Re-examining the origins of William Kay
Introduction
William was the father of Ethel May Kay, who married George Killion. My own connection to this story is indirect — George was my paternal grandmother’s first cousin. While the relationship is distant, the research question is not. The uncertainty surrounding William’s origins has persisted for decades and has been thoughtfully explored before.
In November 2016, William’s great-grandson published a detailed investigation titled “The Mysterious William Kay or Kirkpatrick.” Drawing on family tradition, cemetery inscriptions, newspaper reports and Tasmanian convict records, he constructed a coherent documentary hypothesis explaining William’s identity. [1]
This post does not replace that earlier work. Rather, it builds upon it.
Since 2016, five great-grandchildren of William Kay have tested with AncestryDNA, and I have access to the match lists for four of them. The platform’s large database and shared-match feature enable structured cluster analysis across independent lines, allowing the earlier documentary reconstruction to be examined biologically and introducing an alternative possibility.
Unlike the original investigation, which relied solely on documentary alignment, this phase of the inquiry rests on a specific working hypothesis: that William Kay may have been the biological son of Joseph Rambler Barber and Agnes Kelly.
The purpose of this post is not to assert a conclusion prematurely, but to assess how well competing explanations fit:
the autosomal DNA evidence from five of William’s great-grandchildren,
the structure of shared-match networks,
and the historical context of mid-nineteenth-century New South Wales.
As with any genealogical problem, the question is not which narrative is most compelling, but which explanation is most consistent with the totality of available evidence.
The Kirkpatrick hypothesis
- William married Margaret Deitz at Grafton, New South Wales, on 4 April 1876.
- On the 1882 birth certificate of his daughter Ethel, William stated that he was twenty-seven years of age and born at Burwood, New South Wales.
- That birth certificates of his children place his birth dare between 1852 and 1855.
- The ages recorded across his children’s birth registrations are broadly consistent with a birth in the early to mid-1850s.
- He died on 24 September 1888 at Grafton as the result of a fall from a horse.
- Family folklore maintained that he had been born a Kirkpatrick but was raised under the surname Kay.
A headstone in the Church of England Cemetery at Grafton reads:
“Sacred to the memory of William Kirkpatrick, a native of Tasmania, the beloved son of John Kirkpatrick who died by a fall from a horse, 24 September 1888 aged 36 years.”
No NSW death registration for a William Kirkpatrick in 1888 has been located, suggesting that the man buried under that name was in fact William Kay.
The research then identified a John Kirkpatrick who died at Don Dorrigo in 1895, aged 78, recorded as “not known to have been married,” with no children listed. Newspaper references portrayed him as an early cedar-getter, with rumours of earlier activity in Tasmania and New Zealand.
Attention then turned to a Tasmanian convict named John Kirkpatrick transported on the Triton in 1843. His age and border origins in Carlisle, Cumberland, appeared consistent with family recollections that William claimed Scottish ancestry.
The 2016 conclusion was therefore a reasoned working hypothesis: that William Kay may have been born in Tasmania as the son of convict John Kirkpatrick and later raised under the surname Kay in New South Wales.
At the time, this interpretation rested entirely on documentary alignment.
Introducing genetic evidence and method
Autosomal DNA now provides an additional layer of analysis.
The Testers
For this investigation, autosomal DNA results from five great-grandchildren of William Kay have been analysed. Four descend from one child of William and Margaret — three siblings and their half first cousin — while the fifth descends from a separate child of the couple.
Including more than one descendant line is methodologically important. Patterns that recur across independent branches carry greater evidentiary weight than matches confined to a single line, where coincidence or inheritance variability may distort interpretation.
The question can therefore be reframed:
Do William’s descendants show a consistent genetic connection to the Kirkpatrick family — or does the DNA point elsewhere?
All autosomal DNA analysis discussed in this post was conducted using results at AncestryDNA.
Analytical framework
The evidence was evaluated using three complementary approaches:
Centimorgan ranges
The amount of DNA shared between two people is measured in centimorgans (cM). In general, the closer the biological relationship, the higher the centimorgan total. However, inheritance is variable. Two fourth cousins may share 40 cM, 15 cM, or no detectable DNA at all. For that reason, centimorgan values were interpreted cautiously and never treated as proof on their own.
Shared-match clustering
Rather than looking at one match at a time, this approach asks: do multiple matches also match each other? If several descendants of a known ancestral couple all match one another as well as matching William’s descendants, they form a cluster. Clusters are significant because they suggest descent from a common ancestral line rather than coincidence.
Repetition across independent descendant lines
Strong conclusions are not based on a single tester or a single strong match. Instead, the analysis looks for patterns that recur across different branches of the family. If descendants from separate children of William show the same clustering behaviour, that repeated pattern carries greater evidentiary weight.
Taken together, these three approaches shift the focus away from isolated matches and toward structural patterns — asking not “does one person match?” but “does the overall genetic behaviour make sense if this hypothesis is correct?”
Platform limitations and reporting thresholds
Interpretation also took into account AncestryDNA’s reporting thresholds and interface limitations.
At AncestryDNA, the amount of DNA shared between a tester and a match can be displayed down to approximately 6–8 cM. However, different thresholds apply when examining shared matches.
In the standard shared-match view, a third person will only appear as a shared match between the tester and another match if that third person shares at least approximately 20 cM with both individuals. At fourth- and fifth-cousin distance — where shared DNA is already highly variable — many genuine relatives fall below that shared-match threshold. As a result, they may be biologically related yet remain invisible within standard shared-match clustering views.
AncestryDNA’s Enhanced Shared Matching tools (available through ProTools) provide additional visibility within the shared-match framework. They allow clearer comparison of how shared matches relate to one another. However, the most detailed analysis becomes possible when the researcher has access to a tester’s full match list.
With this access, it is been possible to examine matches below the 20 cM shared-match threshold and incorporate those lower-level matches into broader pattern analysis. This expanded visibility is particularly important at fourth- and fifth-cousin distance, where detectable sharing is inconsistent and individual segments may be small.
Accordingly, the absence of a visible shared match in the standard shared-match interface does not indicate the absence of biological relatedness. It may reflect reporting thresholds rather than true absence of shared ancestry.
For that reason, conclusions were not based on a single visible cluster or one prominent match. Instead, the analysis focused on whether genetic patterns repeated across multiple independent descendant lines — both within the constraints of shared-match thresholds and, where possible, through broader match-list access.
The goal was not to demonstrate every expected cousin relationship individually, but to determine whether the overall genetic pattern formed a coherent and internally consistent structure — and whether that structure aligned with the proposed Barber hypothesis rather than the Kirkpatrick model.
An emerging genetic pattern: The Barber family
The DNA results do not produce recurring shared matches to identifiable Kirkpatrick descendants.
They do, however, show repeated shared matches connected to descendants of John Barber and his wife, Jane Wyer, traced through four of their children who can be followed into adulthood. Three of their children died in infancy or early childhood, and the documentary trail for two others disappears after birth — not uncommon in the early decades of the colony.
If William Kay were biologically connected to the Barber family, the DNA signal would not be expected to sit narrowly within a single descendant branch. Rather, his great-grandchildren would be expected to share autosomal DNA with descendants of one of John and Jane’s children. In other words, the genetic pattern should appear within the broader Barber sibling network.
Before examining which of those children might plausibly be William’s parent, it is useful to outline who John and Jane were and how their family developed in early colonial New South Wales. The biographical summary that follows draws substantially on the research of other Barber descendants, supplemented by updates and corrections from my own review of the available records. A full documentary re-examination of John and Jane’s origins and early lives lies beyond the scope of this post and could form the subject of further research by William’s descendants. [2]
John Barber: From convict to settler
John Barber was transported to New South Wales in 1815 aboard the Marquis of Wellington after being convicted in 1814 for stealing a horse in England. Born about 1788 in Suffolk, he had worked as a stable hand prior to transportation.
In the colony, his occupational skills translated into steady employment. He initially served as a government servant and later established himself as a farrier. By the 1830s he was settled in the Parramatta district and subsequently moved to the Goulburn area. At the time of his death, he was described as a “horse doctor”.
He married Jane Wyer at Liverpool on 23 April 1821. As he was still under sentence at the time, special permission was required for the marriage.
John received his Certificate of Freedom in 1828. In that year’s census he appears as a free Protestant householder at Parramatta with land, livestock, and employees — a notable progression from transported convict to established colonial farmer and tradesman. By 1838 the family had moved to Bungonia near Goulburn.
His death was recorded in the Anglican parish registers at Goulburn on 1 March 1841, aged fifty-two.
Jane Wyer: A child arriving in the young colony
Jane Wyer’s early life is more complex and less firmly documented.
She arrived in New South Wales as a young child, most likely in 1804 aboard the Experiment, and was approximately three years old at the time. There are differing hypotheses regarding her biological mother. Two convict women on the Experiment — Sarah Tandy (also known as Mary West) and Elizabeth Ware — have both been proposed as candidates.
While some researchers favour Elizabeth Ware, I am presently more persuaded by the evidence linking Jane to Sarah Tandy (Mary West), particularly given the documented associations within the Burgin household in Parramatta and the possible London marriage between a Mary West and a George Wyer in 1800. Further research is required before that question can be considered settled.
Jane grew up in the Parramatta district within the household of William and Sarah (Tandy, also known as Mary West) Burgin. William Burgin, an English convict who arrived in 1799, married Sarah formally in 1818, although children had been born to the couple prior to that marriage.
Jane married John Barber in 1821 while already pregnant with their first child. After John’s death in 1841, records indicate that she remarried in 1842 and was buried as Jane Taylor in 1851, although aspects of this later period warrant further detailed review. The available evidence and source references are recorded in my public Ancestry tree.
The children of John and Jane
John and Jane established a substantial colonial family. Based on my current review of the records, their children included:
Joseph Rambler Barber (1821–1907)
William Barber (1823–1826)
Louisa Lucy Barber (1825–1902)
Andrew Bartley Barber (1828–1908)
Samuel Barber (1830–date of death uncertain)
Elizabeth Barber (1833–date of death uncertain)
Henry Barber (1837–1839)
John Barber (1838–1901)
Mary Ann Barber (1840–1845)
Several children died in infancy or early childhood, as was common in the period. Others can be traced into adulthood, establishing multiple descendant lines across Parramatta, Goulburn, Gosford, Kempsey and surrounding districts.
This sibling network is central to the present analysis. If William Kay were biologically connected to the Barber family, his ancestry would most likely lie through one of these children. In that case, the DNA signal would be expected to appear not merely within a single descendant branch, but more broadly across the Barber sibling network.
The following table sets out the autosomal DNA matches identified between descendants of William Kay and documented descendants of John Barber and Jane Wyer.
Autosomal DNA matches between William Kay’s descendants and the Barber sibling network
This table summarises autosomal DNA matches identified at AncestryDNA between tested descendants of William Kay and documented descendants of John Barber and Jane Wyer. Matches are grouped according to the child of John and Jane through whom they descend — Joseph Rambler Barber, Louisa Lucy Barber, Andrew Bartley Barber, and John Barber.
 |
| Click on the image to expand them — the detail is much easier to see at full size. |
The third column indicates the generational relationship between each DNA match and the relevant child of John and Jane (for example, great-grandchild, 2×great-grandchild, etc.). These relationships are based on documented family trees. I do not have access to the match lists of the Barber descendants themselves; the analysis therefore relies solely on matches visible within the Kay descendants’ AncestryDNA tests.
The columns labelled Sibling 1–3 represent three sibling great-grandchildren of William and Margaret (Deitz) Kay. The H1C column represents their half first cousin, and the 2C column represents a second-cousin descendant. All five individuals are great-grandchildren of William and Margaret.
I had access to the full AncestryDNA match lists of the three siblings and their half first cousin for the purposes of this analysis. I did not, however, have access to the match list of the second cousin. The 2C column therefore reflects only matches visible through AncestryDNA’s shared-match functionality.
The numerical values shown in the table represent centimorgans (cM), the unit used to measure shared autosomal DNA. Because these relationships are several generations removed, shared DNA segments are often small and unevenly inherited. Some genealogically valid relatives will share measurable DNA; others may not.
As noted above, differences in match-list access affect what is visible in each column. The 2C column reflects only shared matches visible through AncestryDNA’s shared-match functionality, whereas fuller visibility was available for the other testers.
Accordingly, absence of a shared match in this table should not be interpreted as evidence of absence of biological relatedness.
Accordingly, the analysis does not rely on the size of any single match. Instead, it focuses on clustering patterns — repeated matches to descendants of the same ancestral line appearing across multiple independent Kay descendants. Replication across branches carries greater evidentiary weight than any isolated centimorgan value.
Observations
Several observations emerge:
- The majority of matches occur within the descendant network of Joseph Rambler Barber.
- Many matches appear across two or three independent great-grandchildren of William Kay, indicating replication across descendant lines rather than isolated connections within a single branch.
- Matches are also present within the Louisa Lucy, Andrew Bartley, and John lines, demonstrating that the genetic pattern sits within the broader Barber sibling network rather than being confined to one descendant pathway.
Stepping back another generation
Having identified clustering within the descendants of John Barber and Jane Wyer, the next question is whether that genetic pattern stops there — or whether it extends further back into their families of origin.
If William Kay’s ancestry lies within the Barber line, we would expect to see matches not only to John Barber’s Australian descendants, but potentially also to descendants of his siblings in Suffolk. Although John’s siblings remained in England, some of their descendants later migrated to Australia and the United States, making autosomal connections across collateral lines genealogically plausible.
The same question may be asked of the Wyer line. If Jane Wyer’s ancestry contributes to the observed pattern, clustering might also extend into her family of origin. At present, however, Jane’s parentage remains less securely established, and comparable collateral networks on that side are not yet sufficiently developed for equivalent investigation.
At this greater generational depth, detectable DNA sharing becomes progressively less consistent. Any connections observed would derive from more distant common ancestors, and absence of detectable sharing would not, in itself, be unexpected.
With those constraints in mind, the analysis now turns to John Barber’s family of origin.
William Barber and Mary Friend of Campsea Ash
Documentary research identifies John Barber (1788–1841) as the son of William Barber (c.1754–c.1815) and Mary Friend (c.1756–c.1849) of Campsea Ash, Suffolk, who married there in 1783.
Their known children included:
Lucy Barber (1786–1865), who married Robert Battle
John Barber (1788–1841), transported to New South Wales in 1815
Charlotte Barber (1798–1869), who married William Youngman
Andrew Barber (1802–c.1869), who married Mary Holman
John was the only sibling who lived in Australia. However, collateral migration occurred in following generations, placing other branches of the Campsea Ash Barber family within the Australian colonial population.
The Lucy Barber line
Lucy’s grandson Edmund Battle (c.1828–1913), son of Mary Barber and grandson of Lucy, arrived in New South Wales in 1853 with his wife Rebecca and four-year-old son, Ephraim. Sarah, their two-year-old daughter, died during the voyage.
After arrival, Edmund and Rebecca settled in the Kempsey district on the Macleay River. There they established a substantial colonial family, having eight further children. Through those children, the Lucy Barber line developed an enduring Australian descendant network in northern New South Wales.
The Andrew Barber line
In the same year — and on the same ship — another collateral branch arrived.
A son of Andrew Barber, William Barber (1823–1887), emigrated to New South Wales in 1853. He arrived with his six-year-old son Andrew and three-year-old daughter Mary Ann. His wife Anne and infant daughter Emily died during the voyage.
William died in Liverpool Hospital in 1887. The later life of his daughter Mary Ann has not yet been documented with certainty. His son Andrew married in Narrabri, establishing the Andrew Barber line within regional New South Wales.
By the mid-nineteenth century, multiple collateral branches of the Campsea Ash Barber family were established in Australia. This is genealogically significant. It creates precisely the historical circumstances under which autosomal matches to Suffolk sibling lines might appear among present-day Australian testers.
Extending the genetic pattern beyond John Barber
If William Kay’s ancestry lies within the Barber family at or above the level of John Barber, autosomal matches may appear not only among John’s direct Australian descendants, but also among descendants of Lucy, Charlotte, and Andrew Barber.
The accompanying diagram illustrates these collateral branches where DNA matches have been identified for one or more of William Kay's great-grandchildren:
Yellow: descendants who remained in England
Green: descendants who migrated to or were born in Australia
Blue: descendants who migrated to or were born in the United States
- Red: John Barber
 |
| Click on the image to expand them — the detail is much easier to see at full size. |
Autosomal DNA matches to descendants of William Barber and Mary Friend
The accompanying table summarises autosomal DNA matches identified at AncestryDNA between William Kay’s great-grandchildren and documented descendants of the wider Suffolk Barber family — specifically the lines of Lucy Barber, Charlotte Barber, and Andrew Barber.
 |
| Click on the image to expand them — the detail is much easier to see at full size. |
Sibling 1–3 represent three sibling great-grandchildren of William and Margaret (Deitz) Kay.
H1C represents their half first cousin.
The second cousin (2C) is not included, as no shared matches of 20 cM or greater were identified using the match list of the other Kay descendants.
The numerical values represent centimorgans (cM) of shared autosomal DNA.
Observations
Several points emerge.
- Matches are observed across more than one Suffolk sibling branch, indicating that the genetic pattern is not confined to a single collateral line.
- Multiple matches appear across two or more Kay great-grandchildren. Replication across descendant branches carries greater evidentiary weight than any isolated centimorgan value.
- The absence of qualifying matches for the second cousin is not unexpected given the generational distance. Without full match-list access, only shared matches of 20 cM or greater are visible via AncestryDNA’s shared-match functionality. At this depth, many authentic shared segments will fall below reporting thresholds or may not be inherited at all due to recombination.
Taken together, the clustering pattern suggests that the autosomal signal extends beyond John Barber individually and into the wider family of William Barber and Mary Friend.
Interpretation within generational limits
At this depth, inheritance becomes increasingly stochastic. The analysis does not rely on the size of any single match. Instead, it evaluates whether matches cluster repeatedly within documented descendant networks of the same ancestral family.
The presence of matches across multiple Suffolk sibling lines is consistent with descent from William Barber and Mary Friend, rather than from a more recent or isolated branch.
The Jane Wyer Line: A more tentative position
At present, the same degree of clustering cannot be demonstrated on the line of Jane Wyer.
Jane’s parentage remains uncertain, with competing hypotheses regarding her biological mother. Until that question is more firmly resolved, it is not possible to construct a comparably robust collateral network for investigation.
To date, I have identified only one DNA match that may connect to a sibling line associated with Jane — specifically, a descendant of Sarah Tandy’s later marriage to Henry William Burgin. While this is potentially significant, a single match with only one of William Kay's great-grandchildren does not establish a replicable clustering pattern. It requires further investigation and independent confirmation.
Several factors may contribute to the current absence of clearer clustering on the Wyer side:
The generational distance involved, which increases the likelihood that shared DNA segments will be small or not inherited;
The incomplete reconstruction of Jane’s biological parentage;
The possibility that relevant descendant lines have not yet tested, or that matches fall below AncestryDNA’s reporting thresholds.
Accordingly, the present analysis cannot yet extend the same level of evidentiary weight to the Wyer line as to the Barber line. That remains an area for further research.
Narrowing the field: Returning to the children of John Barber and Jane Wyer
Taken together, the clustering observed within the descendants of John Barber and Jane Wyer — and extending into the wider Suffolk sibling network — suggests that William Kay’s biological ancestry lies within this Barber family.
If that is so, one of William Kay’s parents would be expected to be a child of John Barber and Jane Wyer as John was the only Suffolk Barber in Australia in the early 1850s when William was born.
The documentary and genetic evidence therefore now converge on a more focused question: which of their children is the most plausible candidate?
John and Jane had at least four children who survived to adulthood, while the later lives of a further two remain uncertain. In principle, any of these individuals could represent the parental link to William Kay. However, the strength and distribution of the observed DNA matches provide a basis for examining one child more closely as the starting point for further analysis.
Why begin with Joseph Rambler Barber?
Four considerations make Joseph Rambler Barber (1821–1907) the logical starting point for closer examination.
First, the number and distribution of matches.
The majority of autosomal matches identified to date fall within descendant lines of Joseph Rambler Barber. Importantly, these matches are not confined to a single child’s branch. DNA connections have now been identified to descendants of seven of the nine children of Joseph and Charlotte who married and established families. The matches appear across multiple independent great-grandchildren of William Kay, forming a replicable clustering pattern rather than isolated or sporadic connections.
Second, chronology.
Joseph was born in 1821 at Parramatta. William Kay’s estimated birth falls between 1852 and 1855, based on age statements in later records. Joseph would therefore have been in his early thirties at the time of William’s birth — entirely consistent with biological paternity.
Third, comparison with his siblings.
William’s likely birth in the early 1850s narrows the field among John and Jane’s children. Louisa Lucy Barber and Andrew Bartley Barber were both married in the 1840s and appear to have been in stable family relationships with established households during the relevant period. While it is never possible to exclude any scenario entirely, there is no documentary suggestion that either had a child who was relinquished or informally placed.
The subsequent lives of Samuel and Elizabeth Barber remain uncertain and require further research. The youngest surviving son, John Barber (born 1828), was of plausible age and therefore cannot be dismissed. Although younger than Joseph, he later moved to northern New South Wales after his 1864 marriage in Wollongong, situating him within the relevant geographic sphere.
In short, while several siblings remain theoretically possible, Joseph emerges as the most chronologically and contextually aligned candidate.
This regional footprint is noteworthy.
William Kay married Margaret Deitz in Grafton in 1876. He lived and died in the Grafton district in 1888, and his descendants likewise remained concentrated in northern New South Wales.
Left: Jospeh Rambler and Charlotte (Avery) Barber from Ancestry.com
On its own, shared geography does not establish biological connection. Movement within the colony was common. However, when considered alongside the breadth of DNA clustering across Joseph’s descendant network and the chronological compatibility, the overlap becomes suggestive rather than incidental.
These considerations do not establish parentage. They do, however, justify closer examination of Joseph’s early adult life — particularly the period prior to his 1853 marriage to Charlotte Emma Avery.
Reconstructing the record: Joseph Rambler Barber before 1853
To evaluate whether Joseph Rambler Barber could plausibly be William Kay’s biological father, the historical record must now be examined alongside the genetic evidence.
Joseph was born in 1821, with the event recorded in the Parramatta registration district, the son of John Barber and Jane Wyer. It is important to note that a registration location does not necessarily indicate the exact place of birth; rather, it reflects the district in which the event was recorded. Likewise, his parents’ 1821 marriage was registered at Liverpool, and the early baptisms of their children appear in registers associated with Liverpool, Parramatta and Wilberforce — administrative centres that may not correspond precisely to the family’s place of residence at the time.
By 1838, the family’s later children were recorded in the Bungonia district near Goulburn, where John Barber’s death was also registered in 1841. These registration patterns suggest progressive movement southward through the colony, while reminding us that the documentary footprint reflects ecclesiastical and civil recording structures rather than exact geographic coordinates.
Joseph therefore came of age within a family already accustomed to mobility between developing rural districts.
Early adult life in the Goulburn District
The first clear glimpse of Joseph as an adult appears in a newspaper notice published in September 1842:
TWO POUNDS REWARD.Joseph Barber, native of the colony, sallow complexion, about 22 years of age, and 5 feet 9 inches in height, has lately been living near Tarlow Gap, within seven miles of Goulburn…
The notice, placed by J. B. Ward of Murrimbah, sought the apprehension of sawyers who had absconded before completing their employment agreements. Such advertisements were common in colonial newspapers, which functioned not only as news sources but as instruments of labour regulation and legal enforcement. [3]
Whether this Joseph Barber is definitively Joseph Rambler Barber requires cautious correlation, but the age, description, and location align closely. Importantly, it situates him in the Goulburn–Bungonia region in the early 1840s — precisely where his family had been living.
Marriage in 1845
The next record identified in this renewed examination of Joseph’s life is a marriage that, on present evidence, most likely relates to him, as no other plausible candidate has been identified.
The New South Wales Marriage Index records that a Joseph Barber married Agnes Kelly in 1845 at Bungonia, in the Goulburn district (Volume V B). Joseph would have been approximately twenty-four years old at the time. [4]
On the balance of probabilities — taking into account the location, his age, and the known movements of the Barber family — this record is most consistent with Joseph Rambler Barber.
It places him in the Bungonia–Goulburn district in the mid-1840s and provides an important chronological anchor for the next phase of analysis.
Agnes Kelly: Background and context
A Scottish baptismal record identifies:
Agnes Kelly
Born 25 March 1819
Baptised 30 March 1819
Dumbarton, Dunbartonshire, Scotland
Father: William Kelly
Mother: Elizabeth McNaught
On the balance of probabilities, this baptism corresponds to the Agnes Kelly later tried at Dumfries in 1838 for theft and transported to New South Wales aboard the Planter in 1839. The age, Scottish origin, and chronology align closely with the details recorded in the convict indent. [5] [6]
By 1844, Agnes held a Ticket of Leave in the Goulburn district. In 1845 she received her Certificate of Freedom. Her marriage to Joseph Barber took place in that same district during that same period. [7]
Taken together, the geographic, chronological and administrative records place Joseph and Agnes in the Goulburn–Bungonia district in the mid-1840s. While absolute identity confirmation requires cautious handling, the convergence of age, location, and timing makes it highly probable that the Scottish-born Agnes Kelly baptised in Dumbarton in 1819 is the same woman later transported and married in New South Wales.
The Goldfields and the death at Sofala
The next significant record appears in 1852.
On 5 September 1852, an inquest was held at Sofala into the death of an Agnes Barber. The coroner recorded the death as sudden, reportedly resulting from a fall. Civil registration confirms that an Agnes Barber died at Sofala in that same year. [8] [9]
On the balance of probabilities, this woman is Agnes Kelly, wife of Joseph Barber. The name, the district, and the timing align closely with the known movements of Joseph and Agnes in the Goulburn region in the mid-1840s, followed by the broader shift toward the goldfields in the early 1850s. No alternative Agnes Barber of comparable age and circumstance has yet been identified in that locality during this period.
If this identification is correct, Joseph’s first wife died at approximately thirty-three years of age during the early gold rush.
This situates Joseph in the Bungonia–Goulburn region, with links to the Sofala goldfields, during the period in which William Kay’s estimated birth (between 1852 and 1855) would have occurred. Agnes’s death in 1852 serves as an important chronological marker in Joseph’s life, immediately preceding his remarriage in 1853.
The absence of recorded children, 1845–1852
No civil registration has yet been located for a child born to Joseph and Agnes between their 1845 marriage and Agnes’s death in 1852.
However, this absence must be interpreted cautiously.
Civil registration in New South Wales did not become compulsory until 1856. Prior to that date, births were recorded primarily in church registers, and coverage was uneven — particularly in rural and frontier districts such as Bungonia, Goulburn and Sofala. Clergy were few, travel distances were significant, and record survival is incomplete.
In addition, Joseph and Agnes appear to have been geographically mobile during this period — moving between Goulburn, Bungonia, and possibly the Sofala goldfields. Such mobility increases the likelihood that baptisms, if they occurred, may have been recorded in registers not yet indexed or digitised.
The absence of a currently located birth record therefore does not demonstrate that no child was born. Rather, it highlights the need for systematic examination of surviving church registers in the Goulburn district and surrounding areas.
Why this period matters
This reconstructed timeline is significant.
Taken together, the available evidence indicates, on balance, that Joseph Rambler Barber married at Bungonia in 1845 and that his first wife had died at Sofala by 1852 during the early gold rush period. He remarried in 1853.
William Kay’s estimated birth window — approximately 1852 to 1855, based on later age statements — falls within this transitional period in Joseph’s life.
The DNA evidence presently available is compatible with connections to Joseph’s descendant line. The documentary record places Joseph and Agnes in New South Wales at biologically plausible ages and within the relevant chronological window. Although this alignment does not establish parentage, it supports the consideration of Joseph Rambler Barber and Agnes Kelly as one plausible parental hypothesis for further investigation.
At present, no competing documentary or genetic evidence points clearly toward an alternative parental hypothesis. The earlier Kirkpatrick tradition is not supported by the DNA findings available to date, and no other family network has produced comparable clustering. This does not exclude other possibilities, but it does mean that the Barber–Kelly hypothesis currently provides the most coherent working framework for continued investigation.
Working hypothesis
Under this scenario, William would have been born in the late 1840s or early 1850s, most plausibly in the Goulburn–Bungonia district, shortly before or around the time of Agnes’s death in 1852. If Agnes died while William was still an infant or very young child, it is conceivable that he was raised within the extended Barber–Burgin family network.
The recurrence of the name William within both parental lines is also noteworthy. Agnes’s father was William Kelly, and Joseph’s grandfather was William Barber. While naming patterns cannot establish parentage, the presence of the name across both sides of the proposed family structure is consistent with nineteenth-century naming conventions and may offer contextual support to the hypothesis.
William’s later statement that he was born at “Burwood” may reflect a phonetic confusion, later association, or simplified recollection. Whether this refers to Bungonia, another locality, or a later residence remains open to investigation rather than assumption.
The surname “Kay” may likewise reflect family circumstance rather than formal registration. It could have distinguished him within a family containing multiple Williams, or it may preserve an adapted form of his mother’s surname, Kelly. At present, this remains interpretive.
The absence of identifiable DNA matches on Agnes Kelly’s Scottish side does not materially weaken the hypothesis. At this generational distance, shared autosomal segments are often small and unevenly inherited. Some relatives share measurable DNA; others, equally related on paper, may not.
Taken together, the genetic clustering, geographic placement, chronological compatibility, and naming patterns form a structured explanatory model. It does not constitute proof. Rather, it establishes a testable framework.
Importantly, this framework also allows for refinement. If the Barber connection ultimately proves correct but the precise pathway differs — for example, through another descendant of John and Jane (Wyer) Barber — the core conclusion that William’s ancestry lies within the Barber network would remain intact, even if the parental route requires adjustment.
Based on the evidence discussed above, the following diagram represents the family structure currently under consideration.
Testing the hypothesis: Where to from here?
A working hypothesis must be capable of withstanding further evidence — and, where necessary, refinement.
If William Kay was the son of Joseph Rambler Barber and Agnes Kelly, we would expect the DNA pattern to continue appearing in predictable ways as additional descendants test and as broader match-list access becomes available. In practical terms, we should continue to see repeated clustering between William’s descendants and multiple branches of the wider Barber family.
At the same time, the emerging evidence suggests that the critical question may not simply be whether the Barber family is involved, but which branch provides the correct parental pathway. Continued testing may therefore strengthen the Joseph–Agnes model — or clarify a more precise route within the broader Barber network.
Equally important is renewed documentary research. Parish registers, goldfields records, and local court material may clarify whether a child born in this period could have been absorbed into an extended family under a different surname.
The next stage therefore turns back to William’s descendants. As the evidence base expands — genetically and archivally — the relative strength of the competing pathways within the Barber family will become clearer.
One such possible refinement within the Barber network arises from a cluster of matches connected to George Burgun.
"The Burgun cluster"
When I first began this analysis seven years ago, a persistent cluster of DNA matches descending from George Burgun (c. 1825–1912) and his wife Ellen Arlott repeatedly appeared in the data. At the time, this cluster seemed to sit outside the developing Barber framework and therefore required careful consideration.
Left: George and Ellen Burgun's headstone from Ancestry.com
As the investigation has progressed, that cluster has taken on new significance — not as a contradiction of the Barber hypothesis, but as a possible refinement of it.
 |
| Click on the images to expand them — the detail is much easier to see at full size. |
The above table summarises those matches by showing:
each match’s relationship to George Burgun (for example, great-grandchild or 2×great-grandchild); and
the amount of DNA shared, measured in centimorgans (cM), with William Kay’s great-grandchildren.
At this generational distance, interpretation rests not on any single centimorgan value but on whether consistent patterns appear across multiple independent descendants
Why the cluster matters
At first, this cluster appeared distinct — almost separate from the broader Barber pattern that was emerging elsewhere in the analysis.
Ellen Arlott’s ancestry is well documented, and there was no evidence of shared DNA connections through the wider Arlott family. That initially suggested that the shared DNA might derive from George Burgun’s ancestry rather than Ellen’s, and perhaps indicate an entirely different parental pathway.
However, closer examination changed that interpretation. The same Burgun matches that seemed to form a discrete group were also visible as shared matches not only with William Kay’s great-grandchildren, but with documented descendants of John and Jane (Wyer) Barber. In other words, the Burgun cluster did not sit outside the Barber network — it overlapped with it.
That overlap fundamentally altered its significance.
No birth registration has been located for George Burgun, and no contemporary record identifies his parents. His obituary states that he was born at Windsor in the mid-1820s — at a time when John Barber was living in the Parramatta–Liverpool district.[10]
The same obituary portrays him as a drover and bushman who overlanded cattle, joined the Meragle gold rush, and later participated in one of the expeditions searching for Ludwig Leichhardt. These details situate him within the highly mobile pastoral networks of inland New South Wales during precisely the decades in which William Kay’s early life remains undocumented.
Closer examination of shared matches has shown that descendants of George Burgun share DNA not only with descendants of John and Jane (Wyer) Barber, but also with extended Barber relatives connected to John’s Suffolk family. The introduction of AncestryDNA’s Enhanced Shared Matches feature has strengthened this observation by making certain smaller shared segments visible when full match-list access is available — particularly useful at this generational distance.
Taken together, this pattern places George Burgun within the Barber family. Given the timing of George’s birth in the mid-1820s, and the absence (so far) of any other identified Barber male in New South Wales who could account for that connection, the most coherent explanation is that George was a biological son of John Barber who was raised within the Burgin household.
This, in turn, raises the question of George’s mother. Two possibilities emerge within the documented social network of the period. The first is Sarah Tandy (also known as Mary West), who later married George Burgin but had been living with him and bearing children from as early as 1805. The second — less likely but not impossible — is her daughter Sophia Burgin, born in 1808.
Either scenario would place George’s early upbringing within the Burgin household and provide a plausible explanation for the surname by which he is known. At present, however, the precise maternal pathway cannot be reconstructed from the available DNA evidence alone, particularly without broader access to match data from Burgun descendants.
Where George fits in the Barber family
On this reading, George represents a second pathway by which William Kay’s descendants could connect genetically to the Barber family:
John Barber → George Burgun → William Kay
rather than
John Barber → Joseph Rambler Barber → William Kay
Chronologically, this is not implausible. George would have been in his early thirties during William Kay’s estimated birth window (c. 1852–1855). His mobility and documented bush experience place him within the same broad historical and geographic context.
However, several considerations complicate this picture.
If George Burgun were William Kay’s father, no likely mother for William has yet been identified in either the documentary record or the available DNA evidence. In nineteenth-century New South Wales, children born outside formal marriage were commonly raised within the maternal family network. The absence of an identifiable mother therefore presents a significant gap in this pathway.
The same issue, though in a different form, arises under the Joseph–Agnes model. Agnes Kelly died in 1852 and had no known family in Australia. If William were her child and she died while he was still very young, he would not have had an established maternal kin network within the colony. That does not render the hypothesis implausible, but it would require an explanation for where and with whom he was raised.
In short, the identity and circumstances of William’s mother remain unresolved in both scenarios. This shared evidentiary gap underscores the importance of continued documentary and genetic investigation.
Second, no record has yet been located of a formal marriage between George Burgun and Ellen Arlott, despite their well-documented descendants. While the absence of a marriage record is not unusual in mid-nineteenth-century New South Wales, it introduces an element of documentary uncertainty into George’s early adult life.
Third, the first recorded birth to George and Ellen is that of a son, George, born at Tumbarumba in 1866 — well after William Kay’s likely birth and at a time when George senior was already in his forties. This leaves a significant undocumented period in George’s earlier adulthood. Such gaps are common in colonial records, but they do not in themselves establish a connection to William Kay.
The Leichhardt reference in George’s obituary adds historical texture rather than proof. Participation in an expedition searching for Ludwig Leichhardt does not imply bushranging, nor should it be conflated with later family legend. It does, however, reinforce the image of George as a highly mobile bushman active during precisely the period in which William Kay’s early life remains unclear. That alignment is contextual rather than evidentiary.
There are also genetic limitations. The DNA clustering does not isolate George’s descendant line as the primary signal. Rather, the matches remain distributed across the broader Barber sibling network. At present, the pattern is consistent with a Barber connection, but does not definitively distinguish between the Joseph–Agnes pathway and a possible George pathway.
Accordingly, George Burgun is best framed as a secondary working hypothesis: chronologically plausible, geographically compatible, and consistent with a Barber connection — but not presently better supported than the Joseph–Agnes model.
This does not overturn the earlier analysis; rather, it sharpens it.
This does not overturn the earlier analysis; it clarifies it.
The genetic evidence now places William Kay’s ancestry within the Barber family. The question is no longer whether the connection lies there, but whether it runs through Joseph Rambler Barber — or through George Burgun.
That represents a more precise stage of investigation. Rather than choosing between unrelated families, we are refining the pathway within a single extended colonial kin network.
Conclusion: An invitation to continue the work
This investigation has moved from family tradition to documentary reconstruction and into genetic analysis. In doing so, it has uncovered more than a possible father for William Kay. It has opened a structured inquiry into his parentage and, in the process, illuminated a wider story of early colonial New South Wales — of convicts and settlers, goldfields and grazing stations, mobility, reinvention, and the fragile formation of family identity.
The genetic evidence now places William Kay’s ancestry within the Barber family. The remaining question is not which unrelated family is involved, but which pathway within that family — and which parental combination — provides the correct explanation.
At present, the Joseph Rambler Barber–Agnes Kelly model offers the most coherent alignment of chronology, geography, and DNA clustering. The George Burgun pathway represents a plausible alternative within the same Barber kin network, particularly in relation to the paternal line. In both scenarios, the identity and circumstances of William’s mother remain central to the analysis and require further investigation.
The next stage belongs, in part, to William’s descendants.
Further testing, broader match-list access, and renewed documentary research may strengthen, refine, or recalibrate the current framework. Each additional piece of evidence — genetic or archival — helps clarify not only names and biological relationships, but the lived realities behind them.
This is not simply a search for a father’s name. It is an effort to understand how families were formed, how children were raised or absorbed into kin networks, and how identities were shaped in a young and rapidly changing colony.
If the Barber connection is fully confirmed — whether through Joseph and Agnes, or through George and an as-yet unidentified maternal pathway — then William Kay’s story forms part of a larger and important chapter of colonial New South Wales. That story deserves to be documented carefully and explored collaboratively.
I hope that descendants of William Kay — and of the Barber, Burgin and Burgun families — will feel encouraged to engage with the evidence, to share information, and, where possible, to participate in further testing.
Our understanding of the past evolves as records and genetic tools evolve. This is not the end of the story, but a thoughtful stage in its reconstruction.
[2] https://web.archive.org/web/20200627214156/http://users.skynet.be/lancaster/Harriet%20Barber.html
[3] Advertising (1842, October 6). The Sydney Morning Herald (NSW : 1842 - 1954), p. 1. Retrieved February 8, 2026, from http://nla.gov.au/nla.news-article12409359
[4] New South Wales Registry of Births, Deaths and Marriages, Marriage Index 1788–1950, Joseph Barber to Agnes Kelly, 1845, Bungonia (Goulburn district), vol. V B.
[5] Ancestry.com, Scotland, Select Births and Baptisms, 1564–1950 (database online), Lehi, Utah: Ancestry.com Operations, Inc., 2014; citing Scotland, Births and Baptisms, 1564–1950, FamilySearch, 2013; entry for Agnes Kelly, 25 March 1819, Edinburgh, Midlothian, Scotland; accessed 8 February 2026.
[6] Reakes, J., comp., Australia, Convict Index, 1788–1868 (database online). Provo, Utah: Ancestry.com Operations Inc., 2001; accessed via Ancestry.com 8 February 2026.
[7] Ancestry.com, New South Wales, Australia, Certificates of Freedom, 1810–1814, 1827–1867 (database online), Provo, Utah: Ancestry.com Operations, Inc., 2009; citing New South Wales Government, Butts of Certificates of Freedom, NRS 1165, 1166, 1167, 12208, 12210, reels 601, 602, 604, 982–1027, State Records Authority of New South Wales, Kingswood; entry for Agnes Kelly, 1845, Series NRS 12210, roll 1021; accessed 8 February 2026.
[8] Ancestry.com, New South Wales, Australia, Registers of Coroners’ Inquests, 1821–1937 (database online), Provo, Utah: Ancestry.com Operations, Inc.; citing State Archives NSW, Series 2921, Item 4/6613, Roll 343; entry for Agnes Barber, 1852; accessed 8 February 2026.
[9] Ancestry.com, Australia, Death Index, 1787–1985 (database online), Provo, Utah: Ancestry.com Operations, Inc., 2010; citing NSW Pioneer Index – Pioneer Series 1778–1888; entry for Agnes Barber, 1852, New South Wales; accessed 8 February 2026.
[10] "DEATH OF A PIONEER" Windsor and Richmond Gazette (NSW : 1888 - 1971) 15 February 1913: 11. Web. 8 Feb 2026 <http://nla.gov.au/nla.news-article85848936>.