Anybody Holmes?

Focusing on Chromosome 22 Using the MyHeritage Chromosome Browser

This presentation focuses on a single reconstructed span: the segment on chromosome 22 (Build 37 coordinates 16,494,186–44,558,812).

My brother Steven matches both Enoch and Cheri across nearly the entire span of chromosome 22. The only interruption occurs beyond 44.5 Mbp, where Steven transitions to DNA inherited from his father's paternal great-grandfather, Weldon Borland. Up to that breakpoint, Steven, Enoch, and Cheri share a continuous segment.

Because Steven matches both descendants of John Forbes across this span, the segment must have been inherited by Steven from his great-grandmother, Elizabeth (Forbes) Borland-Tripp, the sister of their ancestor John Forbes.

We can narrow the inheritance path further. Elizabeth and her brother John inherited this chromosome 22 segment from one of their parents. Because the segment is continuous and shows no internal recombination, that parent must have transmitted the same intact chromosome 22 copy to both siblings. Therefore, the ultimate source of this segment is one of Elizabeth and John's four grandparents.

Elizabeth Forbes' parents were Francis Morrison Forbes and Pearl Beatrice (Price) Forbes, known in their time as Frank and Dollie. If a recombination event had occurred when Frank or Dollie transmitted chromosome 22 to Elizabeth, there would have needed to be a recombination at precisely the same location when transmitting to John in order for the shared segment to remain continuous. That scenario is statistically unlikely. The far simpler explanation is that both siblings inherited the same unrecombined chromosome 22 copy from one grandparent.

Steven shows no matches to relatives from Frank's side of the family on chromosome 22. He does, however, have multiple matches to relatives descending from Dollie's line within this span. Therefore, the evidence strongly supports that this entire chromosome 22 segment traces through Pearl Beatrice (Price) Forbes and ultimately through one of her parents — either her father Rezin Price or her mother Virginia.

All of the Price matches mentioned above fall on the Rezin Price side; there were no matches on the Virginia Kail side.

Tapping Ancestry's Enormous Database

With that chromosome 22 lineage established, the next step was to organize Steven's Ancestry matches so we could map the Holmes-side matches to specific ancestral branches.

Preliminary Match Organization on Ancestry

Before beginning the chromosome-level analysis, I first organized Steven's paternal AncestryDNA matches into clearly defined groups. All matches sharing 20 cM or more were reviewed and sorted by family side.

The first division separated paternal grandmother matches from paternal grandfather matches. This step was straightforward because numerous close relatives from Steven's paternal grandmother Helen Freudenberg's side have tested with Ancestry (including my father's maternal half-brother Michael). Any kit appearing as a shared match with those relatives was assigned to the paternal grandmother category. Most of the remaining paternal matches belonged to the line of Steven's paternal grandfather, John Borland, or were otherwise easy to identify as belonging to the Freudenberg side.

The matches designated as John Borland side were then subdivided into the two branches: the Elizabeth Forbes line and the Weldon Borland line (corresponding to John's father). This subdivision was facilitated by the fact that John's paternal half-brother, Jeff Borland, also tested with Ancestry. Having previously analyzed Jeff's matches, I was already familiar with the close relatives on the Weldon Borland side who had tested at Ancestry, and used their shared matches to sort as I had done in the previous stage.

The next layer of separation involved dividing the Elizabeth Forbes matches into Forbes versus Price descendants. This proved manageable because Frank Forbes' ancestors immigrated to the United States relatively recently and therefore Steven had comparatively fewer DNA matches on this side. The Price line, by contrast, has deep roots in Ohio and traces back to Colonial times in America.

While pedigree collapse can sometimes complicate DNA analysis, in this case it proved advantageous. I descend from the Kail family twice — once through Weldon Borland and once through Dollie Price. Because I had already worked extensively with Jeff Borland's matches and maintain a comprehensive Kail family tree on Ancestry, I was very familiar with surnames of the female lines in the Kail descendant lines. This made separating the Rezin Price matches from the Virginia Kail matches straightforward.

Separating the Price from the Holmes matches took just a couple days for the same reason. Weldon descends from this same Price family via multiple lines as they were early pioneers of Leesville Ohio and surrounding areas. There were just 3 matches I couldn't figure out at this level of splitting, but that's well under 1%, and I'll save those for a challenge on a rainy day, as they will not make or break this study.

The resulting breakdown of Steven's 20+ cM paternal matches, categorized according to this framework, is summarized below.

Identifying the Chromosome 22 Matches On Ancestry

With the Mary Holmes branch isolated to 182 paternal matches at or above 20 cM, the next objective was to determine which of those matches corresponded to the reconstructed chromosome 22 span.

The approach was straightforward. I reviewed the matches I had tagged with the "Mary Holmes" group label on Ancestry and looked for names that were recognizable from prior work on MyHeritage. A small number of those names appeared among the Mary Holmes–tagged matches. I then examined the shared matches for several of these individuals to determine whether a consistent pattern emerged.

A clear signal quickly became apparent. In each case, the highest shared match was a kit belonging to Justine, who matches Steven at 45 cM on Ancestry. Justine shares exactly 180 matches with Steven above the 20 cM threshold. All of those shared matches fall below Justine's 45 cM level, making her the highest-centimorgan match within that network of shared matches.

Of Justine's 180 shared matches, 11 had already been categorized elsewhere in the pedigree breakdown:

• 1 was a descendant of Clunn Holmes Price (previously grouped with the James Price branch as a Price–Holmes descendant rather than as a Mary Holmes cousin),
• 9 were coincidental descendants of my ancestor Daniel Smith (grouped under Weldon Borland), and
• 1 was a documented cousin on the Helen Freudenberg side.

After accounting for those 11 individuals, the remaining shared matches totaled 167 — effectively corresponding to 180 + Justine − 11 − 2 = 167 of the 182 matches I had initially tagged as belonging to the Mary Holmes line by process of elimination.

Given this overwhelming overlap, I applied the Mary Holmes group tag to the 11 "double cousin" matches as well, recognizing that their connection to Mary Holmes was clearly represented within this shared network.

The difference between the 167 shared matches connected through Justine and the 182 matches initially tagged to the Mary Holmes line requires explanation. The answer lies in additional inherited segments outside chromosome 22. According to the Borland Genetics Chromosome Browser, Steven matches the 2023 reconstruction of Mary Holmes not only on the 41 cM segment on chromosome 22 that forms the basis of this study, but also on three additional segments: a 16 cM and a 17 cM segment on chromosome 12, and a 22 cM segment on chromosome 14. These additional segments account for the remaining Holmes-side matches that do not appear among Justine's shared matches, as they correspond to different inherited portions of Mary Holmes' genome.

Steven's Match With Justine

With Justine now functioning as the central shared match within the Mary Holmes–tagged group on Ancestry, the next logical step was to determine precisely how she connects to Mary Holmes.

Although Justine's tree was public, it displayed only a single individual — her paternal grandfather. However, she used her full name as her match name, and that, combined with the limited information visible in her tree, provided enough detail to begin reconstructing her lineage. Over some time, I built out her family tree at least five generations in every direction using publicly available records and subscription resources on Ancestry.

This effort revealed that one of Justine's fourth great-grandmothers was Rebecca Holmes (1823–1910), who already existed in my own family tree. Rebecca was the daughter of Samuel Holmes, a brother of Isaac Holmes — the father of Mary Holmes. In addition, Rebecca's mother, Mary McNabb, was the sister of Elizabeth McNabb, who was Mary Holmes' mother.

In other words, two Holmes brothers married two McNabb sisters. Steven descends from one of those couples through Mary Holmes, while Justine descends from the other through Rebecca Holmes. Their shared DNA therefore reflects inheritance from both the Holmes and McNabb parental lines.

One noteworthy point about the Justine match is its reported length. Ancestry lists the shared segment between Steven and Justine as 45 cM in a single segment, which is actually larger than the 41 cM span that the Borland Genetics reconstruction reports for the same region. I interpret this difference not as a contradiction, but as a consequence of the way different tools calculate centiMorgan totals. Borland Genetics uses a polynomial recombination map that interpolates recombination rates continuously along the chromosome, whereas Ancestry's proprietary map is based on discrete reference points. The different algorithms can yield slightly different centiMorgan totals for the same physical span. As a result, the Ancestry and Borland Genetics cM figures will not match exactly even when they reflect the same shared DNA. In this case, both platforms are describing the same underlying shared inheritance across chromosome 22; the slight numerical difference reflects the underlying map models rather than a real biological discrepancy.

Within this same segment, I identified ten additional descendants of Samuel and Mary (McNabb) Holmes: two through their daughter Rebecca (Trisha and Marcos), three through their daughter Natalie (Chris 2, Douglas, and Valerie), two through their daughter Phoebe (Chris 1 and Frank 2), and three through their daughter Elizabeth (Frank 1, Jean, and Robert). This pattern indicates that the segment was transmitted from either Samuel or Mary to at least four of their children — Rebecca, Natalie, Phoebe, and Elizabeth — and subsequently passed down through multiple independent lines of descent.

When descendants of double first cousins match across a continuous span of a chromosome — absent coincident crossover events that would mimic continuity — the segment must have been inherited intact from a single one of the cousins' shared great-grandparents. In this sense, double-first-cousin analysis can provide even greater mapping resolution than sibling comparison. The challenge, of course, is determining which of the eight shared great-grandparents is the source.

Fortunately, additional Ancestry matches allow us to narrow the possibilities further. Both Steven and Justine match C.S., a descendant of Jacob Holmes (brother of Isaac and Samuel Holmes), and two descendants of Sarah (Holmes) Osbun, another sibling of Isaac and Samuel: Martha and Stuart. Although these additional matches do not span the entire chromosome 22 segment (C.S. and Martha measure 31 cM and 21 cM respectively with Steven), Justine's 45 cM match establishes the full inherited interval. Taken together, these matches allow us to confidently map the entire span to one of Isaac Holmes' four grandparents.

Getting to Know the Holmes Family

At this stage of the analysis, we have pushed the evidence back to a generational depth beyond the reach of Ancestry's automated tools. Shared Ancestor Hints and ThruLines extend only to fifth great-grandparents — and Isaac Holmes occupies that exact generational boundary in Steven's pedigree. As a result, those features can no longer assist us in identifying earlier ancestors connected to this segment.

Before developing a strategy for evaluating the substantial body of remaining matches — 167 by my count — it is necessary to step back and understand the Holmes family itself. To interpret the genetic evidence properly, we must know who Isaac Holmes was, where his family lived, and what ancestral lines appear in his pedigree.

Below is Isaac Holmes' pedigree as documented in my Ancestry tree. Of particular note is that Isaac's grandparents — one of whom transmitted the entire chromosome 22 span in question — were all born around the year 1700. That generational placement will become important as we evaluate the historical and genetic context of this segment.

Isaac Holmes (1764–1851), son of Obadiah Holmes (1728–1794) and Mary Clunn (c. 1730–1812), was born in Trenton, New Jersey. As a child he moved west with his family through Virginia and Pennsylvania before establishing himself in eastern Ohio, where he became an early settler of the Leesville area in present-day Carroll County. He married Elizabeth McNabb in 1794, raised a large family on the Ohio frontier, and lived there until his death in 1851 at age eighty-seven.

Isaac's father, Obadiah Holmes (1728–1794), was born in Upper Freehold, Monmouth County, New Jersey, to Joseph Holmes (1698–1777) and Elizabeth Ashton (1700–c.1750). In early adulthood he lived in New Jersey and on Staten Island, working as a boat builder before later engaging in merchandising in Trenton. Beginning in the late 1760s, he moved his family steadily south and west along the Great Wagon Road through Virginia and Pennsylvania before settling in what is now Brooke County, West Virginia. His children would continue that westward movement into Ohio.

Joseph Holmes (1698–1777), Isaac's grandfather, was born in Middletown, Monmouth County, New Jersey, the son of Obadiah Holmes (1666–1745) and Alice Throckmorton Ashton (1671–1716). He married Elizabeth Ashton in 1722 and raised a large family in Upper Freehold and Freehold, including Obadiah (1728–1794), through whom this line migrated west.

Obadiah Holmes (1666–1745), Isaac's great-great-grandfather, was born in Newport, Rhode Island, to Jonathan Holmes (1634–1713) and Sarah Borden (1644–1701). By 1696 he had relocated to Monmouth County, New Jersey, where he married Alice Throckmorton Ashton.

That Obadiah was the grandson of the Reverend Obadiah Holmes (1606–1682), an early Rhode Island settler and Baptist minister. Born near Manchester, England, he emigrated to New England in the 1630s. After disputes over religious practice in Massachusetts and Plymouth colonies, he settled in Newport, Rhode Island, where he served as pastor of the Baptist church for thirty years. In 1651 he was publicly whipped in Boston for refusing to renounce his religious beliefs. He died in Newport in 1682, leaving descendants who would spread from Rhode Island into New Jersey and eventually westward.

Isaac's mother, Mary Clunn (c. 1730–1812), was the daughter of John Clunn (1704–1759) and Margaret Lynch (1698–1781). John Clunn was born in London and emigrated to America in 1725. Margaret Lynch was born in Dublin, Ireland. They married in New Jersey in 1734 and raised their family in Mercer County. Mary married Obadiah Holmes about 1751 and accompanied him through decades of migration from New Jersey to Staten Island, down the Shenandoah Valley, and ultimately into the upper Ohio frontier, where she died in 1812.

Secondary Clustering Strategy

Since working through the remaining 167 matches individually from the top down would require substantial time and effort, I developed a more efficient strategy to organize the workflow. The objective is to divide the list into smaller sub-groups of matches who are likely related to Isaac in the same way, and then attempt to solve each group as a unit — "solve" meaning to identify and connect their shared ancestor(s) to Isaac's family tree.

Although the 167 matches already function collectively as a segment-level cluster, because they all match one another on the same chromosome 22 segment, what I am now seeking are closer genealogical relationships within that larger group. Specifically, I am looking for matches who also share significant DNA with one another outside of the focal segment — suggesting they may be first, second, or third cousins to each other.

To define these sub-clusters, I applied a threshold of 90 cM. Any match sharing 90 cM or more with at least one other match in our list is considered part of a sub-cluster. Not every member must share 90 cM with every other member, but each must meet that threshold with at least one other member in the group. I selected 90 cM because, according to the Shared cM Project 4.0 tool, a 90 cM match has approximately a 99% probability of being fourth cousin or closer, and roughly a 90% probability of falling within the third cousin range. At that level of relatedness, the shared ancestor should typically be identifiable in their trees.

By approaching the data this way, I can work outward from clusters anchored by closer relationships, rather than attempting to analyze each of the 167 matches independently.

Preliminary Application: Sub-Clusters Where Lead Match Matches Steven at 30+ cM

The first cluster, which we will call the Bishop/Hayes cluster consists of fifteen of the remaining matches on our list. The highest cM match to Steven in the cluster is 43 cM. Of those for whom I was able to connect to one another, twelve are documented descendants of William Henry Bishop (1848-1924) of Monroe County, Ohio, and his wife Rosella (Hayes) Bishop (1847-?) of Trumbull County, Ohio. Ten of the matches in this cluster descend through their daughter Sarah Elizabeth (Bishop) Kidder, while two more descend through their daughter Cynthia Elisia (Bishop) Joy. It is unclear at this time whether or how Isaac Holmes is related to either Henry or Rosella.

The second cluster, the Rolf/Depew cluster consists of seven of the remaining matches, the highest being a 34 cM match to Steven. These are descendants of Orlando H. Rolfe (1819-1879) and wife Esther (DePew) Rolfe (1827-1894), both from New York. Five descend through their daughter May Deborah (Rolfe) Sowles, while two descend through their son Anson M. Rolfe. Again, the relationship, if any, to Isaac Holmes is obscure.

The third cluster, knocking out five matches, the highest being a 34 cM match to Steven, are descendants of Lorenzo Dow Thomas (1815-1878) and Sarah Jane Bailey (1817-1880), both of Harlan County, Kentucky. Two descend through son John James Thomas; one through son Taylor Thomas; and one through son James Calhoun Thomas.

Next was a singleton Susan. Of her great-grandparents, the one that stands out is Ida M. (DePue) Ney. While I could not connect her to Esther (Depew) Rolfe, my initial suspicion is that the two surnames Depew and DePue are variants. Ida was born 1860 in Midland County, Michigan, but her direct DePue line hails from Sussex County, New Jersey, which is promising given that Isaac's ancestors were New Jersey colonists.

Next we have a cluster of five matches at 20+ cM to Steven, starting at 33 cM (and like all of the previous clusters also has matches below our somewhat arbitrary 20 cM cut-off). Four of the five descend from Gunelda Elaine (Gassaway) Hall (1914-2006) of Richland County, Ohio, and the fifth does not provide sufficient information to build a tree. Browsing through some of the other shared matches in this cluster below 20 cM, I notice at least one match that descends from Gunelda's uncle Lafayette Gassaway. Therefore, we will name this cluster the Gassaway/Hilton cluster, after their most recent common ancestor pair Thomas Gassaway (1888-1977) of Muskingum County, Ohio, and his wife Catherine (Hilton) Gassaway (1861-1931) of Clark County, Indiana.

The next cluster, having just two matches that share 20 cM or more with Steven, we'll call the Lyons/Lurtz cluster, although since one of the namesakes of the cluster appears to be living, I will refrain from providing further identifying information. The closest match in this cluster is at 32 cM. This is a very interesting cluster, and in the next stages of this project it will probably be worth digging deeper into low cM cluster members, since Mr. Lyons is a documented cousin of Isaac Holmes' wife Elizabeth (McNabb) Holmes via shared ancestors Louis du Bois (1626-1696) and Catherine (Blanchan) du Bois (1627-1713), who emigrated from Wicres, France and settled in Kingston, Ulster County, New York. At this point in the study, it is not apparent whether or not this is coincidental. Mr. Lyons descends through their son Louis (1677) whereas Elizabeth (McNabb) Holmes descends through their daughter Sarah (DuBois) Van Meteren.

Next, we have the Waldron/DePue cluster, with only one representative member sharing 20+ cM with Steven, the match being C.S. at 32 cM. C.S. is a descendant of Isaac Henry Waldron (1837-1916) from Albany, New York, and wife Ruth Ann (DePue) Waldron (1842-1922) of Crawford County, Pennsylvania. He descends through their daughter Julia Myrtle (Waldron) Smith. Another match in the cluster but sharing only 15 cM with Steven descends through Julia's sister Lettie May (Waldron) Howell. Of note, Ruth Ann (DePue) Waldron is a cousin of Ida M. (DePue) Ney. Both are descendants of Daniel DePue (abt 1750) and Annatje (Westbrook) DePue, both from families originating in Kingston, New York. Ruth descends from their son Daniel Westbrook DePue, whereas Ida descends from their son Nicholas DePue.

With the next Standifer/Branham cluster we take eight more matches off the preliminary to-do list. This cluster contains descendants of Andrew Jackson Standifer (1844-1923) and wife Rebecca (Branham) Standifer (1844-1912), both of Hancock County, Tennessee. Two matches descend through daughter Martha Ann (Standifer) Howard. One descends through their son George Clinton Standifer. One descends through their daughter Pearl Bell (Standifer) Cates. The other half of the 20+ cM matches in this cluster do not have sufficient information to determine their exact relation to Andrew and/or Rebecca. No connection to Isaac Holmes or to any of the other clusters is evident.

Our ninth cluster is the Dunbar/Steele cluster, with five 20+ cM shared matches (closest match at 31 cM). The central ancestor pair to this cluster appears to be Robert Dunbar (1801-1870) and wife Mary Elizabeth P. (Steele) Dunbar (1805-1870), both of Amherst County, Virginia. Two matches in the cluster descend through their son Thomas Madison Dunbar (via different wives), and only one of those two matches is a perhaps coincidental cousin of Elizabeth (McNabb) Holmes via shared ancestors John VanMeter (1683-1745) and Margaret (Molenaar)VanMeter (1687-1745). Another in this group, himself sharing 20 cM with Steven, and sharing 90+ cM only with a cluster member having less than 20 cM shared with Steven, is a descendant of a Paul Ogden Wynkoop (1900-1978) but has no identifiable connection to the Dunbar or Steele lines. The Wynkoop family has early ties to the DePue/DePew family, which we will discuss in a subsequent section.

Tenth is the Frease/Saunders cluster of four matches with the closest again at 31 cM. Two of these matches descend from Margaret May (Frease) Conine, daughter of Charles William Frease (1824) of Pennsylvania and Mary Elizabeth (Saunders) Frease (1834-1870) of Virginia. There is a smaller match (13 cM) in this cluster not counted in the four above that descends from Margaret's brother Elizar Eugene Frease. I was unable to find information about the parentage of either of Charles or Mary.

Our eleventh cluster is the last with a head match in the 30+ cM range. Two of the three 20+ cM matches have private trees, and the third has a tree showing only maternal lines. However, one of those maternal lines happens to lead to a Harriet (DePew) Church (1810-1881), who was from Sullivan County, New York (which borders Ulster County). While we're missing half the tree, I don't believe the connection is coincidental.

Making Sense of the Data So Far

We were expecting to see surnames like Holmes, Ashton, Clunn, Lynch, etc., we were expecting locations like Trenton, New Jersey, London, England, or Dublin, Ireland, and instead, after systematically reviewing over one third of the remaining matches (56/167), we're repeatedly seeing the surname DePue/DePew, and it appears this segment came from Kingston, New York. Let's pause from further analyzing new matches and reassess our assumptions in light of the unexpected results.

What if I was mistaken in mapping this segment to Isaac Holmes, and I shouldn't have ruled out his wife Elizabeth (McNabb) Holmes so quickly? Elizabeth's father was a Quaker from Chester, Pennsylvania, but her mother Martha (Shepherd) McNabb had as her mother Elizabeth (Van Meter) Shepherd, of Dutch and French ancestry, whose father John Van Meter (1683-1745) was born in Kingston, New York.

Reliance on Ancestry User Trees

Earlier in this article, I stated: “Fortunately, additional Ancestry matches allow us to narrow the possibilities further. Both Steven and Justine match C.S., a descendant of Jacob Holmes (brother of Isaac and Samuel Holmes), and two descendants of Sarah (Holmes) Osbun, another sibling of Isaac and Samuel: Martha and Stuart.”

Let us first examine the supposed Sarah (Holmes) Osbun connection. While I initially relied on Ancestry member trees indicating that such a person existed and that she was a daughter of Obadiah Holmes and Mary (Clunn) Holmes, that reliance may have been misplaced. WikiTree, a widely used collaborative genealogy database, lists the wife of Samuel Osbun simply as “Sarah Unknown,” indicating that no reliable evidence has been found for the maiden name Holmes. Furthermore, more reliable sources indicate that Obadiah Holmes married Mary Clunn in Trenton during the Christmas season of 1755, whereas Sarah Osbun's gravestone records her birth in 1752—three years before that marriage.

Additional inconsistencies appear in the Ancestry trees themselves which state that Sarah was born in Westchester County, New York, a location not consistent with the documented movements of either Obadiah Holmes or Mary Clunn during that period. A more plausible explanation for the confusion appears to be that one of Sarah Osbun's daughters, also named Sarah, married Jacob Holmes in Cadiz, Ohio. Jacob was indeed a member of the Holmes family—specifically the son of Abraham Holmes, who was himself a son of Obadiah Holmes and Mary Clunn. However, this does not explain the shared DNA with Steven, because the two matches in question (Martha and her son Stuart) are not descendants of Jacob Holmes.

After reviewing Martha's tree and finding no clear connection to Elizabeth McNabb or to any of the other clusters identified so far, I placed Martha and Stuart into a new sub-cluster for later analysis. Martha shares 21 cM with Steven, which determines the cluster's resolution priority.

Jacob Holmes Descendant Matches

While the Osbun descendants can be reasonably dismissed as false Holmes hits, the same cannot be said for the Jacob Holmes descendant that we relied upon earlier when determining whether the segment belonged to the Holmes or McNabb side of the family. I carefully rechecked each generation in the lineage connecting the match C.S. to Jacob Holmes, the brother of Isaac Holmes, and confirm that the documented genealogy is sound. I also verified that C.S.'s DNA match with Steven occurs on the chromosome 22 segment under examination and is not merely a shared match between Steven and Justine on some unrelated segment.

C.S. is not the only descendant of this Holmes line appearing among the matches. A second match, James, descends from James Holmes and shares 18 cM with Steven and 22 cM with Justine. James was not included in the earlier analysis because his match to Steven fell below the 20 cM threshold used at that stage. At this point, however, his presence cannot be ignored. We can be confident that James also matches on the same chromosome 22 segment because he shares matches with representatives of ten of the eleven sub-clusters identified above, including those containing the DePue/DePew lines. C.S. likewise shares matches with eight of those clusters. Neither individual matches representatives from every cluster, likely because their matching segments do not extend across the entire 45 cM span being examined, or possibly due to occasional pile-up effects that create weak false positives. This issue will be revisited later when we examine clusters whose lead matches fall below the 30 cM range.

Importantly, both C.S. and James descend from the couple James Crossley and Alvernia (Bickerstaff) Crossley. C.S. and James are second cousins to one another. Jacob Holmes was Alvernia's second-great-grandfather. Therefore, if the chromosome 22 segment did not originate with Jacob Holmes, the only alternative explanation is that the segment was inherited by C.S. and James through another ancestral line of either James Crossley or Alvernia Bickerstaff. In other words, we would need to demonstrate that C.S. and James share an additional ancestral connection with Steven outside of the Holmes line. The most plausible mechanism would be a form of double-cousin relationship in which another branch of their ancestry intersects with one of the families appearing in the sub-clusters already identified. If such a connection exists, it could explain why descendants of the Crossley/Bickerstaff line appear in the chromosome 22 cluster even if the segment itself did not originate with Jacob Holmes. The fact that only descendants of one of Jacob Holmes' second great-grandparents match on this segment rather than descendants of multiple of Jacob's children, makes this a very realistic possibility which needs to be explored. From this point forward, the primary objective will be to identify stronger evidence explaining the Crossley matches and to determine whether additional data supports or refutes a connection between the DePue/DePew family and Elizabeth (VanMeter) Shepherd.

Re-examining the Dunbar/Steele Cluster

The Dunbar/Steele cluster is the most complex cluster examined so far. Among the matches in this group are three individuals whose documented ancestries point in different directions.

First is Katryn, a documented descendant of the Van Meter family and therefore a distant cousin of Elizabeth (McNabb) Holmes. Katryn also descends from Robert Dunbar (1801–1870) and his wife Mary Elizabeth P. (Steele) Dunbar (1805–1870), although the Dunbar/Steele line itself does not appear to have any known connection to Steven's ancestry.

Second is Miles, who also descends from Robert and Mary (Steele) Dunbar but does not descend from Katryn's documented Van Meter line.

Third is John, a documented descendant of the Wynkoop family, whose ancestry leads to the Dutch lines from which the DePue/DePew families of Kingston, New York ultimately descend. However, John's tree does not appear to overlap either Katryn's Van Meter line or the Dunbar/Steele line shared by Katryn and Miles.

Because these matches connect to three different ancestral networks—Van Meter, Dunbar/Steele, and Wynkoop—the structure of the cluster is ambiguous. To obtain additional data that may clarify these relationships, the threshold for membership in this cluster will be temporarily reduced from 100 cM to 45 cM, allowing additional matches to be included in the analysis in hopes of revealing a clearer pattern of shared ancestry.

My working hypothesis is that a clearer structure will emerge showing the Van Meter connection as the common source of DNA inheritance across this cluster, with the Dunbar/Steele connection between Katryn and Miles being coincidental rather than causative. Additional insight into the relevance of John's Wynkoop line would further strengthen the analysis.

To explore this possibility, the next stage of analysis will rely heavily on Ancestry's Pro Tools, which provide access to the amount of DNA shared between the matches themselves. By lowering the cluster membership threshold from 100 cM to 45 cM, we can incorporate additional matches and construct a chart showing how the matches cluster based on their shared cM with one another, rather than simply their shared DNA with Steven.

For purposes of identifying meaningful structure within the cluster, shared matches below 45 cM will be treated as noise, since the lead match in this cluster shares only 31 cM with Steven on chromosome 22. The goal is to determine whether stronger relationships among the matches reveal a coherent pattern that helps distinguish the true ancestral source of the segment from coincidental overlaps in the trees. Where matching donors have also tested their descendants, those descendant matches are omitted from the analysis to simplify the resulting diagram.

After performing the clustering and visually depicting the results (see above diagram), it becomes clear that my initial hypothesis was incorrect. Not only do additional matches appear who descend from Robert Dunbar and Mary (Steele) Dunbar and cluster closely with Katryn and Miles, but the structure that emerges also reveals a previously unrecognized sub-cluster consisting of descendants of Sarah (Steele) Mottesheard, sister of Mary (Steele) Dunbar.

This pattern strongly reinforces the Steele family connection within the cluster. The presence of descendants from two Steele sisters indicates that the shared DNA likely traces through that family rather than through the alternative lines previously considered. Consequently, Katryn's documented Van Meter ancestry and John's Wynkoop ancestry must be treated as coincidental in the context of this segment, while the Dunbar/Steele connection is now supported by the clustering evidence.

Based on this procedure, the extended Dunbar/Steele cluster resolves into five distinct sub-clusters. The first corresponds to the original Dunbar/Steele group identified before the membership threshold was lowered to 45 cM. Members of the second sub-cluster appear to descend from John Abbott Independence Lee (b. 1838) and his wife Phoebe (Williams) Lee (b. 1836), both of Craig County, Virginia. The third consists of a single match, Marcie, who does not currently have a tree. The fourth sub-cluster is composed of descendants of Sarah (Steele) Mottesheard, sister of Mary (Steele) Dunbar. The fifth contains three matches, only one of whom has a tree, and that individual descends from a John Shepard Clark, born around 1805 in what is now West Virginia.

The Steele sisters—Mary (Steele) Dunbar and Sarah (Steele) Mottesheard—were daughters of Robert H. Steele (b. 1778) and Letitia (Champ) Steele (1781–1850) of Monroe County, Virginia (now West Virginia), a county directly adjacent to Craig County, Virginia, where the Lee family resided. The fact that the only match in the fifth sub-cluster with a tree has a distant ancestor named John Shepard Clark is noteworthy because Steven's ancestor Elizabeth (McNabb) Holmes was the daughter of George McNabb and Martha (Shepherd) McNabb of Shepherdstown, West Virginia—a town named for our ancestral Shepherd family.

While Shepherdstown lies nearly 400 miles north of Craig and Monroe Counties along the Shenandoah Valley, and the middle name "Shepard" alone would ordinarily be too tenuous to pursue, the connection becomes more plausible when considering that Martha (Shepherd) McNabb's parents, Thomas Shepherd and Elizabeth (Van Meter) Shepherd, were married in Winchester, Virginia in 1733, at a time when both the Van Meter and Shepherd families were already established in the Shenandoah Valley. By contrast, the Holmes and Clunn families are not known to have settled in the Shenandoah Valley during this period.

Based on the refined clustering analysis, my revised hypothesis is that the extended Dunbar/Steele cluster—hereafter referred to as the Steele/Champ cluster to reflect the deeper ancestral couple—represents either descendants of Thomas Shepherd and Elizabeth (Van Meter) Shepherd as a couple, or descendants of a sibling of either Thomas or Elizabeth (most likely descendants of a sibling of Elizabeth). As we continue to analyze Steven's chromosome 22 matches, we will hopefully refine the hypothesis.

Weekend Analysis: Merging the DePuy/Wynkoop Descendant Clusters

Over the weekend of March 7–8, 2026, I continued the analysis by performing additional clustering exercises on the chromosome 22 dataset and expanding the family trees of many matches. In the previous section, we were able to expand the Dunbar/Steele cluster by lowering the clustering threshold, producing a consolidated Steele/Champ super-cluster that added additional members and effectively moved the analysis one generation closer to the unknown most recent common ancestor responsible for the segment. For this weekend's project, I decided to form a super-cluster using a completely different method.

Among the eleven clusters identified so far, four already contained members connected to the DePew/DePue family: the Rolfe/DePew cluster, the Waldron/DePue cluster, the Church/DePew cluster, and a singleton match (Susan) who descends from Ida (DePue) Ney. While reviewing and expanding the trees of additional lower-cM chromosome 22 matches during the weekend, I identified many more descendants of Moses DePuy and Maria Cornelia (Wynkoop) DePuy, who appear to be the progenitors of the New York DePew/DePue family. Having concluded that either Moses DePuy, Maria Cornelia Wynkoop, or both are likely immediate family of the ancestor of Mary Holmes who contributed the chromosome 22 segment under investigation, I decided to create a new analytical grouping.

With that working assumption, I created a new tag group in Ancestry for all matches on this segment who have a documented DePew/DePue ancestor, along with any additional matches who cluster with them using a 90 cM shared-match threshold. This tagging step allowed the DePuy-related matches to be tracked as a unified analytical group as the clustering work continued. The resulting super-cluster currently consists of thirty-four matches, of which sixteen match Steven at 20+ cM.

Creation of Cluster Cards

The next step in our analysis is organizational. I created "cards" that represent the clusters of matches we have identified so far. One of these is a cluster of documented cousins of Elizabeth (Van Meter) Shepherd. A second is the DePuy/Wynkoop super-cluster, which we believe connects to our tree through one of Elizabeth (Van Meter) Shepherd's ancestral lines from Kingston, New York. The remaining seven clusters are primarily Appalachian in their geographic focus; we hypothesize that some of them may consist of descendants of Thomas Shepherd and Elizabeth (Van Meter) Shepherd as a couple.

Continuing Cluster Analysis: Clusters with Lead Matches Between 25 and 29 cM (Part 1)

The next task was to continue working downward through the more distant match clusters, beginning with those whose lead matches fall in the 25–29 cM range. After expanding the trees of several matches in the first cluster within this range, I believe the study reached a turning point. I was able to place several of these matches directly into my own tree, as they are descendants of Thomas Shepherd and Elizabeth (Van Meter) Shepherd, from a branch of the family that settled in Kentucky.

This is significant because it provides our first documented descendants of a sibling of Martha (Shepherd) McNabb among the matches on this chromosome 22 segment, just as we have already identified descendants of siblings of Elizabeth (McNabb) Holmes, Mary (Holmes) Price, and others among the clusters examined earlier. Furthermore, this discovery is consistent with the growing body of circumstantial evidence suggesting that the segment passed through Elizabeth (Van Meter) Shepherd. Because these matches could be directly connected within my existing tree, I did not create a new cluster card for them and instead simply added the matches and their intermediate ancestors directly to my tree, linking them to the Shepherd/Van Meter family.

Next, at 28 cM, was a singleton match from a donor named Colleen with no tree. I did a quick public records search only to find there are dozens of people with her name living in the United States. Since her closest shared match to anyone else on this segment was 36 cM, I simply applied a suspected Van Meter cousin tag and moved on.

The next cluster, the Vail/Yorke cluster, is documented in a card below.

The next cluster, whose lead match shares 27 cM with Steven, consists of descendants of François de Puy, uncle of the Moses DePuy discussed earlier. This finding suggests that the connection to Steven is through the DePuy line rather than the Wynkoop line. The DePuy family originated in the historical Artois region of France, whereas the Wynkoop family came from the Netherlands, which further points to Artois as the likely European homeland of the chromosome segment under study.

This observation is also consistent with the working hypothesis that the segment passed to Steven through Elizabeth (Van Meter) Shepherd. Elizabeth's paternal grandmother, Sarah (DuBois) Van Meter, who is well known in colonial history for her capture by the Minisink Indians, lived in Kingston, New York, but both of her parents were originally from the Artois region of France.

Rather than create a separate cluster card for this group, I incorporated its members into the existing DePuy cluster (dropping the Wynkoop name from the card), applying the same DePuy tag used earlier in the analysis. With the addition of these matches, the DePuy super-cluster now contains 39 members, 18 of whom share 20 cM or more with Steven.

The next cluster, the Monk/Brown cluster, is documented in a card below.

Next was a singleton, Duston, for whom I was able to construct a tree leading six generations in every direction, but no connection to Elizabeth Van Meter or any of the other clusters was obvious. His 27 cM match, per Ancestry, was split across two segments, so he may simply be a false positive match. I gave him the tag indicating likely Van Meter side and moved on.

Next was another singleton, Richard, with a 27 cM match, no public tree, and insufficient information to create one.

The next cluster, with a lead match of 27 cM, consisted of just two matches, only one with a tree. I expanded the tree but found no familiar surnames or places. The donor's mother's family has French-Canadian roots.

The next cluster, with a lead match at 26 cM, only had one member with enough identifying information to build a tree, but I was not able to connect it to Elizabeth Van Meter or any of the other clusters.

The next cluster, with a lead match of 26 cM, consists of two second cousins, both descending multiple times from Elizabeth Van Meter's du Bois and Blanchan lines from Artois, France, thus sealing the deal on the origin of this segment. I simply connected them to the family tree rather than creating a cluster card.

Taking a Break From Clustering to Explore Elizabeth Van Meter's Ancestors

At this stage of the analysis, it is useful to briefly step away from the clustering results and examine the pedigree of Elizabeth (Van Meter) Shepherd herself. Elizabeth was the daughter of John Van Meter and Margaret (Molenaar) Van Meter. Through her father, she descends from the Van Meteren family of Gelderland in the Netherlands and from the French Huguenot couple Louis du Bois and Catherine (Blanchan) du Bois, whose families originated in or around the Artois region of France before settling in Kingston, New York. Through her mother Margaret Molenaar, Elizabeth also traces to the Molenaar family of New York City, as well as to the Hay and Capoens families of early New Netherland. The pedigree shown below summarizes these ancestral lines and provides the framework for the next stages of our analysis.

Early in this study, we determined that the entire span of chromosome 22 that Steven shares with Justine was inherited from one of Mary (Holmes) Price's great-grandparents. Through the subsequent examination of matches across the full span of the segment, the weight of the evidence now strongly indicates that the great-grandparent in question was Elizabeth (Van Meter) Shepherd. This means that all 45 cM of the segment must have been present in Elizabeth's genome. We have not yet definitively identified descendants of Elizabeth's siblings among the matches, though the strongest candidates so far may be members of the Bishop/Hayes cluster, where matches to Steven reach as high as 43 cM. If that cluster ultimately proves to descend from one of Elizabeth's siblings, the connection is not currently visible in the available trees, and may reflect an error or omission somewhere in the pedigree of either William Henry Bishop or Rosella (Hayes) Bishop.

We have likewise not yet identified any matches who clearly descend from siblings of Elizabeth's father, John Van Meter. Such individuals would share ancestry through both the Van Meter and Molenaar lines, and remain an important category to watch for as the clustering work continues. One difficulty is that the husbands of the daughters of Joost van Meteren and Sarah (DuBois) Van Meter are not presently known to me, which makes recognizing their descendants considerably more challenging. It is therefore quite possible that descendants of those lines are already present among the remaining unexplained clusters, but are currently hiding in plain sight.

Based on the data gathered so far, we can establish a minimum boundary for the portion of the segment that must have been present in Sarah (DuBois) Van Meter. The strongest indicator is the 32 cM lead match in the Lyons/Lurtz cluster, which is shared with a documented descendant of Sarah's brother, Louis du Bois. Because siblings Sarah and Louis share this 32 cM sub-span, it necessarily follows that this portion of the segment was inherited from one of their parents and ultimately from one of their grandparents. In other words, at least 32 cM of the chromosome segment must trace back to one of Sarah's grandparental lines. While the estimated age of the full 45 cM segment suggested a meiosis event occurring around 1715, this evidence indicates that the 32 cM sub-span itself has survived without recombination since roughly 1600, approximately the generation in which Sarah's grandparents were born.

A Little Math: Getting from 45 cM to 32 cM

For a 45 cM span, the standard model is that recombination events across that interval in a single meiosis follow a Poisson distribution with parameter λ = 0.45 because 45 cM = 0.45 Morgans.

For one meiosis, the probabilities are:

So in any single generation, a 45 cM segment has only about a 7.5% chance of experiencing two or more recombinations. That is why it is usually reasonable to think in terms of zero or one recombination across a span this size.

Stepping back two generations

Between Sarah (DuBois) Van Meter and Elizabeth (Van Meter) Shepherd, there are two meioses:

• Sarah/Joost → John Van Meter
• John/Margaret → Elizabeth Van Meter

Across two meioses, the expected number of crossovers is λ = 0.9 and the probabilities become:

So even across two generations, there is still about a 77.3% chance that the 45 cM span underwent zero or one recombination only. That makes complicated multiple-breakpoint scenarios the minority outcome.

What that means for our segment

Given the 32 cM minimum sub-span that we can attribute to Sarah (DuBois) Van Meter, the simplest possibilities are:

1. 45 cM Sarah DuBois. No recombination occurred in the two-generation line from Sarah to Elizabeth, and the observed reduction from 45 cM to 32 cM occurred instead somewhere in the line of descent from Louis du Bois to Mr. Lyons. Probability: P(0 recomb over two meioses) ≈ 40.7%
2. 32 cM Sarah DuBois / 13 cM Joost van Meteren. Exactly one recombination occurred in the first meiosis, when Sarah and Joost's son John Van Meter was formed, and no recombination occurred in the second meiosis. Probability: P(1 recomb in meiosis 1, 0 in meiosis 2) = 0.287 × 0.638 ≈ 18.3%
3. 32 cM Sarah DuBois / 13 cM Margaret Molenaar. Exactly one recombination occurred in the second meiosis, when Elizabeth Van Meter was formed from John Van Meter and Margaret Molenaar, and no recombination occurred in the first meiosis. Probability: P(0 in meiosis 1, 1 in meiosis 2) = 0.638 × 0.287 ≈ 18.3%

Summary of the three main scenarios

If we restrict ourselves to the simple, likely cases of zero or one recombination across the two generations, the probabilities are:

The remaining 22.7% represents scenarios with two or more recombinations, which would create more complicated mosaics.

In conclusion, because a 45 cM segment has only about a 22.7% chance of experiencing two or more recombinations across two generations, the most likely explanations are the simpler ones: either the full 45 cM segment was present in Sarah (DuBois) Van Meter, with the shortening occurring somewhere in the line of descent to Mr. Lyons, or the segment was shortened by a single recombination event in the Sarah-to-Elizabeth line, producing a 32 cM Sarah (DuBois) portion combined with a 13 cM portion from either Joost van Meteren or Margaret Molenaar.

At this stage of the study, it is unlikely that continued clustering analysis alone will allow us to determine which of these scenarios is correct. One possibility is that additional evidence may emerge if one or more of the yet-unanalyzed clusters helps resolve the Bishop/Hayes cluster, whose 43 cM match has the potential to rule out certain scenarios depending on its ultimate placement. Otherwise, it is possible that we may eventually identify 13 cM matches to either the Margaret Molenaar or Joost van Meteren lines among Steven's matches. Such matches, however, would not appear through the clustering methods used here, because a match sharing only 13 cM on this segment would not register as a shared match under Ancestry's 20 cM shared match threshold unless it also matched Justine at 20 cM or greater. Even if a descendant of one of those lines were identified among Steven's smaller matches, we would not know from Ancestry alone whether the match occurred on chromosome 22 or elsewhere in the genome (because we would not know whether they also match Justine). Confirming such a connection would therefore require examining segment-level data on platforms such as MyHeritage or FamilyTreeDNA, where it would be possible to determine whether a 13 cM match falls on either end of the chromosome 22 span under study.

For the moment, therefore, the clustering analysis will continue. As additional clusters are examined, we will look for evidence that might clarify the Bishop/Hayes connection, such as clusters descending exclusively from either the Bishop or Hayes families, or for matches descending from siblings of Elizabeth (Van Meter) Shepherd or her father John Van Meter, which would further strengthen the conclusions reached thus far.

Continuing Cluster Analysis: Clusters with Lead Matches Between 25 and 29 cM (Part 2)

The next cluster, with a lead match at 26 cM, had two members. After expanding the tree of one of them, I discovered that they were descendants of Samuel and Mary (McNabb) Holmes through their daughter Natalie.

The next cluster, with a lead match at 26 cM, was comprised of descendants of Sarah's brother David DuBois, also of Kingston, New York.

Next, also at 26 cM, was Angela, a singleton with no tree.

After examining our next cluster, with a lead match at 26 cM, we now are able to drill our Bishop/Hayes cluster back two generations further, as this new cluster consists of descendants of William Henry Bishop's grandparents Jacob and Sarah (Salkeld) Smith. I created an updated cluster card below.

The next cluster, with a lead of 25 cM consisted of closely related donor matches. See the Stocker/Wagner card below.

The next match was a singleton, B.R., with a 25 cM lead and no public tree.

The next cluster, with a 25 cM lead match, consisted of a mother and son, both descending from William Henry Kuykendall (b. 1837, Bradford County, Pennsylvania). While I was not able to discern William's parents, the genetic connection to Steven strongly suggests he must be a descendant of Hendrick Kuykendall (b. 1720, Port Jervis, NY) and Elizabeth Cool (b. 1716, Kingston, NY), as Elizabeth (Cool) Kuykendall is known to be a descendant of Sarah DuBois' brother David DuBois—which would explain the match. No card was created for this cluster, but instead a DuBois descendant tag was added to the matches in Ancestry.

The final cluster in the range, having a 25 cM lead, wound up being descendants of Moses DePuy, so I simply added it to the existing DePuy super-cluster, which now has forty-six matches, twenty-two in the 20+ cM range.

A Serendipitous Finding on One of Steven's Other Holmes Segments

While reviewing Steven's matches, I noticed a familiar name from having previously examined my own DNA results: a cousin named Jay, whom I match on one of my own Holmes segments inherited from Margaret (Molenaar) Van Meter. Jay descends from Janetje (Molenaar) de la Grange, Margaret's half sister. Curious whether this connection might shed additional light on the segment under study, I examined Jay's shared match list with Steven. Among those shared matches were several names that were already familiar from the clustering analysis of Steven's chromosome 22 du Bois segment.

If this overlap is not coincidental, it strongly suggests that these individuals descend from both the du Bois and Molenaar families. The most straightforward explanation would be that they are descendants of John Van Meter and Margaret (Molenaar) Van Meter, meaning they descend from siblings of Elizabeth (Van Meter) Shepherd. Jay himself matches Steven at 19 cM, which indicates that the shared segment between them cannot be the chromosome 22 segment examined earlier, since even under the most generous interpretation only about 13 cM of that segment could plausibly derive from the Molenaar side. Instead, Jay's shared match list provides a useful reference set for identifying matches who carry Molenaar DNA. By examining Steven and Jay's shared matches, it may be possible to determine whether some of the clusters or super-clusters that remain unexplained in the chromosome 22 analysis can be accounted for by the presence of Molenaar ancestry among their members. To explore this possibility further, I compiled a chart of Steven and Jay's shared matches down to 20 cM, which is presented below, in order to determine whether these matches represent Molenaar-only descendants or individuals who also match Justine, in which case they would likely descend from the Van Meter/Molenaar couple rather than the Molenaar line alone.

We have not yet discussed the Riggle/Jenner cluster, because its lead match is only 23 cM. However, I know from both Steven's Ancestry and MyHeritage match lists that descendants of the Riggle/Jenner family match Steven on chromosome 22. In addition, three more Riggle/Jenner descendants appear among Jay's shared matches below the 20 cM threshold shown in the chart, suggesting that members of this family cluster with Steven on more than one segment.

When examining Steven's shared matches with James (the hypothesized Molenaar-only match listed above), we find that he also matches the McNabb/Holmes, Gassaway/Hilton, and Riggle/Jenner descendants appearing in the chart. This indicates that each of these individuals also matches Jay on the Molenaar segment. Unfortunately, the same cannot be said with certainty for Paula. Although Paula appears among Jay's shared matches, the evidence suggests that she matches Jay on a different segment that neither Steven nor I inherited, which may or may not derive from a common Molenaar ancestor.

For these reasons, the matches in the Gassaway/Hilton and Riggle/Jenner clusters can be confidently classified as descendants of John Van Meter and Margaret (Molenaar) Van Meter. The evidence for the Smith/Salkeld super-cluster is somewhat less direct, since Paula is its only representative among Jay's shared matches. Nevertheless, given that every other shared match above 20 cM in this comparison (with the exception of James) can be explained through the Van Meter/Molenaar line, it remains highly plausible that members of the Smith/Salkeld super-cluster also descend from this couple.

There is also one other match in common with Steven, Jay and James named Michael, who only matches Steven at 14 cM. However, he is a member of the Wood/Busse cluster on chromosome 22 that we haven't yet discussed since the lead match shares only 23 cM with Steven.

Continuing Cluster Analysis: Clusters with Lead Matches Between 23 and 24 cM

I undertook the job of building/extending trees for many of the matches at 24 cM over the weekend of March 14–15, 2026. The first cluster, having a lead match of 24 cM, consisted of a closely related family group and I was not able to connect them to any of the clusters we've examined thus far. I did not make a card for this sub-cluster for privacy reasons, but the family is from Monroe County, Ohio, and surnames include Nalley, Davis, Bates, Dougherty, Davis, Luke, Groves, and Hunter.

The next cluster consists of more descendants of Moses DePuy, descending through Nicolas and Weyntjen (Roosa) DePuy. They were added to the DePuy super-cluster.

For the next cluster, I have added a new Wert/Magnan card below. The matches have ancestors from Westchester, New York that probably descend from either the du Bois or de Puy family, although the connection is not yet clear.

Next was a singleton match with insufficient information provided to create a tree.

The next cluster consisted of just two matches, neither of which provided sufficient information to build trees.

The final 24 cM cluster consisted of three closely related matches, one of whom, Priscilla, had a public tree that I was able to extend. Upon doing so, I found that she descends from Abraham Hays and Sarah Tuttle, who are the maternal grandparents of Thomas Gassaway (see the Gassaway/Hilton card). Based on this connection, I reclassified the cluster as part of the Hays/Tuttle super-cluster and created an updated card below. As established earlier, these matches ultimately descend from John Van Meter and Margaret (Molenaar) Van Meter, and this new information moves the analysis one generation closer to identifying the precise point of connection. The most likely avenue for further progress lies on the Hays side, where the pedigree remains less developed than the Tuttle line and therefore presents greater opportunity for uncovering the missing link.

As I broke into the clusters with lead matches of 23 cM, I encountered several in a row where the matches descended from DePuy lines and I added them to the DePuy super-cluster, now having 58 total matches, 30 of which share 20 cM or more with Steven.

Then, there was a treeless singleton David, followed by a cluster of matches from Harlan County, Kentucky, for which I made a Pace/Fields card below.

Next came the Harrison cluster. The title of this cluster has only one surname because the maiden name of Rollin Harrison's wife Clarissa is not known. She is from Bradford County, Pennsylvania though, and we know that there was a branch of the Kuykendall family living in that county who were DuBois descendants. Further research is required, but for now I added a card below.

The next cluster of matches descend from a George and Sarah Meadows of Jones County, NC. See below for cluster card.

Next was the Wood/Busse cluster mentioned earlier. Little is known about the progenitors of this cluster other than that the couple lived in Zanesville, Ohio. However, recall that the context we've seen this cluster before is that the matches in this cluster are Van Meter and Molenaar descendants, as the cluster intersected the shared matches with known Molenaar cousin Jay. Possibly relevant to the connection, John and Margaret (Molenaar) Van Meter had a daughter Magdalena who married a Robert Pewsey. I have not attempted to trace the descendants of this line. I am pointing this out because Busse and Pewsey could theoretically be variant spellings of the same surname. This is an area for further research.

Next, was a singleton that I was able to trace to a DePuy line.

Then, there was a small cluster with no obvious connection to any of the families we discussed thus far. For privacy reasons, I did not create a card for this cluster.

The following cluster, I traced back to David DuBois via a Kuykendall line. This is further evidence in support of our hypothesis that the segment in question was passed down through the Kuykendalls to the Harrison cluster, as we now know that multiple Kuykendall branches passed this segment down to modern-day descendants.

Next was a singleton Kenneth whose tree showed no obvious connections. Then Lynn, another singleton with ancestors from Kingston, New York, although I could not find a direct link to the DuBois or DePuy families.

The last cluster at the 23 cM level differs from the others in that its members look like immediate family to one another, not a spread of distant cousins around one historical couple. A normal cluster card would have to spell out recent common ancestors and could edge into identifying people who may still be living, so I did not publish one for that group. Without going into those recent lines, everyone in the cluster still appears to descend from a Daniel Riggle born 1784 in Washington County, Pennsylvania, who later resided in Tuscarawas County, Ohio. That detail matters for a possible tie to the Riggle/Jenner cluster briefly mentioned in the section on clusters overlapping the Molenaar segment; when we take up Riggle/Jenner, we can weigh whether both groups descend from the same Riggle line.

A New Riggle Super-Cluster

We still have several clusters to walk through in descending order before we reach the Riggle/Jenner cluster, but our first cluster with a 22 cM lead—the Ohler/Newton cluster—already ties into the Riggle family as well. Because the relationships among these groups are somewhat tangled, it makes sense to discuss them now.

Ranson Newton (1809–1875) married twice: first to Elizabeth Price and second to Sarah Riggle. Elizabeth belongs to the same Price family as my ancestor Rezin Price, though that point is not relevant here. Ranson had a daughter by each wife—Cyrena Newton (b. 1828) with Elizabeth Price, and Amadilla Newton (b. 1853) with Sarah Riggle.

Descendants of Amadilla belong to the Ohler/Newton cluster (named for Amadilla and her husband James Neely Ohler). Descendants of Cyrena belong to what I have been calling the Riggle/Jenner cluster: Cyrena married John Wesley Riggle, and the Riggle/Jenner members appear to be descendants of Cyrena’s son John Wesley Riggle II and his wife Mary Adelaid Jenner.

Ordinarily, when descendants of two half-sisters match a tester’s kit, I assume the shared DNA comes through their common parent—in this case Ranson Newton. Here, however, both lines also run through the Riggles (the older daughter married a Riggle, and the younger daughter’s mother was a Riggle), so that shortcut is not safe.

The cluster tied to Daniel Riggle (b. 1784) in the previous section—the last 23 cM group discussed before the cluster cards—further suggests the signal may ride the Riggle side rather than the Newton side. That reading is strengthened because Amadilla’s maternal grandfather was the same Daniel Riggle.

How Daniel Riggle connects to Cyrena’s husband John Wesley Riggle is still unclear. John’s father appears to have been Daniel A. Riggle, b. 1793, also in Washington County, Pennsylvania. Because John Wesley was already married by the 1850 census and does not appear in his parents’ household, published trees may be assigning him to the wrong Daniel Riggle.

For this discussion, Ranson Newton’s ancestry traces back to New England, and I see no plausible connection to the Van Meter family of the Appalachian valley.

Because I am fairly confident the thread runs through the Riggle family, I have added a Riggle super-cluster card at the bottom of the next section that surveys clusters with lead matches between 21 and 22 cM.

Continuing Cluster Analysis: Clusters with Lead Matches Between 21 and 22 cM

I take up the 22 cM end of this band first: each of the next three clusters had a lead match sharing 22 cM with Steven.

The first cluster is represented in the Neff/McClung card below. William Taylor Neff and Margaret Catherine McClung were both of Fayette County, West Virginia. I found no obvious connection to any of the other families we've discussed thus far.

The second cluster is represented in the Bryant/Mings card below. The subject family was midwestern with roots in Iowa and Minnesota. I found no obvious connection to any of the other families we've discussed thus far.

The third cluster was another tight cluster, and for privacy reasons I did not create a card. I did not find any connection to any of the other families we've discussed thus far.

Next we move on to clusters whose lead matches fall at 21 cM.

The first cluster in the 21 cM range is one we already briefly discussed, i.e. the mother and son who descend from Samuel Osbun, who was mistakenly connected to the Holmes family. Once again, the cluster is very small, so I did not create a card, but any connection to the families we've explored is obscure.

The next cluster was dominated by private trees. I attempted to create trees for two of the individual matches who had enough information to be identified, but I did not find an intersection, and therefore no card could be created for this cluster.

Next was J., a singleton with no tree.

The final cluster in the 21 cM range had insufficient information from which to build trees.

2026 ECGGC Conference

At this point in the study, it is March 26, 2026, and it happens to be the deadline for submitting speaking proposals for this year's East Coast Genetic Genealogy Conference, which will be taking place in Boston. As this article is technically a continuation of a previous ECGGC talk, I put together a proposal for a talk based on the study.

I proposed a session titled Mining a Chromosome 22 Pile-Up: Clustering Through a Documented Colonial Lineage, framed as a practical walkthrough of the same workflow documented here—defining the segment, organizing Ancestry matches into clusters and super-clusters, building selective trees and cluster cards, and closing with how to think about a crowded match list versus genuine pile-up—so the audience can reuse the steps on their own chromosomes.