Parents for Florida Pioneer, Seth Howard: Part 2 - Gathering DNA Matches
Summary of the Project Through Step 1
So far in the process, I have extensively studied the documented life of Seth Howard in Florida. This did not seem to lead to any clues regarding his birth-family. I decided to implement autosomal DNA (atDNA) into my analysis to help generate some leads.
DNA Test-Taker
I needed to determine who was the best person to take an autosomal DNA (atDNA) test to help answer my research question. I had multiple kits to chose from, including my own, but using my Great-Uncle David’s match list (he will just be called “Uncle David” from here on out) was the best option since he was the closest generationally to my research subject, Seth Howard.
Descendancy chart created using diagrams.net
Here is my Uncle David along with his mom, my great-grandmother, Granny Irene, through whom we descend from Seth Howard.
From left to right: Aunt Mildred, Uncle David, Uncle Howell, Granny Irene, and R.S. Date unknown, although likely the early to mid-1980s. Digital picture in researcher’s personal collection. The image above was lightly edited for clarity and improvement using Nano Banana in Google Ai Studio.
Starting Point: Previous DNA Project
My jumping off point was a previous research project which I have discussed before. I confirmed that 11 DNA matches including my great-uncle David were descendants of Seth Howard and Harriet Weeks.
The line of descent from Seth Howard and Harriet C. Weeks for eleven test-takers is shown in blue and yellow. The test-taker names were anonymized to protect their privacy. The centimorgan (cM) values shared between David Gordon Montgomery (denoted as “MMH1”) and each documented test-taker are shown in the blue rectangles. The cM values shared between Mary Romeo (1942–2017) (denoted as “LVH1”) and each documented test-taker are shown in the yellow rectangles.
Additionally, after this project had wrapped up, I found 29 more confirmed descendants of Seth and Harriet for a total of 40 matches. Let’s refer to this group as the “Howard-Weeks Cluster”. It served as the basis for searching for 4th to 5th cousin matches who could potentially descend from the siblings of Seth Howard or the siblings of his parents.
Organizing DNA Matches Visually
To make sense of Uncle David’s DNA matches, I needed a way to group them that made analysis useful and easier to see. I tested out many different tools and auto clusters worked best for me.
When working with autosomal DNA results, one of the most powerful strategies for identifying genetic relationships is to analyze shared matches which are the individuals who appear on more than one person’s match list. To assist with this process, the tool DNAGedcom offers a feature that allows users to download "In Common With" (ICW) files from testing services such as AncestryDNA.
An ICW file identifies groups of matches who are “in common with” one another, meaning they all share DNA with the original test-taker and with each other. The ICW file shows which matches cluster together, making it possible to infer genetic networks that may correspond to particular branches of a family tree.
For example, if Match A appears on the test-taker’s match list and Match B is listed in the ICW file for Match A, this indicates that both A and B share DNA with the test-taker, and that they are also matched to each other on the platform. This can be a strong indicator that these individuals may descend from the same ancestor or ancestral couple.
Researchers use ICW data to form genetic clusters for a test-taker. These clusters are then examined for patterns such as repeated surnames in family trees, geographic overlaps, relationship predictions based on shared cM values, and known cousin relationships that help assign the cluster to either the maternal or paternal side of the family.
While ICW clustering is an excellent hypothesis-generation tool, it is not conclusive on its own. Because services like AncestryDNA do not provide segment data, ICW connections cannot confirm that matches share the same segment of DNA. Therefore, ICW clusters should be used in conjunction with family trees, documentary evidence, and other DNA tools to build reliable genealogical conclusions.
Gathering from DNAGedcom
Because no hypothesized parents existed for Seth Howard at the beginning of this project, the “Howard-Weeks” group of confirmed descendants noted earlier was used to create an ICW file using DNA Gedcom. The tool also created a Match file which is foundational for analyzing DNA matches and building cluster hypotheses. It is used in conjunction with the ICW files to form the genetic networks.
My Great Uncle David’s pedigree is deeply colonial on both his maternal and paternal side. When I downloaded his matches out of ancestry at the time of this original project, he had over 86K matches. I just checked again and he now has over 100K matches in the ancestry database. To save time and effort, instead of downloading all of his matches, I used the Howard-Weeks cluster identified to “fish” for the matches that would be most useful for this project. I only downloaded Uncle David’s matches if they met two criteria:
1. I wanted the sweet spot of 4th to 5th cousin range, so they had to share between 10 and 50 cM with Uncle David.
2. To be included in the download, they had to be a shared match with at least one member of the Howard-Weeks cluster.
This methodology gave me a much more manageable 3,347 matches to work with.
Creating Genetic Clusters
The next step was to run a “Collin-Leeds Method” (CLM) clustering of the matches extracted above. CLM clustering works by identifying those matches who are in common with each other and with the test-taker, and then grouping those matches into clusters when multiple people appear in one another’s ICW lists. Finally, the program color-codes each cluster, which often corresponds to a single ancestral line.
The algorithm essentially says: If Match A shares DNA with Match B, and Match B shares DNA with Match C, and they all match the test-taker, they likely come from the same ancestor. This method builds clusters without segment data, making it ideal for AncestryDNA users.
The application of the Collin-Leeds Method to autosomal DNA results using data exported from DNAGedcom produced approximately twenty-nine distinct clusters, each containing three or more individuals.
DNA GEDcom, Collin-Leeds Method Tool, “Shared Clusters for David Gordon Montgomery 8cM - 50 cM”.
A manual review of each cluster was conducted to identify potential common ancestors. Many of the clusters could be clearly traced to the known ancestral couple Seth Howard and Harriet C. Weeks, indicating that those matches were genetically tied to the documented direct line. These clusters were excluded from further analysis since they did not provide new information relevant to the research goal.
Among the remaining clusters, two groups stood out following comparison of user-provided family trees. These were designated as Cluster 2 and Cluster 303 in the output, and appeared to contain matches who descended from either:
A woman named Appy Howard
A woman named Mary Ann Howard
Or a man named Aaron Howard
Of these three individuals, documentary evidence could only be located for Appy and Mary Ann Howard. By the time of the 1850 U.S. Census, both women were enumerated in Wayne County, Georgia, living with their respective husbands. This is notable because Wayne County lies just two counties north of the Florida-Georgia state line, making it a plausible origin location for family members who later settled in Florida, such as Seth Howard.
Other than the name “Aaron Howard” listed as the father on the death certificate of Mary (Howard) Thornton’s oldest child, no additional definitive records were found for Aaron Howard, and his identity remains unverified. However, the geographic proximity of Appy and Mary Ann Howard to the Florida line, combined with the shared surname and appearance in DNA clusters, presents a compelling lead. These findings suggest a possible collateral line or siblingship with Seth Howard that warrants further research.
Shown below are the two relevant DNA clusters.These clusters may represent shared genetic ancestry that predates or parallels the known descent from Seth Howard and Harriet C. Weeks, and as such, they serve as key targets for ongoing documentary and genetic investigation. For the purpose of this research, these two clusters were combined into one supercluster, as all the included DNA matches descend from Mary Ann and/or Appy Howard. This group will be referred to as the 'Howard Sisters Cluster' going forward, because the evidence presented will support that they were full biological siblings.
Looking Ahead
In the next post in this series, I will explore pedigree triangulation and we address the Aaron Howard problem head-on.
