Ep. 36: The Highland Line - Tracing Scottish Ancestors with AI

If you have Scottish Highland ancestry and your family tree hits a wall before 1855, this episode was made for you.

In Episode 36 of Ancestors and Algorithms, Brian traces a MacLeod family from the Isle of Skye using three AI tools and a research workflow that works for any Scottish Highland, Hebridean, or Inverness-shire ancestry. Whether your ancestors were MacLeods, MacDonalds, Morrisons, Campbells, Camerons, or any of the great Gaelic families of the northern parishes, the techniques in this episode apply directly to your research.

What you will learn:

• How to use Perplexity to orient yourself to an unfamiliar Scottish archive system before you search a single record, including a four-part orientation prompt that maps ScotlandsPeople, the Disruption of 1843, Gaelic naming conventions, and the historical context of the Highland Clearances in one session
• How to use Gemini via AI Studio (free) to transcribe 19th-century Old Parish Register (OPR) handwriting with expert-level accuracy, including the exact prompt structure that handles Scottish ecclesiastical abbreviations correctly
• How to use Claude to correlate evidence across multiple documents simultaneously, resolve census age conflicts, analyze the Highland naming tradition, and surface the Free Church records gap that explains why so many Highland families vanish from the OPR after 1843
• Why the Great Disruption of 1843 is the single most important historical event in Scottish Highland genealogy research, and how to find your ancestors in the Free Church records at the National Records of Scotland when the OPR goes silent
• What the 1841 Scotland census's age rounding convention means for your research and how to use it to resolve apparent conflicts between census records
• How the Highland naming tradition works as genealogical evidence, including its limits, and how to use it correctly without overstating what it proves

Records and archives referenced in this episode:

• ScotlandsPeople (scotlandspeople.gov.uk): Old Parish Registers, Scotland census 1841 to 1921, Highland and Island Emigration Society records 1852 to 1857
• National Records of Scotland: Free Church records, reference CH16, 1843 to 1977
• Trove (trove.nla.gov.au): For Australian listeners researching Scottish emigrant families in colonial newspapers
• Highland Archive Centre, Inverness (highlandarchives.org.uk)

AI tools demonstrated:

• Perplexity (perplexity.ai): Free tier
• Gemini via AI Studio (aistudio.google.com): Free, used for OPR handwriting transcription
• Claude (claude.ai): Free tier, used for multi-document correlation and evidence analysis

For Australian and UK listeners:

Australian researchers will find the Highland and Island Emigration Society records on ScotlandsPeople an essential starting point. Trove at trove.nla.gov.au holds Scottish emigrant community newspapers from Victoria and New South Wales that name home parishes in Scotland. UK researchers: the National Records of Scotland in Edinburgh holds the Free Church collection that covers the records gap created by the 1843 Disruption.

Connect with Ancestors and Algorithms:

📧 Email: ancestorsandai@gmail.com
🌐 Website: https://ancestorsandai.com/
📘 Facebook Group: Ancestors and Algorithms: AI for Genealogy - www.facebook.com/groups/ancestorsandalgorithms/

Golden Rule Reminder: AI is your research assistant, not your researcher.

Join our Facebook group to share your AI genealogy breakthroughs, ask questions, and connect with fellow family historians who are embracing the future of genealogy research!

New episodes every Tuesday. Subscribe so you never miss the latest AI tools and techniques for family history research.

Transcript

SPEAKER_00 0:11

There's a woman in an eighteen fifty-one census. Her name is written as Catherine McLeod Gillis. She's thirty eight years old. She's listed as born on the Isle of Skye, and she's living in Inverness, hundreds of miles from where she started, married to a blacksmith. The Isle of Sky in the mid 1800s was not a quiet place to be from. It was a place of clearances and evictions, of entire communities pulled off land their families had worked for generations, of a church that split in two and took its records with it, and of a Gaelic naming tradition so layered and complex that it can either hide your ancestors from you or hand them right to you, depending on whether you know what you're looking at. I went looking for Catherine's parents. Three AI tools later, I had a working hypothesis grounded in good evidence, a clear explanation for why the direct proof is missing, and a very specific list of exactly what to search next. That's what today is about. Let's dive in. Welcome to Ancestors and Algorithms, where family history meets artificial intelligence. I'm your host Brian, and today we are heading to the Scottish Highlands. If you have Scottish ancestry, especially Highland or Island ancestry, this episode is going to feel like someone finally handed you a map for a territory you've been wandering in without one. We're going to walk through the exact AI workflow I use for Scottish research. From orienting yourself to an unfamiliar archive system, to transcribing the handwriting in those old church registers, to correlating evidence across multiple documents when the direct record you need isn't there. And if you don't have a drop of Scottish blood, don't go anywhere. Because the techniques we're covering today apply to every research tradition you'll ever encounter. The orientation prompt, the cross document correlation, the honest assessment of what a partial answer actually proves. These are universal tools. So settle in. Our subject is Katrina McLeod. That's the Scottish Gaelic form of Catherine McLeod, and it's the name her family would have used at home, even if the Church of Scotland's session clerk wrote Catherine in the register. Katrina is a composite case built from the kinds of challenges that come up again and again in Highland Scottish research. And if you've got McLeods or McDonald's or Morrisons or Cameron in your family tree, I promise some part of today's story is going to feel very familiar. We start in the eighteen fifty one Scotland Census, which is on Scotland's people, Scotland'speople.gov.uk That's Scotland's official genealogy website, and I want to pause for a second to make sure everyone knows it exists. Census records from eighteen forty one through nineteen twenty one, church registers going back to the fifteen hundreds, wills, valuation roles, immigration records. It's the single most important resource for Scottish research bar none. In that eighteen fifty one census, we find Catherine McLeod Gillis. She's thirty eight years old, born on the Isle of Skye. Her husband is Donald Gillis, a blacksmith. They're living in Inverness, and three children are listed with them Murdo Gillis, age nine, Mary Gillis, age seven, and young Katrina Gillis, age three. Hold those children's names Murdo, Mary, we'll come back to them. Catherine was born around eighteen thirteen, and the question I set out to answer is who were her parents? Now before we go one step further, I want to give you some historical context for the world Katrina was born into, because the records of early nineteenth century highlands do not exist in a vacuum. They exist in a very specific historical moment, and understanding that moment changes how you interpret what you find and what you don't find. The Isle of Skye sits in the county of Inverness, off the northwest coast of Scotland. In eighteen thirteen it was almost entirely Gaelic speaking. The Church of Scotland held services in English, some schooling existed and it was in English. But at home, in the fields, in the fishing boats, people spoke Gaelic. They told their stories in Gaelic. They understood their family relationships through a naming tradition that was centuries old and deeply tied to who their people were. The early nineteenth century was also the period of the highland clearances, the large scale evictions of crofting families to make way for sheep farming. On Skye specifically, multiple landlords served over seventeen hundred writs of removal, affecting tens of thousands of people in the four decades following eighteen forty. Some families moved to coastal townships on the same island. Some immigrated to Canada or Australia. Some moved to Inverness or Glasgow or Edinburgh for factory work and trades. A blacksmith's wife born on Sky who turns up in Inverness by eighteen fifty one fits this historical pattern exactly. And here's something that the clearances created for genealogists, something that AI can help you navigate. A diaspora that scattered people from the same parish across multiple continents, left behind place names and family connections embedded in naming traditions rather than written records, and produced a documentary record that stops in one archive and picks up in another. Understanding the historical forces is not only interesting background, it's the research strategy. In Kilmuir specifically, the northern tip of Skye where our Katrina's family lived, the land was largely worked by crofting tenants who owed their tenure to Lord MacDonald's estate. A crofter wasn't a freeholder. He could not sell his land, he could not improve it without permission, and could be removed if the landlord chose to use the land differently. This precarious position shaped everything, where families stayed, when and why they left, and what records were kept about them when they moved. And then there's the church. Before eighteen fifty five, Scotland had no civil registration. No government mandated recording of births, marriages, and deaths. That didn't begin until the first of january eighteen fifty five. Before that date, the Church of Scotland kept what are called old parish registers, or OPRs. These are the records of baptisms, marriage proclamations, and burials kept parish by parish by a session clerk whose diligence honestly varied enormously from one place and one decade to another. The OPRs for highland parishes like Kilmere on the northern tip of Skye can be extraordinary. Or they can have significant gaps. And in the eighteen forties, something happened to the Church of Scotland that created a genealogical problem that researchers still bump into today. I'll explain that in a few minutes. But first, let me show you how we started the search. Because remember what we always say around here. AI is your research assistant, not your researcher. What AI can do is help you understand the landscape of a research problem before you start digging. It can map the territory. It can surface the historical forces that explain the gaps. That context saves hours. Sometimes it saves days. And that's where we went first. The first tool I opened was Perplexity at perplexity.ai. And I want to explain why because the choice of tool is intentional. Perplexity is my go-to when I need to orient myself to unfamiliar research territory quickly. It searches the live web, it cites its sources, and it can synthesize context from genealogy guides, archive documentation, and historical resources in a way that would take me an hour to assemble manually. The free version is completely capable for this kind of orientation research. I also had the Comet browser open on my laptop, which lets me research right alongside any Scotsland people page I'm viewing. Comet is free as of twenty twenty five, available for desktop and mobile and worth knowing about. Here's the prompt I gave perplexity. Read it carefully because the construction of this prompt is part of the lesson. Prompt number one. I am researching Scottish Highland ancestry. My ancestor Catherine McLeod appears in the 1851 Scotland census in Inverness age thirty eight, born Isle of Sky. That places her birth around eighteen thirteen in Sky. What are the most important genealogical records available for researching someone born on the Isle of Sky circa 1813? Please cover the following. What is available on Scotland's people, the significance of the disruption of 1843 for Highland Genealogy Research, Scottish Gaelic naming conventions I should understand for families in this era, and any historical context about the Isle of Sky in the early 19th century that might help me understand what records would or would not exist, end quote. Notice what I did there? I gave perplexity a specific person and a specific research problem, a place, a date, and four distinct areas of inquiry. A vague prompt gives you a vague answer. A specific prompt gives you a specific one. What came back was genuinely a masterclass in Scottish genealogy. Let me walk you through the four things Perplexity surfaced that shaped everything that followed. First, Scotland's people. Perplexity confirmed the site holds the full run of OPR records for every Scottish parish, indexed from the 1500s through 1854, all ten Scotland census records from 1841 to 1921, statutory registers of births, marriages, and deaths from 1855 forward, wills and testaments, valuation roles showing property occupiers, and something specific to the Highland experience that I was not expected to see highlighted, the records of the Highland and Island Immigration Society, which operated from eighteen fifty two to eighteen fifty seven and assisted thousands of residents of Skye and the Western Isles to emigrate to Australia. If your Scottish ancestors ended up in Victoria or New South Wales, this collection may have their names. I'll come back to it in the outro. Second, naming conventions. Perplexity confirmed a tradition that I'd read about but hadn't fully internalized. In Scottish Gaelic Highland families, there was a powerful tradition of naming children to honor grandparents on both sides of the family. The first son and first daughter typically carried grandparental names. Which side of the family went first varied by household. Some families honored the father's side first, others the mother's side first, and many mixed the pattern depending on who was living, who was closest, and sometimes who would help pay for the wedding. There was no universal rule about the order. What the tradition does establish reliably is that grandparental names were recycled across generations. Go back to Catherine's children. Her oldest son Murdo, her oldest daughter Mary. By the Highland Convention, those names almost certainly honor grandparents. The question is, whose grandparents? Donald's or Catherine's? That's what we still need to determine. And we don't know until we find out what Donald Gillis' parents were called. That's a key nuance, and I want to be honest about it. The naming tradition is a clue, not proof on its own. We'll come back to this. Third, and this is the one that changed the shape of the entire investigation, the disruption of eighteen forty three. In May of that year, approximately four hundred and seventy four ministers walked out of the Church of Scotland's General Assembly in Edinburgh and founded the Free Church of Scotland. This was called the Great Disruption and it reshaped Scottish religious life permanently. And in the Gaelic speaking highlands and islands, the Free Church took hold with extraordinary force. Sky, Lewis, the Western Isles, most of the Northwest mainland, these communities followed their ministers into the new church in huge numbers. Here's what this means for genealogy. The Free Church didn't keep OPRs. Their records are a completely separate collection held at the National Records of Scotland, and they are not indexed in the same way as the Church of Scotland registers on Scotland's people. If your Highland ancestors switched to the Free Church after eighteen forty three, any events from that year until civil registration began in eighteen fifty five may not be in the OPR. They are somewhere else, or they were not recorded at all during the upheaval of a congregation rebuilding itself without a church building, without a stipend for the minister, and sometimes without anywhere to meet. For Catherine, born around eighteen thirteen, the disruption doesn't affect her own baptism, but it absolutely shapes what happened to her family's records in the years surrounding eighteen fifty five, and it shapes the absence of records we'll find later in this episode. Fourth, perplexity explained the Gaelic form of the McLeod surname. In Scottish Gaelic, McLeod is written McLeod. The female form used for daughters is Nicolade, which is a contraction meaning daughter the son of Laud. So the same woman might appear as Katrina McLeod in Gaelic context, Katherine McLeod in an English language census, Kathneen McLeod or Kath McLeod in an OPR where the session clerk was abbreviating heavily, or even as Katherine M. Laud in some nineteenth century records where the apostrophe replaces the AC. Every one of those is the same person. That's why the orientation step matters. With that context, I went into Scotland's people and started looking. The eighteen forty one Scotland Census is on Scotland's people, and you can search the index for free. You pay a small fee of roughly two US dollars to view the original image, but the index will tell you whether the name appears and in which parish. That matters because you want to confirm you're looking in the right place before spending your research credits. I search for Catherine McLeod in Kilmure. That is the northernmost parish on Skye, occupying the tip of Trotternish Peninsula. McLeod families had deep roots in the northern parishes of Skye and it felt like the right place to begin. She was there. In the township of Kilmalog, within Kilmere Parish, Murdoch McLeod, age fifty five, Crofter, Mary McLeod, age fifty, and Catherine McLeod, age twenty five. Now I need to tell you something important about that age of twenty five. In the eighteen forty one Scotland census, the standard instruction was that adults over fifteen should have their age recorded as the nearest multiple of five below their actual age. In practice, many people followed this exactly. Some gave their precise age and some fell somewhere in between. The pattern to watch for any age that is a multiple of five, like twenty five, fifty, or fifty five could represent someone up to four years older than listed. So Catherine at twenty five might actually be twenty five, twenty six, twenty seven, twenty eight, or twenty nine. That places her birth anywhere from roughly eighteen twelve to eighteen sixteen. The eighteen fifty one census says she's thirty-eight, putting her birth around eighteen thirteen. These line up. This is GPS element four happening in real time. I found an apparent conflict between two sources, and understanding the census convention resolves it as a non conflict. But here's what matters most. In eighteen forty one, Catherine McLeod is living in Kilmere Parish with a man named Murdoch McLeod, age fifty five, and a woman named Mary McLeod, age fifty. The age relationships fit for a parent child connection well. Murdoch, born around seventeen eighty six, Mary around seventeen ninety one, marrying in eighteen ten, and having a daughter in eighteen thirteen is perfectly plausible. And those names Murdoch, Mary, go back to Catherine's children in eighteen fifty one, Murdo and Mary. This is where I needed to verify something before going further. I needed to know what Donald Gillis' parents were named. Because if his parents were also named Murdo and Mary, the naming tradition evidence points equally to both sides of the family and proves nothing. But if his parents had completely different names, the Murdo and Mary and Catherine's children become significant evidence. A quick search on Scotland's people for Donald Gillis and Straith Parish, which is in the southeast of Skye and a historically common area for the Gillis family, turned up a household in an early return that identified his parents as Duncan Gillis and Effie Gillis, not Murdo, not Mary. That matters. If Donald's parents were Duncan and Effie, then a first son named Murdo and a first daughter named Mary can't be explained by Donald's side of the family at all. By the Highland tradition of honoring grandparents, those names almost certainly point toward Catherine's parents. The naming evidence now carries real weight. But I still need a documentary evidence connecting Catherine to the Kilmure household. For that, I needed the OPR. This is where Gemini via AI Studio comes in. The Kilmure OPR marriage proclamation section, like most Thailand OPRs, is handwritten in dense, faded script that varies based on which session clerk was working in which year. The earlier entries from the first two decades of the nineteenth century tend to use abbreviations that are not always familiar to modern eyes. This is exactly the job Gemini 3 via AI Studio was built for. As of April 2026, Gemini 3 achieves a character air rate of around 1.67% on eighteenth and nineteenth century English in Scots handwriting. That is expert level accuracy. The critical detail is to access it through Google AI Studio at aiStudio.google.com, not through the regular Gemini app or the Gemini website. The AI Studio gives you direct access to the full model capability without additional application level guardrails that reduce transcription quality and other interfaces. And it's completely free. When you upload your document image, use thinking mode for difficult records. It takes a few extra seconds, but the accuracy on faded or cramped handwriting improves noticeably. Here's the prompt I used when I uploaded the OPR page image I had downloaded from Scotland's people. Prompt number two. Quote I am transcribing a page from the old parish register of the parish of Kilmure, Inverness Shire. Isle of Sky, Scotland. This is a Church of Scotland marriage proclamation register approximately 1808 to 1820. Please transcribe this image exactly as written, preserving all abbreviations, original spelling, and punctuation. If any text is unclear or illegible, write the word illegible in square brackets. Where you can confidently identify a standard Scottish ecclesiastical abbreviation, expand it in square brackets immediately following the abbreviation. Do not guess at unclear surnames, end quote. Two things to notice about that prompt. First, I gave Jim and I context, the document type, the geographic and temporal setting, the specific record format. An AI with no context might misread an archaic abbreviation as something modern. With context, it understands it's working with a Church of Scotland proclamation record from a specific era and region. Second, I gave it an explicit instruction for uncertain text, write illegible in brackets. That single instruction protects me from fabricated readings of smudged or damaged words, which is where AI transcription most commonly goes wrong. Jim and I came back with a clean transcription of the full page. And about two thirds of the way down, here's what it found. Murdoch McLeod and Mary Morrison, both in this parish, were proclaimed in order to marriage the ninth and sixteenth and twenty third days of december eighteen ten. James McLeod witness. There it is. Murdoch McLeod and Mary Morrison proclaimed for marriage in december eighteen ten in Kilmere Parish. A proclamation in this context means the bans were read on three consecutive Sundays, which was the Church of Scotland's requirement for all marriages. The actual wedding would have occurred shortly after that third proclamation. Murdoch McLeod and Mary Morrison married in Kilmure around late december eighteen ten or early january eighteen eleven. Catherine McLeod was born around eighteen thirteen. The chronology fits. I also want you to notice something about what Jim and I transcribe beyond the names. The entry includes the word both in this parish. That phrase matters. It tells me these two people were both residents of Kilmure at the time of the proclamation. They weren't from different parishes, which would have required a different administrative process. Both Murdoch and Mary were established Kilmore residents before they married. And the witness, James McLeod, was likely a close family member. That's the kind of contextual detail a transcription unlocks that an index alone never could. Jim and I also correctly flagged two abbreviations J A M E S for the abbreviated first name and W I T N E S for the word witness. I would not have been confident in those expansions myself from the image alone. Having them identified explicitly in the transcription, with the brackets making clear these are Gemini's interpretations and not original text is exactly what you want from a transcription tool. It preserves accuracy and signals uncertainty in one pass. Now I took everything to Claude. This is where the GPS framework comes fully alive, and this is what I find most powerful about Claude for genealogical analysis. Claude can hold multiple documents in its attention simultaneously and identify the connections and the tensions between them. I gave it everything we had. Prompt number three Quote I am researching Catherine McLeod born approximately eighteen thirteen, parish of Kilmure, Isle of Skye in Vernesshire, Scotland. Here's the evidence gathered. Document one eighteen fifty one Scotland Census in Vernus. Catherine McLeod Gillis, age thirty eight, born Isle of Skye, husband Donald Gillis, blacksmith. Children Murdo Gillis, age nine, Mary Gillis, age seven, Katrina Gillis, age three. Document two eighteen forty one Scotland Census, Township of Kilmalog, Kilmere Parish. Household includes Murdoch McLeod, Age fifty five, Crofter, Mary McLeod, age fifty, Catherine McLeod, age twenty five. Note that eighteen forty one census ages for adults were often recorded to the nearest multiple of five. Document three Kilmere OPR Marriage Proclamation, december eighteen ten. Murdoch McLeod and Mary Morrison, both of this parish, proclaim for marriage. Additional note, research into Donald Gillis' family indicates his parents were named Duncan Gillis and Effie Gillis, not Murdo and Mary. Additional note, no baptism entry for Catherine McLeod born approximately eighteen thirteen has been located in the Kilmir OPR, and very few McLeod family entries of any kind appear in the Kilmir OPR after the early eighteen forties. Please do the following. First, create a comparison table showing how the ages across the two census records align or conflict. Second, analyze whether the naming pattern of Catherine's children supports the identification of her parents, taking into account that Donald Gillis's parents were not named Murdo or Mary. Third, evaluate the overall strength of the circumstantial case that Catherine McLeod Gillis is the daughter of Murdoch McLeod and Mary Morrison. Fourth, explain what genealogical gaps remain, what might explain the near absence of this family from the OPR after the early eighteen forties and what archives I should search next, end quote. Claude's response did exactly what I was hoping it would do. Let me walk you through each piece. On the age comparison table, Claude flagged the eighteen forty one census convention immediately, noted that Catherine's recorded age of twenty five represents a possible actual age of twenty five to twenty nine, and calculated that birth years of eighteen twelve to eighteen sixteen are all within this range. The eighteen fifty one census age of thirty-eight points to a birth year of approximately eighteen thirteen. Claude assessed these as consistent within normal variation for nineteenth century age reporting, and specifically noted that without understanding the census convention, a modern researcher might wrongly treat these as conflicting evidence pointing to different individuals. On the naming pattern, with Donald Gillis's parents confirmed as Duncan and Effie, Claude assessed the name evidence as meaningful. If Catherine's oldest son is named Murdo and her oldest daughter is named Mary, and those names cannot be attributed to Donald's family, then by the Highland Convention of Honoring grandparents, these names most likely point toward Catherine's parents. Claude noted this was strong circumstantial support for the identification but is not direct proof, because naming traditions were followed inconsistently and families sometimes honored other relatives. On the overall strength of the case, Claude described it as a well-supported working hypothesis. Multiple independent lines of evidence point to the same conclusion, the shared household in the 1841 census, the naming evidence now that Donald's family has been ruled out, the OPR marriage of Murdoch and Mary Morrison in the right parish and the right time frame, and the geographic consistency between Kilmure and Catherine's stated birthplace of Skye. Claude rated this as a reasonable basis for proceeding with research, but correctly stated that without a direct baptism entry or another document simultaneously naming Catherine and her parents, this remains probable rather than proven. And then Claude did something that made the whole investigation snap into focus. In answering that fourth question, why would a family nearly disappear from the OPR after the early eighteen forties, Claude connected two things I had in the prompt but hadn't explicitly leaked. It said The timing you describe is consistent with the disruption of eighteen forty three. Kilmere was a strongly Gaelic speaking community, and these communities were among the most enthusiastic supporters of the Free Church. If Murdoch and Mary McLeod followed their minister out of the Church of Scotland in eighteen forty three, their records after that date would not be in the OPR at all. They would be in the Free Church collection at the National Records of Scotland held under reference CH sixteen. The absence of records in the Church of Scotland registers after eighteen forty three may not mean the family disappeared. It may mean they changed churches in quote. I went back and looked at the Kilmuir OPR. Almost nothing for this McLeod family after the early eighteen forties. Perfectly consistent with a family that went free church after the disruption. That's not a failure. That's a discovery. We have now learned something specific and historically grounded about who these people were and where their records are likely to be. Now for the dead end I want you to see honestly. I searched the Kilmuir OPR for every year from 1812 to 1817. Every variant spelling, every abbreviation, nothing appeared. The Kilmir OPR for those years is fragmentary. There are sections where entries exist and sections where the page coverage is visibly incomplete. This is typical of Highland OPRs, where registration was inconsistent, depended on the diligence of individual session clerks, required payment to record in some periods, and was not universally observed even by Church of Scotland families. The absence of an entry does not mean the family wasn't there. It means the record wasn't made or was made and didn't survive. GPS element one includes knowing when you have genuinely exhausted the available sources in a category. I searched every OPR image for Kilmere in the relevant years. The entry's not there. That is the part of the honest record of this research. So here's where we stand. This is a partial answer. And I want to spend a moment on why that's not a defeat because I think the genealogy community sometimes treats partial answers as failures. They're not. They're an honest account of what the evidence actually supports. What we found through AI assisted investigation is genuinely significant. We identified a Murdoch, McLeod, and Mary Morrison, who married in Kilmure Parish in december eighteen ten. We found a Catherine McLeod in the eighteen forty one census living in the same parish in the specific township of Kilmelog in a household with Murdoch and Mary McLeod at ages consistent with them being her parents. We confirm that the names of Catherine's children, Murdo and Mary, cannot be explained by Donald Gillis' family background and align directly with the Highland naming tradition pointing toward a father named Murdo and a mother named Mary. We have a geographic consistency, chronological consistency, and naming pattern consistency across multiple independent sources. The circumstantial case is solid. But we don't have the direct proof. No baptism entry linking Catherine by name to parents Murdoch McLeod and Mary Morrison. And that distinction matters enormously in genealogy done to professional standards. This is exactly what the genealogical proof standard is asking us to do. Not to find evidence but to evaluate it honestly. To say here is what I know, here is what I believe with good reason, and here is what remains unproven. That intellectual honesty is what separates a family tree worth trusting from a chain of hopeful guesses dressed up as fact. Claude Summary said it clearly. The preponderance of evidence supports the identification of Murdoch McLeod and Mary Morrison as the probable parents of Catherine McLeod Gillis. This relationship should be recorded as probable in any genealogical database entry with a research note documenting the evidence and specifying what direct documentation is still needed to confirm it. That is a soundly reasoned, coherently written conclusion. That is GPS element five. The conclusion follows from the evidence, it accurately describes the evidence's limits, and it maps exactly what needs to happen next. I know some of you are nodding along right now. You've been exactly here. You found the right family in the census, you found the marriage that fits. And then the direct baptism record isn't there, and you're left with strong evidence and no clear answer. You're not doing genealogy wrong. That wall is part of the landscape. The question's not whether you can always find the direct proof. The question is whether you understand the evidence you do have well enough to describe it honestly, and whether you know enough about the archive system to know where to look next. Today AI helped with both of those things. Your homework for this week. If you have Scottish ancestry, go to Scotland's people and find one ancestor in the eighteen forty one or eighteen fifty one census. Look at their household. Note all the names and ages. Then go to Perplexity and ask it to explain what OPR records exist for that specific parish, what the disruption of eighteen forty three might mean for your research in that area, and what the highland naming tradition might tell you about the names of the people in that household. That's all. See what you learn. If you don't have Scottish ancestry, run this same exercise for any unfamiliar research territory. The orientation prompt, giving perplexity a specific person, place, date, and four distinct questions is the most underused AI technique in genealogy. It costs five minutes and it can save you days. And for those of you keeping track of the methodology, today we demonstrated GPS elements one, three, four, and five. Reasonably exhaustive search when we combed every OPR year in range and honestly reported finding nothing. Analysis and correlation of evidence when Claude built the comparison table across three documents. Resolution of conflicting evidence, when we understood the census rounding rule and used Donald Gillis' family to validate the naming pattern, and soundly reasoned conclusion, when we stated clearly what we know, what we believe with good reason, and what remains unproven. That framework is what separates solid research from guestwork. And AI helps us work within it faster than we could before. And for my Australian and UK listeners, today's research territory is especially yours. UK researchers, Scotland's people is your primary starting point, and the National Records of Scotland and Edinburgh holds the free church registers and Kirk Session archives that go far beyond the Church of Scotland's OPRs. The Highland Archive Centre in Inverness is worth knowing for local Invernessre collections. For my Australian listeners, particularly those in Victoria and New South Wales where thousands of Sky and Lewis immigrants settled in the eighteen fifties, the Highland and Island Immigration Society records on Scotland's people list many of those immigrants by name. The Public Record Office Victoria holds ship arrival records for the period. And Trove, the National Library of Australia's free research archive at Trove.nla.gov.au, has notices from Scottish immigrant communities that can make connections no other source reveals. Thank you so much for listening to Ancestors and Algorithms. If you enjoyed this episode, please leave a review wherever you listen to podcasts. And if you know a fellow genealogist or family history researcher who could benefit from what we cover today, share this episode with them. That is the best way to help our community grow. Now, for my Patreon members, the companion guide for this episode is in your library. Twelve advanced prompts, including a full multi-tool workflow for correlating OPR and census evidence across multiple Scottish parishes, a GPS research checklist built specifically for pre-civil registration highland research, and detailed case studies for working through the Free Church Records and the Kirk Session Archives. And if you've been thinking about joining, I host monthly live QA sessions on YouTube exclusively for members where we work through research problems together. Head to ancestorsandai.com to learn more. I'm your host Brian, and I will see you next week for another journey into the past powered by the future. Until then, happy researching.