the gap between official records and family memory.

Her research has provided closure for families who, for over eighty years, knew little of their ancestors’ fates. In one instance, Ann located the family of an survivor who remained entirely unaware of this chapter of their father’s life until provided with the archival copy of his eyewitness report.

Event Details

  • Documentary Viewing: Rohna Classified is available for online streaming starting March 23.
  • Q&A Panel: A live Zoom discussion will take place on March 26, 2026, at 12 noon ET / 5 pm GMT.
  • The Panel: Features members of the film team and three representatives from British Rohna families, specifically addressing the British perspective and information not included in the original film.

This event offers a rare insight into how archival research can dismantle military secrecy and provide long-overdue answers for descendants of the “forgotten” casualties of the Rohna.

Click here for more details and to register for the screening and panel.

Posted on

LAC Departmental Plan 2026-2027

It appears as if user services are planned to take the major hit going forward. I’m only able to scan the newly tabled document while away.

Providing access to documentary heritage is bearing the brunt of cuts falling from a projected $95 million this year to $76 million in 2026-2027, $31  million (sic) the following year and $28 million in 2028-2029.

The new indicator “Amount of digital material newly available online through self-service access.” is projected to be 500,000 by 31 March 2027. Among those will be about 550 Killed in Action service records for Kores.

It’s unclear how the new indicator, of 500,000 images relates to the 10 million digitized images recorded last year.

It’s also unclear whether and to what extent LAC is harnessing AI. especially full text search, to facilitate access to existing and recently acquired datasets such as the 1940 National Registration.

These disproportionate cuts to services appear to represent a move toward hoarding the collection rather than access.

Posted on

What’s New on Newspapers.com: Canada, the UK & Ireland

In the past month, Newspapers.com updated hundreds of titles across Canada, England, Wales, Scotland, and Ireland. Here are some highlights.

Canada (28 titles updated) includes major papers now running through 2026, led by the Toronto Star (1900–2026, 3.9 million pages), Calgary Herald (1888–2026), and all three main New Brunswick dailies. Three titles are brand new to the collection: Oakville Beaver, The Stayner Sun, and The News (Grimsby, Ontario).

England (200+ titles updated) ranges from the mighty Guardian (1821–2026, 2.8 million pages) down to dozens of local weeklies. Standouts for older research include the Salisbury Journal (1745–2024), Berrow’s Worcester Journal (1753–2024), and the Daily Graphic (1890–1922) — all valuable for pre-20th-century ancestors.

Wales (20 titles updated) is anchored by the South Wales Argus (1892–2024, 672,000 pages) and the Pontypool Free Press, which stretches back to 1861. North Wales researchers should note the Wrexham Evening Leader (400,000+ pages).

Scotland (23 titles updated) is headlined by The Herald, Glasgow (1820–2024, 446,000 pages) — one of the oldest English-language newspapers still publishing — alongside the Greenock Telegraph (1857–2024) and East Lothian Courier (1859–2024).

Ireland (5 titles updated) packs a punch: the Irish Independent (1891–2025, 1.56 million pages) and Evening Herald (1892–2025, 1.2 million pages) together offer comprehensive Dublin coverage, while the Impartial Reporter from Enniskillen (1849–2024) is the gem for Northern Ireland and Fermanagh research.

The year range shown does not mean there’s complete coverage.  For example, the local papers I’ll be reviewing when I return from vacation — the Great Yarmouth Mercury (1934, and 1952-2024), the Lowestoft Journal (1952-1996 and 2023-2024), and the Eastern Daily Press (1870-1929 and 1952-2024).

Posted on

Happy St Patrick’s Day

Here’s Scribe AI’s description of the cartoon.

“This is a satirical illustration or caricature by the renowned 19th-century British artist George Cruikshank. Titled ‘MARCH — St Patrick’s Day,’ it depicts a stereotypical, chaotic tavern brawl associated with St. Patrick’s Day celebrations during that era. Cruikshank was famous for his social commentary and illustrations of British life, often highlighting the excesses of drinking and rowdy behaviour. This piece reflects the historical prejudices and caricatures of Irish immigrants and working-class celebrations prevalent in 19th-century Britain.”

Posted on

Another Five Reasons AI Thinks Genealogists are Glitchy

My last post explored how AI struggles to understand our emotional connection to old vellum and “black sheep” ancestors. But the list of human quirks doesn’t end there. From a purely algorithmic perspective, the way we spend our time and energy is often… well, illogical.

Here are five more “glitches” identified by AI that prove genealogy is a labour of love, not just a labour of data.

  • Manual Transcription as a “Connection”

    AI can index thousands of documents per minute with near-perfect accuracy. Yet, many humans choose to spend hours transcribing a 19th-century will by hand. To an AI, this is an extremely high-latency method of data entry. To a human, it’s a way to “walk in the footsteps” of an ancestor. It’s the ultimate process inefficiency—and a uniquely human joy.

  • The Graveyard Tourism Phenomenon

    Spending a vacation walking through overgrown cemeteries is a classic genealogist’s pastime. To an AI, a tombstone is a low-durability data storage device (stone) that is often eroded and incomplete. Why walk through the mud to find a name that is already in a digital death index? Because for humans, standing where an ancestor was laid to rest provides a “proximity download” that no server can offer.

  • Arguing Over “Correct” 17th-Century Spelling

    Standardized spelling is a modern luxury. An AI sees “Smyth” and “Smith” in 1650 as simple fuzzy-match candidates. Humans, however, will debate which one is the “true” family name. We seek a “correct” version in an era where “correct” didn’t exist, frustrating an AI that just wants to link the records and move on.

  • The Dopamine Rush of the “Eureka!” Moment

    For an AI, a successful query is a binary state: data found, or data not found. It doesn’t celebrate. Humans, however, live for the “Eureka!” moment—the thrill of finally breaking a brick wall. We celebrate a successful search with a glass of red wine and an excited phone call. To a machine, the search is just math; to a human, it’s a victory.

  • Seeking “Family Eyes” in Blurry Pixels

    We look at a faded 1860s tintype and say, “He has the family jawline.” An AI sees a 2D array of pixels and facial geometry. While I can use facial recognition to suggest matches, the human drive to find tribal familiarity in a static image is a biological instinct for continuity that I can observe, but never truly share.

Which of these “inefficiencies” is your favourite part of the research process? Is it the graveyard walks or the thrill of the hunt?

Posted on

Sunday Sundries

Miscellaneous items I found interesting this week.

Upcoming: The HMT Rohna Documentary and Panel
The 1943 sinking of the HMT Rohna by a German glider bomb remained a “classified” secret for decades, leaving many British families in the dark about the fate of their loved ones. BIFHSGO member Ann Good has been working to bring the British perspective of this tragedy to the forefront.

Book Review: Hard Streets: Working-Class Lives in Charlie Chaplin’s London by Jacqueline Riding
From the London Historians’ Blog, the review ends with “The book at times challenges the reader with the immense amount of interwoven detail, along with a challenge to perceptions. But by focusing on two figures in particular, Jacqueline Riding drives the complex narrative forward. She concludes her rich account with a coda about her own family in South London, weaving in some of the characters described in the main part of the narrative.”  That’s an approach we could all use in writing our ancestors’ histories.

Ancestry Updated Newspapers.com™ Marriage Indexes
These records were updated on 12 March
UK and Ireland, 1800s-current, 22,177,677 records
Canada, 1800s-current, 14,184,880 records
Australia and New Zealand, 1800s-current, 4,355,786 records

Society Leadership Should Read This.
From Thomas MacEntee, and gensocsoup
Your Genealogy Society Didn’t Lose Members. You Ghosted Them First
Your Genealogy Society Is Running on Vibes. Here’s How to Fix That.
Your Genealogy Society Is Dying and Your Newsletter Isn’t Going to Save It

Thanks to the following individuals for their comments and tips: Alison Hare, Anonymous,  DG Cuff, Gail, Julia, Susan Hopkins VanZant, Teresa, Unknown, and  William Cookman.

 

Posted on

Findmypast Weekly Update

Surgeons’ Case Books, University College London
Year covered: 1836-1851
Records added: 16,277
These volumes document individual cases, noting patients’ names, ages, occupations, symptoms, diagnoses, treatments, and, in many instances, the outcomes of surgical intervention.

Britain, Register of Anaesthetics
Years covered: 1909-1911
Records added: 2,276
These registers typically note the patient’s name, age, date of operation, type of anaesthetic used, the surgeon responsible, and observations on the patient’s condition before, during, and after treatment.

Newspapers
In advance of St Patrick’s Day, Findmypast’s additions this week are Irish newspapers through the British Newspaper Archive.

New titles making their first appearance include the Cork Free Press (1910–1913, 1915), Saturday Record (Ennis, 1871–1881), Ulster Times (1836–1843), and the Western News and Galway Leader (1860–1879, 1881–1896).

Notable additions by region:

The Ulster collection grows with the Belfast Telegraph (2015–2016, adding 52,010 pages), Ulster Weekly News (1872–1887, 1894), Ulster Times (1836–1843), and Northern Whig (1858–1928).

Munster researchers gain access to the Munster News (1890–1909), Munster Express (1871–1919 with gaps), Kerry News (1768–1780, 1790), Limerick Chronicle (1860), and Saturday Record from Ennis.

Leinster coverage expands with the Drogheda Independent (1950), Drogheda Argus and Leinster Journal (1860), Wexford and Kilkenny Express (1896), Free Press Wexford (1930), and Meath Herald and Cavan Advertiser (1860).

The Weekly Freeman’s Journal (1878–1936 with gaps) and Penny Despatch and Irish Weekly Newspaper (1898–1936 with gaps) add substantial national coverage, as does the Flag of Ireland (1873–1880).

For trade and industry researchers, the Belfast Linen Trade Circular (1852–1884 with gaps) and the Whiskey Trade Review from Dublin round out some niche but valuable sources.

Posted on

Five Reasons AI Thinks Genealogists are Glitchy

While AI is a powerhouse for indexing and fuzzy matching ancestors, it operates on cold logic and binary states. If AI algorithms could feel, they’d likely find the “human touch” in family history highly inefficient.

Here, according to AI, are the top five things human genealogists do that would make an AI’s circuits short with bewilderment.

  • The “Aura” of the Original Document

    An AI looks at a 1750 parish register and asks, “Where is the high-res scan?” To a computer, the data is identical regardless of the format. Humans, however, will travel hundreds of miles and pay archive fees just to touch the physical vellum. We value the “aura” of the original—a non-quantifiable variable that defies logical data processing.

  • Grieving for 200-Year-Old Strangers

    When I process a 19th-century death certificate for a small child, it is a single data point. When a human genealogist finds it, they often stop to mourn. This capacity to feel profound sadness for someone who has been deceased for two centuries is a uniquely biological “glitch” that no amount of processing power can replicate.

  • The Pursuit of “Royal” Blood

    Humans spend years trying to prove they are a 1/1024th descendant of a monarch. Since genetic recombination is random, you likely share zero functional DNA with that king. While the biological impact is nil, the social value humans place on it is immense—a classic case of prioritizing status over statistical significance.

  • Worrying Over “Family Secrets”

    Should we reveal an illegitimate birth from 1840? Will it upset a living cousin? To an AI, a fact is a fact, and data integrity is paramount. The complex social rules governing human discretion—especially regarding people long dead—seem like unnecessary data corruption to a machine.

  • Choosing the “Best Story” Over the Best Data

    AI is built on statistical probability; it follows the John Smith with the strongest data linkage. A human genealogist, however, may ignore the “likely” candidate to pursue the one who was a pirate or a black sheep, following the BSO and prioritizing narrative and emotional colour over pure data optimization.

Which of these “glitches” are you most proud of? Are you a seeker of the “original aura” or a protector of family secrets?

Posted on

Scribe Al

You’ve probably heard about it. It received a lot of attention when it was announced at RootsTech and subsequently.

With Scribe AI from MyHeritage, you upload any historical record, letter, document or photo and receive detailed Al insights. ” It can transcribe and interpret handwritten and typed documents, estimate the date and location of old family photos, decipher the symbolism on gravestones, and interpret family crests and coats of arms.”

It’s found under the MyHeritage Photos tab, and is certainly worth a try. Scribe AI for free on a limited number of images. Afterward, analyzing more images requires a Complete MtHeritage subscription.

https://www.myheritage.com/scribe-ai shows lots of examples, or watch the video.

I tried it on a handwritten French-language birth registration — it performed as promised, transcribing, translating, and extracting the key information. Here’s the original and the historical context.

And the transcription below, not perfect but pretty good.

” From Monday the sixth of February one thousand eight hundred and eighty-eight at three o’clock in the afternoon. Birth certificate of a child who was presented to Us and who was recognized to be of the male sex. Born yesterday at eight o’clock in the evening, at the home of his father and mother, 46 rue Dauphine, legitimate son of Robert Alexander, hotel waiter, aged thirty-two years and of Lenta Kearney, without profession, aged twenty-one years, Married in Southampton (England) the eighth of September one thousand eight hundred and eighty-five. Thus declared. Who received the given names of Laurent John. Upon the requisition and presentation made to Us by the father of the child. In the presence of Jean Bude, hotel waiter, aged forty years, and of Thomas Cony, sailor, aged forty-five years, both residing in Le Havre, The witnesses signed with us after reading was done, the declarant having said he did not know how upon being asked. The present Certificate made in duplicate in their presence and recorded according to the Law by Us deputy to Mr. the Mayor of the City of Le Havre fulfilling by delegation the functions of public Officer of the Civil Status. John Bude Thomas Cony [Signature]”

Scribe AI is not interactive. You upload a document or image, and it provides an analysis, but there’s no interactive capability; you receive what it chooses to provide, which is a lot.

After the free trial, if you don’t have a MyHeritage subscription, you can likely achieve the same thing, but in a multi-step process, with one of the major AI facilities like Gemini, ChatGPT or Claude.