Ep. 40: Seven Heirs - How AI Decodes a Tennessee Probate Mystery

You've found the estate settlement. You've counted the heirs in the distribution sheet. And the math doesn't add up.

Nine children appear in the 1860 census. Seven names appear in the 1874 settlement. Two heirs are gone with no explanation, no death notation, and no trace in the legal record.

That is where this episode begins.

Probate records are among the most underused sources in American genealogy. When an ancestor died without a will, the intestate settlement process required the court to document every legal heir by law. When someone is missing from that list, there is always a reason. This episode shows you how to find it.

In Episode 40 of Ancestors and Algorithms, host Brian works through an 1870s Tennessee intestate estate using four AI tools: Claude, Perplexity, ChatGPT, and NotebookLM. Step by step, prompt by prompt, you watch AI transform a dense 19th-century legal document into a focused research roadmap for tracing missing heirs.

WHAT YOU'LL LEARN:

How to use Claude to analyze a probate distribution sheet and surface the anomalies that are easy to read past, including a buried legal clause that completely changed the research direction.

How to use Perplexity to research historical intestate succession law so you understand exactly why a legal heir might be absent from an estate settlement with no death record to explain it.

How to use ChatGPT to generate every plausible reason a family member might be missing from a probate document, including scenarios most researchers never consider: daughters recorded only under married surnames, guardianship proceedings filed separately from the estate, and creditor debt attachment.

How to use NotebookLM to cross-reference census records, tax lists, and estate documents together and identify what the evidence actually establishes versus what you are inferring.

THIS EPISODE IS FOR YOU IF:

• Your ancestor's probate distribution lists fewer heirs than the census records suggest there should be. 
• You are facing a 19th-century estate settlement full of archaic legal terms you cannot parse. 
• You research Tennessee ancestry from the Civil War era through the early 1900s. 
• An ancestor disappeared from the records after a death in the family and you have no idea where to look next. 
• You want to see exact, copy-paste AI prompts designed for genealogy research before trying them yourself.

The outcome is honest: this mystery is not fully solved. The research produced two legally grounded theories and identified a precise next record set in a physical archive that has not yet been digitized. Sometimes the win is knowing exactly where to look. That is real genealogy.

TOOLS: Claude, Perplexity, ChatGPT, NotebookLM RECORDS: Intestate estate settlements, distribution sheets, probate inventories, Tennessee county tax records, FamilySearch Tennessee Probate Court Files 1795-1955 GPS: All five elements of the Genealogical Proof Standard

Companion Guide and free resources at ancestorsandai.com.

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

0:11

There's a moment in genealogy research that stops you cold. Not the exciting kind of stop where you find the record that answers everything, the other kind, where the math doesn't add up, where a legal document tells you something that contradicts everything you already know. I found a sixty three page estate settlement in a Tennessee courthouse file, a formal legal document from eighteen seventy four required by law to name every living heir of a man who died without a will. It listed seven names. I had nine children documented in the census records. Nine and two of those people are nowhere in that legal document. No explanation. No mention of their deaths. No notation from the administrator. Nothing. That mystery sent me through four AI tools, two weeks of dead ends, a crash course in 1870s Tennessee inheritance law, and a wall I still haven't broken through. Today I want to show you exactly how that research unfolded. Let's dive in. Welcome to Ancestors and Algorithms, where family history meets artificial intelligence. I'm your host Brian, and today we're going deep into one of the most information rich and most underused record types in all of American genealogy, probate records, wills, estate settlements, letters of administration, inventories of personal property. These documents can name every child, every spouse, every sibling, every neighbor who was owed money. They can give you a complete family picture at the moment of death in ways that no other record type can match. But they are dense. They are full of archaic legal language. They reference court processes that haven't existed for 150 years. And they assume you understand things most of us have never studied. That is exactly where AI comes in. Today I'm going to walk you through the workflow prompt by prompt, including an honest ending that I think a lot of you are going to recognize. Let's get started. Let me introduce you to Silas Emery Cravins. Silas was born around 1822 in Cannon County, Tennessee, hilly country on the eastern highland rim, southeast of Nashville, roughly halfway between Nashville and Chattanooga. The terrain rolls and pitches across the upper Cumberland landscape. The farms run smaller than what you would find in the river bottoms to the west, and the limestone breaks through the surface of the fields and long gray shelves. But by the eighteen sixty Federal Census, Silas was doing well. He was listed as a thirty eight year old farmer with one hundred sixty acres, a wife named Polly and nine children, ranging from a teenage son working the farm alongside his father down to an infant daughter who hadn't yet learned to walk. It is the picture of a midcentury Tennessee farm family doing exactly what they were supposed to do. Then the Civil War arrived. Cannon County sat in the divided heart of Middle Tennessee. The county was occupied by Union forces for most of the war, but populated in large part by families who held divided loyalties or who primarily wanted to survive and hold their land. The economic disruption to agriculture communities in that zone was real and lasting. Land values fell, farm productivity dropped. Families scattered. By the eighteen seventy census, the picture had changed substantially. Silas was a widower. Polly had died somewhere in the decade between eighteen sixty and eighteen seventy. His real estate value had declined from what it had been before the war. But Silas was still farming, still in Cannon County, with two of his younger children still in the household. Seven of the nine children appear in or near the eighteen seventy census record in the same township or in adjacent ones. The two I couldn't account for had stopped showing up. Silas Emory Cravens died in the spring of eighteen seventy two. He died interstate, which is the legal term for dying without a valid will on record. And that triggered a court process I had to learn almost entirely from scratch. When someone in eighteen seventies Tennessee died interstate, the county court opened an administration proceeding. A close male relative, typically a son or son in law, was appointed administrator. That administrator was legally required to take an inventory of the decident's property, pay outstanding debts from the estate, and distribute whatever remained to the heirs according to the state's entity statutes. Every step of that process was documented by the court, and if those documents survived, which in Cannon County many of them did, they are an extraordinary genealogical resource. I found Silas's estate file through the Family Search Catalog. The collection to search is the Tennessee Probate Court Files 1795 through 1955, a massive ongoing digitization project that has made thousands of county probate files available online for free. I pulled up the images for Cannon County in the early 1870s and spent an afternoon working through a stack of fragile handwritten documents. What I found was a sixty three page estate settlement filed in Cannon County in 1874, roughly two years after Silas' death. The file included the administrator's appointment, a detailed inventory of personal property that itemized everything from livestock down to the individual farm tools, a record of debts paid, and at the center of it all, the distribution sheet. The distribution sheet is the document that names every person legally entitled to receive a share of what remained after debts were paid. Under Tennessee Intimacy Law with no surviving spouse, the estate passes to the children in equal shares. I read that distribution sheet four times. Seven names. Seven people listed as receiving equal shares of what remained after the debts were paid. The shares were modest as you'd expect for a small farm in a county still recovering from the war, but the legal process was properly followed. The administrator had done his job. I want you to understand what I was feeling in that moment, because I think many of you will recognize it. It wasn't panic, it was that slow, creeping discomfort of realizing that something you thought you understood doesn't add up. I had built what I thought was a solid family reconstruction. I had documented Silas's nine children from the eighteen sixty census. I had traced as many of them as I could through the eighteen seventy census, and I had assumed the way genealogists sometimes do in a record set feels comprehensive, that the estate settlement would confirm what I already knew. Instead, it was contradicting it. Nine children in the 1860 census, seven in or near the eighteen seventy census with two unaccounted for. Seven names in the eighteen seventy-four estate settlement. Did the two missing children die between eighteen seventy and eighteen seventy-four? Tennessee didn't begin statewide death registration until 1914, so the absence of a death record proves nothing. But there was nothing in the estate file either. No notation, no reference to their existence at all. In a document that meticulously itemized Silas's farm equipment down to a worn out plowblade, two of his children had ceased to exist on paper. That's the mystery I brought to AI. And here's the key thing to hold on to as we work through this. AI is your research assistant, not your researcher. The tools I'm about to show you cannot go to a Tennessee courthouse and pull microfilm. They cannot find a record that doesn't exist online. But they can compress weeks of solo research into hours and help you ask much better questions of the records that do exist. That's an enormous amount of value. Let me show you exactly how it works. The first thing I needed was a translator. Nineteenth century probate documents are written in a legal style that is nearly its own dialect. Terms like relic, dower, intimacy, distribute, and commissioners of appraisement appear throughout like speed bumps. I don't have a law degree. I needed someone to walk me through this document in plain English and flag anything that seemed unusual. So I open Claude. Claude's strength when working with dense documents is its ability to take a set of specific analytical tasks and work through them methodically. I don't want a general summary when I'm working with a legal document. A summary smooths over the details that matter most. What I want is four distinct analytical passes, who is named, what does the language mean, what is strange, and what is implied but not stated. I worked from a transcription of the estate settlement and gave Claude this prompt. Quote, I'm going to paste in the text of a nineteenth century probate estate settlement document from Tennessee dated 1874. Please do the following one, list every person named in the document along with their role as stated or implied, heir, administrator, witness, creditor, or appraiser. Two, identify every legal term a modern reader might not recognize and explain each one in plain, clear English. Three, flag any phrases or clauses that seem unusual, incomplete, or that raise a question a genealogist would want answered about this family. Four, note any implied relationships or circumstances not stated explicitly but reasonably inferable from the legal language. Four numbered tasks, each targeting a different layer of the document. That third task is the critical one. I am explicitly asking Claude to find anomalies, not explain the document in general. That is a fundamentally different instruction and it produced a fundamentally different result. Claude worked through all four tasks in sequence. On task one, it identified eleven people named in the document, the administrator, seven named distributees, two witnesses, and one creditor. The basic cast of characters established. Task two turned the legal terminology into plain English. Relic, Claude explained, is an archaic term for a surviving widow or widower who retained certain legal rights in the estate. Dower referred to a widow's statutory right to a portion of her deceased husband's real property, a right that existed independently of any will or intestacy proceeding. The document referenced Polly Cravens as a relic in a historical clause, which initially confused me since I knew Polly had predeceased Silas. Claude's explanation helped me understand that the administrator was formally acknowledging dower rights had been considered and resolved, even in the absence of a surviving widow. That's the kind of nuance I wouldn't have caught without the translation. Task four added one more observation. Claude noted that the administrator named in the settlement was also one of the seven distributees, meaning he was receiving his own share of the estate while simultaneously managing the process. In nineteenth century Tennessee, this was common practice. A son serving as administrator was the norm. But Claude flagged it as a detailed worth keeping in mind, since an administrator who is also a beneficiary had a personal interest in how the settlement was structured. And then task three stopped me cold. Buried midway through the settlement language was a phrase I had read several times without fully processing. Claude flagged it with a clear note. And the phrase was and to the assignees of those heirs as may have conveyed their interest prior to this date. Claude's explanation was direct. If individuals who are known heirs do not appear in the distribution list, it is possible they transferred their interstate shares to another party through a deed of conveyance executed prior to the settlement date. The genealogist pursuing this question would want to search the county deed books for the years 1872 through 1874 for any transactions involving the decident's surname or the names of known heirs, in quote. My working assumption from the start had been that the two missing children were dead. Claude had now handed me a legitimate alternative. They might have been alive, sold their inheritance rights to a third party before the settlement was filed, and their names replaced in the distribution list by whoever bought those rights. That changes which records I need to look for entirely. That flag clause changed the shape of the mystery. Before going further, I needed to understand exactly what Tennessee law said about assigning interstate shares in the eighteen seventies. That is a legal history research question, and legal history research questions are where perplexity does its best work. It searches the current web with citations and I need verifiable sources, not general knowledge. Here's the prompt. Quote I'm researching Tennessee into state secession law as it existed in the early eighteen seventies. Please find and explain the following one, how did Tennessee law in this era determine the order of inheritance when someone died without a will? Who was entitled to inherit and in what proportions? two, what was the legal status of a widow's dowers right in Tennessee from approximately eighteen seventy to eighteen seventy five? And how would those rights interact with an interstate estate settlement? Three, were there any circumstances under which a legal heir could be absent from or excluded from an interstate estate settlement? For example, could an heir sell or assign their share of an estate before the settlement was completed? Please cite specific sources wherever possible, end quote. The response confirmed what Claude had flagged. Under Tennessee law in the early 1870s, the estate after payment of debts was divided equally among the surviving children. If a child had predeceased the decident but left living children of their own, those grandchildren would collectively inherit their parents' share. A surviving widow held dower rights, a life interest in one third of the real property, separate from the children's distribution. Since Polly had predeceased Silas, no active dower claim existed at settlement time, though the document's formal acknowledgement of this was legally required. And on the question of assignment, yes. Under Tennessee law in this period, an heir's right to an interstate share was a property right that could be transferred by deed of conveyance before the estate was formally settled. This was not unusual. Heirs who needed cash immediately, who lived at a distance, or who had pressing debts would sometimes sell their interstate share to another heir, a neighbor, or an attorney for a lump sum payment well below the eventual settlement value. Now I had two legitimate, legally grounded theories for my two missing heirs. Theory one, they died between eighteen seventy and eighteen seventy four before the estate was settled and left no children of their own. Theory two, they were alive but had sold their interstate shares to another party before the settlement was filed. Both theories were consistent with the document language. Both were consistent with Tennessee law, and I had no evidence yet to distinguish between them. This is exactly the branching scenario problem where Chat GPT earns its place in the toolkit. Not document analysis, not legal history, lateral thinking helping me generate possibilities I might not have considered on my own. Here's the prompt. Quote I'm researching a genealogical mystery. A Tennessee farmer died interstate in eighteen seventy two. The official estate settlement lists seven heirs, but I have documented evidence from the eighteen sixty and eighteen seventy census that the decident had nine living children at or near the time of death. The settlement contains a clause suggesting some heirs may have conveyed their interest prior to the settlement date. Please give me a comprehensive list of every plausible explanation for why two heirs might be absent from an eighteen seventies Tennessee interstate estate settlement. For each explanation, tell me what records would typically exist to prove or disprove it, and where a genealogist might find those records, end quote. ChatGPT generated eleven distinct scenarios. I want to walk through several that I hadn't considered. The obvious ones were already on my list, death before settlement or assignment of interest by deed of conveyance. Those are the top two and most researchers would stop there. Scenario four stopped me. The two missing heirs may have been daughters who married before Silas' death, and their married names appear in the distribution list in places where I don't recognize them as Cravens descendants. I had been scanning the settlement for variants of the surname Cravens, but if a daughter married someone named Lankford or Foltz or Minot, she might appear in that document under her married name with no visible connection to the family I was tracing. That was a search I hadn't run. Scenario six introduced a record type I hadn't thought to look for, guardianship files. If one of the missing heirs had been declared legally incompetent or incapacitated, a separate guardianship proceeding might exist in the county court records. Their share of the estate would be managed by a guardian, and the distribution list might reflect the guardian's name rather than the heir's. Scenario nine was sobering in a different way. In the post Civil War South, an heir who was known to be deeply in debt could have their interstate share attached by a creditor before the estate was settled. The share would go to pay the debt rather than the heir directly, and the heir's name would not appear in the final distribution at all. Given that Cannon County families were navigating genuine economic hardship in the early 1870s, this wasn't a remote possibility. That's eleven total explanations across scenarios ranging from the straightforward to the genuinely unexpected. Now I was genuinely excited, and I made the classic genealogy mistake. I latched on to theory two, the assignment of interest. The document language supported it, the legal confirmation from Perplexity supported it, and so I spent two full days searching for a deed conveyance in the Cannon County records that would show one or both missing heirs selling their interstate share. I found nothing. The online deed indexes for Cannon County in this period are incomplete. Ancestry has partial coverage. Family Search has some additional records, but gaps remain. A deed executed in 1872 or 1873 for a small farm family transaction in a rural Tennessee County might not be accessible online at all. I had to pivot. I stepped back and asked a more honest question. What can I actually verify with the records available to me right now? That question led me to Notebook LM. Notebook LM is not a chatbot. It doesn't search the web and it doesn't draw on general knowledge. What it does and does exceptionally well. Well, is analyze and answer questions based solely on the documents you upload to it. Every response comes from your sources and only your sources. When you're deep in a research case with a collection of documents gathered over multiple sessions, Notebook LM helps you understand what those documents say in relation to each other. It will catch contradictions between sources that you might miss when reading them individually weeks apart. I uploaded four things to a Nobook LM notebook, the state settlement text, the 1860 census page, the 1870 census page, and my own research notes formatted as a plain text document. Then I asked, quote, based only on the documents I have uploaded, please identify any contradictions or inconsistencies between the sources, then list the specific questions these documents raise but cannot answer on their own. For each open question, suggest what type of additional record might resolve it, end quote. Notebook LM confirmed that two names present in the eighteen sixty census were absent from both the eighteen seventy census and the eighteen seventy four estate settlement. It flagged that pattern as significant. And it did something important. It explicitly declined to confirm whether the two individuals absent from the eighteen seventy census were the same two absent from the eighteen seventy four settlement. That might seem like an obvious inference. Nobook LM correctly refused to treat inference as fact. The documents it had didn't actually establish that connection, and it wasn't going to pretend otherwise. It also flagged something useful from my own research notes. I had recorded the approximate birth years of all nine children from the 1860 census. The two missing individuals, based on their ages, would have been young adults by 1872, somewhere in their mid to late twenties. That age range is consistent with marriage, with migration out of the county, and frankly, with death from disease, all of which were common outcomes for young Tennesseans in the decade following the Civil War. That is the genealogical proof standard working the way it should. AI is your research assistant, not your researcher. A tool that tells you when the documents don't establish the connection you're hoping for is protecting you from yourself. That is a real service. Notebook LM's list of suggested next record types included Canon County deed books from 1872 through 1874, county marriage bonds to check whether the absent individuals married and changed surnames, guardianship files, and Tennessee County tax records from 1871 and 1872. That last one I hadn't considered. I searched the available Tennessee County tax records through family search and ancestry for that period, and the two missing individuals did not appear paying property taxes in the county. That nudges me slightly toward theory one that they had died, but it doesn't prove anything. Okay, I have to be straight with you about where this stands. I didn't crack this one. After three weeks of research, four AI tools, dozens of searches, and two days chasing a deed conveyance that might not exist in any online database, I don't have a definitive answer for why two of Silas Emory Craven's children are absent from his eighteen seventy four estate distribution. A brick wall in genealogy is not failure. Let me tell you exactly what I know and what I don't because that distinction is the whole point of today. What I know The estate settlement is authentic and legally complete. Seven people received shares. The document contains an explicit legal clause acknowledging that some heirs may have conveyed their interest prior to the settlement date. That clause exists because such conveyances were real, common, and legally recognized. Tennessee law in the eighteen seventies allowed any heir to sell their interstate share by deed to a willing buyer at any time before the final settlement was filed. The 1871 and 72 Tennessee tax lists for Cannon County don't show the two missing individuals paying property taxes, which is a weak signal they may have been deceased or absent from the county, but weak is the right word. It's not evidence of death. What I don't know whether the two individuals absent from the eighteen seventy census are the same two absent from the eighteen seventy four settlement, it's the most logical inference. But inference is not proof, and connecting those absences requires records I haven't accessed. Whether either of the missing heirs died before eighteen seventy four. The absence of death records is expected, not evidential, in a state that didn't begin statewide death registration until nineteen fourteen. Whether a deed of conveyance exists in the Cannon County Deed Book showing one or both missing heirs selling their interstate interest, I couldn't reach the books online. Whether either of the missing heirs was a daughter who married before Silas' death and appears in the settlement under a married name I haven't yet identified. Your homework. For most states that's before 1900. Go to the family search catalog, search your ancestor's county and time period, and find out what probate collections exist. You don't have to read everything today. Find the collection, confirm it's there, and note the format. That 20 minute step could change your research entirely. For my Australian and UK listeners, today's techniques translate differently. Australian researchers should start with the Public Record Office Victoria for Victorian Era Probate Records, the NSW State Archives Collection through Museums of History NSW for New South Wales estates, and Trove for probate notices in historical newspapers. For UK researchers, the critical dividing line is january twelfth, eighteen fifty eight, when the Principal Probate Registry replaced more than 300 ecclesiastical courts that had handled probate jurisdiction for centuries. Before eighteen fifty eight, the prerogative court of Canterbury held highest jurisdiction and its records are accessible through the National Archives at Kew. After eighteen fifty eight, records became centralized and far more accessible. By my past has strong coverage across both periods. The same structured prompts I use today on an eighteen seventy four Tennessee estate settlement work equally well on an eighteen fifties Lincolnshire one. 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 who could benefit from what we cover today, please share this episode with them. That is the best way to help our community grow. For my Patreon members, the companion guide for this episode is now in your library. It includes 12 advanced prompts, a multi-step workflow for working through any probate document from initial transcription to written proof argument, and a GPS research checklist tailored to Estate and Will Research. Head to ancestorsnai.com for everything, including information about the monthly live QA sessions I host exclusively for members on YouTube. 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.