DNA Day Sale on at FTDNA - ends 28th April

Rox2 SNPs: R1b-P312>ZZ11>DF27>ZZ12>ZZ19>Z34609>Z2571>FGC11380>(FGC11397 and a few dozen phylogenetic equivalents). YFull name: R-Y8397

A yDNA cluster is essentially a group of present-day men who share similar Y-STR results - ones that indicate possible descent from the same earlier male common ancestor. SNP testing can confirm a STR cluster match to be a member of a SNP defined subclade (a related group). 'Rox2' is the nickname for this particular cluster/subclade and its 'signature' STR pattern was first noticed in 2005. The Y chromosome is passed down virtually unchanged to male descendants of a common ancestor - from father to son - although small random differences accumulate over time. This is due to mutations on the Y chromosome that occasionally occur when it is replicated. Those mutations form distinct patterns that are passed on and are unique within the general population. The pattern can be used to identify potential matches and also estimate age. The more recent and prolific the common ancestor, the more similar and numerous the haplotypes are in today's population. Rox2 can be clearly identified using STR markers alone (see below under STR IDENTIFICATION). Those who closely match the key off-modal Rox2 'signature' markers over 111 markers and take SNP tests consistently find they are also positive for FGC11397, FGC11369, FGC11394 or any other currently equivalent Rox2 defining SNPs when they test for them. The subclade formed centuries before the fixing of surnames and as a consequence there are many different surnames represented in the subclade today.

TMRCA (Time To Most Recent Common Ancestor) estimates suggest the Rox2 subclade founder (one man) lived around 1250 years ago + or - a generous margin of error (i.e. 'born' in about 750 AD at the end of a c. 3000 year bottleneck period). In broad terms, allowing for the possibility that the bottleneck is somehow making the subclade appear younger, it is highly likely Rox2 was 'founded' in the Early Middle Ages (some time between the 5th century and the 10th century AD). That Rox2 founder had at least five 'sons'. Those five brothers are the ancestors of all present-day Rox2 matches. Prior to relatively recent emigration to worldwide colonies, Rox2 appears to have had exclusively northern European geographical distribution, the most southerly match being from northern France so far. There is no firm evidence yet as to where Rox2's Early Medieval ancestor came from or his family background, that cannot be determined with modern yDNA alone. 

Note: FTDNA upgraded Big Y results from build hg19 (2009) to hg38 (2013). The 'hg' stands for 'human genome'. If you have pre-2018 Big Y results on the Big Tree, send your newly updated hg38 files (once they are completed) to the DNA Warehouse at the Big Tree (not Yahoo Groups as before) to update your results. Extra SNPs could be found in previously unmapped areas of the genome.


DF27, the c. 5000 year old subclade Rox2 descends from, is the largest, most widespread but possibly least well-understood of the three big R1b-P312 subclades. The technology used in SNP tests has difficulty identifying DF27. However, 'NGS' (Next Generation Sequencing) testing does highlight the downstream SNPs and understanding is improving with the sharing of results on phylogenetic trees and analysis of ancient archaeological yDNA. R1b-P312>ZZ11 subclades look to have originated in the east and moved west with Corded Ware and Bell Beaker cultures in the Bronze Age, in the 2nd millennium BC. The earliest ancient DF27 found so far is from a man buried with a Corded Ware axe and a Bell Beaker in c. 2300 BC in what is now Germany. For discussion of ancient DF27's origins see Distribution Maps.

After more extensive post-2014 DF27 NGS testing (using FGC and FTDNA's Big Y) and analysis by Alex Williamson, two ancient groups just below DF27 were revealed. They are ZZ12 and the smaller, parallel subclade that had already been known for several years, Z195/Z196.
Rox2 is DF27>ZZ12+. Its ancestral branch below ZZ12 is ZZ19/ZZ20 (hg19 25938772 T>A, 22224478 A>G). Subclades under ZZ19 are widespread and present-day descendants are found all across Europe, including in Sweden, Finland, Poland, Denmark, Czech Republic, Germany, Austria, Armenia, Italy, France, Spain, Portugal, Britain and Ireland. ZZ19 is a very large subclade, making up over one third of the total of all DF27>ZZ12 on the Big Tree. Below ZZ19 is Z34609These ancient 'top layer' subclades were in existence before around 4500 BC. Those who remain Z34609* indicate that more subclades await discovery below Z34609. Kits below Z34609 that are negative for Z2571 are from widespread areas, Armenia, Czech Republic, Italy and Iberia.

'Son' of Z34609, Z2571, has two ancient parallel subclades downstream of it, they are FGC11380 (YFull now has a mini-phylogenetically equivalent block here, R-Y8841: FGC11385 * FGC11384 * FGC11380/Y8841) and CTS11567. Rox2/FGC11397 is downstream of FGC11380/Y8841. I got a FGC11397+ (Rox2) result from YSEQ in June 2014, as have several other STR matches who have since tested single SNPs with either YSEQ or FTDNA. See this diagram.


Z2571 (hg19 23076115 C>G, hg38 20914229 C>G) A deep DF27 SNP. ISOGG: R1b1a1a2a1a2a6.

FGC11380 (hg19 23098886 T>C, hg38 20937000 T>C) Early branch point. (Big Tree FGC11380 Y8841, 22220783-C-T, YFull's R-Y8841: FGC11385 * FGC11384 * FGC11380/Y8841)

FGC11397 (hg19 7332619 G>A, hg38 7464578 G>A) 'Rox2'. A currently phylogenetically equivalent SNP, one of a block of dozens of equivalent SNPs below FGC11380/Y8842 that define the subclade. (Available to test at YSEQ.net and FTDNA). The Big Tree leads with FGC11388. YFull lead with Y8397 (FGC name FGC11374). Another, possibly early, equivalent is FGC11381 (hg19 6720487 C>T/hg38 6852446 C>T).


The best and simplest yDNA testing approach to clearly identify a Rox2 match is to get a NGS SNP test (eg. Big Y, YSEQ or FGC) and then, if happy sharing results, upload those NGS SNP results to an external public phylogenetic tree (Big Tree). See 'RESEARCH' paragraph, below. Analysis on the Big Tree is free and arranges your results into an easily understandable 'family tree' format. Additionally, there are a large number of existing Rox2 111 STR marker matches, so an upgrade to the full 111 markers at FTDNA can also be useful. Please join the FTDNA DF27 project and encourage your 67 and 111 matches to do so - there are many 67 and 111 marker Rox2 matches across many different FTDNA projects who are unaware of the fact they match Rox2. A match with the eleven off-modal signature markers over 111 markers is as clear an identification of Rox2 as a single FGC11397/FGC11369 etc. SNP test. 

NGS results, when added to a phylogenetic tree, like The Big Tree or YFull's tree, can highlight which specific branch of Rox2 you belong to, if currently known. If your Rox2 branch is not currently known, a branch will form when a more close match subsequently uploads their NGS results to the tree in the future. However, little will be learned by a living y relative, or a known cousin (from one's own close family tree), taking a Big Y NGS test, the results may be identical to yours - Big Y identifies on average one hg19 SNP approximately every 100 years for Rox2. Progress can only be made if new results are shared - for a subclade of this size it's a team effort. There are occasional sales at FTDNA - there is often a sale at Christmas on Big Y or STR upgrade tests.

Turner (FTDNA kit N3036, FGC kit number GYBGZ) received FGC results on 3rd April 2014 (link). Many SNPs were seen to be shared with an anonymous kit in the 1000 Genomes project, HG00107. HG00107 has Orkney ancestry and matches the Rox2 off-modal STR pattern. Dickinson (FTDNA kit 134765, FGC kit WBAFF, Cumbria) received FGC results on 12th April 2014 He and Turner share dozens of SNPs in a phylogenetically equivalent block below DF27>Z2571. FTDNA's Big Y appears to read under half of the 45-ish Rox2 equivalent SNP block present in the two FGC tests but 'private' SNPs below the shared block compare well between the two tests.

The very large phylogenetically equivalent SNP block shared by all Rox2 subclade matches might be imagined as a father-to-son chain of descent with dozens of SNP 'links' in that chain below Z2571>FGC11380. However, we don't know the (chronological) order of the links. So far, that block/lineage is unbroken in thousands of years, from at least around 4300 years ago +/- a wide margin of error (YFull estimate) up to about the eighth century AD (+/-). In that large passage of time no branches survive, or they have yet to turn up. A branch in the large shared block might turn up in the future and could potentially give some clues about the origin of the subclade. This very long SNP 'bottleneck' ends some time around 750 AD when several branches suddenly flourish all at once - a 'founding event'.

Underneath the Rox2 shared block, a smaller string of more recent SNPs have accumulated over time for each lineage down to the present day (on average roughly just over one dozen 'definite'' hg19 SNPs in Big Y - i.e. SNPs that are not insertions/deletions or other more complex mutations that the automated technology has difficulty reading. Those SNPs have happened in the time between the founder's birth in around 750 AD (+ or -) and the present. If any of those SNPs are shared with another Rox2 kit (after comparison on a phylogenetic tree like the Big Tree) a new branch will form. That branch highlights the point where both kits share a common ancestor. There is wide variation in the numbers of SNPs different individual lineages accumulate during the same passage of time, SNPs don't happen like clockwork and can be random and affected by factors like the age of father and possibly environmental influences.
Automated SNP age estimates treat all subclades the same across the entire board, potentially skewing results if that age estimate formula is applied to isolated and  idiosyncratic subclades. Different subclades can have different characteristics. However, the SNP numbers average out across several Rox2 kits. As a very rough guide, Big Y might on average identify one definite hg19 SNP in approximately 100 years (3-4 generations) for a Rox2 kit. Again, this may vary in other subclades. Big Y kits returned after 2017 are hg38, not hg19 ('BigY2'). Rox2 kits with hg38 results get more SNPs than hg19. Eventually, when there are sufficient numbers of hg38 kits, it will be possible to make an estimate of years-per-SNP for hg38-build kits and compare them with earlier results from hg19 tests. At the moment things have become 'messy' as far as SNP counting goes - with 'lo-res' hg19 and new 'hi-res' hg38 .vcf files together on the Big Tree. From initial impressions (March 2018) it looks like the hg38 'BigY2' rate will be on average closer to around 90 years per Big Y 'family' SNP, rather than 100 years per SNP for hg19 files.

If for example a young subclade was descended from several sons of a recent common ancestor (say, a man born in 1750 AD), then there might be few unique family SNPs in evidence in each separate line of descent from that man. Also, the genetic distance (GD) between the cluster's STR haplotypes would be very low. If, however, the common ancestor was born in 1350 AD, then half a dozen reliable hg19 SNPs could have occurred in each brother lineage since that founder was born. The STR haplotypes would usually have a higher GD too.

In summer 2017, FTDNA began changing their haplotree and gave their own 'BY' (Big Y) names to some 'family' downstream Rox2 SNPs. In autumn 2017 they began changing Big Y hg19 SNP position numbers to hg38. FTDNA, YFull and YSEQ appear to have their own ideas/criteria about what SNPs, names and positions are included in their trees. They can leave out some of the important SNPs, Indels or STRs , i.e. ones that are difficult for the current (automated) technology to call. I am focusing on the Big Tree here because it most clearly illustrates key Rox2 branching. The new Big Tree age estimates are a work in progress (currently miles out for Rox2 and some other subclades). The combination of a couple of higher-resolution FGC results and a mix of lower-resolution hg19 Big Y and higher-resolution hg38 'BigY2' results on the Big Tree might be affecting age estimates for some subclades. YFull use a formula across their database to allow for this distortion in their age estimates. Again, this does not allow for different subclades potentially having different SNP numbers and mutation rates.

There are currently five deep subclades below Rox2 represented on the Big Tree. These deep subclades might be thought of as 'sons' of the Rox2 founder but not necessarily actual sons - they could well be - but the SNP mutation might have happened a generation or two down the line. They are parallel lineages descended from different common ancestors, their connection is their Rox2 'father' - the founder. NGS testing followed by uploading results to phylogenetic trees, like Big Tree, highlight actual linear descent with SNP-defined branching. Two close Rox2 STR matches might occasionally have very low genetic distance from each other by random coincidence at lower resolution. A 'lack of divergence' between two 67 marker haplotypes can sometimes give the incorrect impression that the two matching kits are closely related. Fortunately, NGS tests using SNPs see through this. For example, a member of one Rox2 subclade BY21578 is an almost identical STR match (66/67 markers) to a member of parallel 'brother' Rox2 branch Y17787 but is still unrelated in over 1000 years.

Those five NGS-identified Rox2 deep subclades (so far), as seen on the Big Tree, are:
  • FGC11414 is itself the result of a large founding event and has several branches below, or 'downstream' of it - they are 'sons' of FGC11414. Three have been identified so far. One branch, FGC11407 (plus equivalents hg19 7031104 C>T/hg38 7163063 C>T and FGC11419 at FTDNA's haplotree) has the two FGC tested kits. FGC has better coverage than Big Y. BY21591 contains Big Y kits with different very early (pre-colonial migration/pre-Industrial Revolution) ancestors from Yorkshire. A currently exclusively Swedish Rox2 branch of similarly early ancestry is represented by BY21590 (plus equivalents Y23589 and Y23590). BY21590 has two branches on FTDNA's Haplotree, Y60859 (plus three equivalents) and Y61403 (plus five equivalents) but Big Tree does not show these branches. Lastly there is FGC11414* (the asterisk indicates no further branching known below FGC11414). See the Phylogenetic Tree page. The three subclades of FGC11414 have been existing in parallel since the earliest days of Rox2. Their connection is their 'father', FGC11414. The places of birth of different FGC11414 subclades could be nearby, or distant from, the location that their FGC11414 father was in nine months previously. The presence of FGC11414* kits indicate more 'brother' subclades will develop downstream of FGC11414 as new NGS results are added to the trees.
  • A17453 Immediately below this is A16311, two kits are A17453*. Notes regarding A16311 on the Big Tree say, '*Mutations whose exact position can't be determined precisely from NGS tests, such as those in palindromes, are shown with a pink background.' FTDNA's haplotree doesn't have A16311. There are three 'sons' of A16311. They are BY21588 (plus two equivalents), A16155/BY21584 (plus four equivalents) and Y33105 (plus three equivalents, BY21592, Y33106 & Y32852). FTDNA and YFull's haplotrees may differ. A17453 has the potential to be a large subclade and its evolving structure and size on the Big Tree is interesting. It is important that new Big Y results are uploaded to the Big Tree to help understanding of the increasingly complex branch structure. Instructions for doing so can be found here.
  • Y17787 (YFull use equivalent Y17484 as lead SNP, Big Tree and FTDNA use Y17787) kits share four equivalent SNPs after the Rox2 founder's time. That shared SNP block indicates that kits below it have common ancestry with kit 185183 for a 'bottleneck' period of a few hundred years. Then, the present-day DF27+ Macauley families branch off from a common ancestor identified by a couple of equivalent SNPs, Y17484 and FGC45672. There are currently two identified early 'sons' of the Macauley 'father of the family', Mr. Y17787. Those two sons are Y21522 (plus three equivalents) and Y30689. Y30752 is a son of Y30689, so he is a 'grandson' of Y17787. There are potentially two more early 'sons' of Y17484 waiting to be discovered with further testing, indicated by the presence of two 'Y17484*' kits.
  • BY21578/YSEQ's A12539 leads a SNP block on The Big Tree containing six other currently phylogenetically equivalent SNPs and has three members of the Hickey group immediately below it. BY30716 branches off with an equivalent block of six SNPs below BY21578.
  • FGC39762 leads a block of 15 phylogenetic equivalents on the Big Tree that so far contains two relatively recently related Simpson men.

* Bold = FTDNA names. The kits on the Big Tree represent a small proportion of overall Rox2 matches. Existing subclades will gain more definition with the addition of new NGS results. The above lineages, descendants of early 'sons' of Rox2, have lived in parallel for 1000+ years. It takes at least two kits from the same early lineage to form a new branch and that can only be identified after inclusion and analysis (.vcf and BAM files) on a phylogenetic tree.

Now that the wider subclade has several families researching their own recent downstream 'family' subclades, it is becoming impractical to include all details here. Members of the families/subclades below Rox2 could create dedicated webpages/blogs that cover latest information about more recent areas of their part of the tree. It would be great to be able to keep up to date with your progress. That more specific information might then be linked to from here.

Rox2 SNPs, mentioned above, can be ordered or requested individually at a reasonable price from YSEQ.net. This is not necessary if you have a NGS/Big Y test, all one's SNPs are covered by that test. In order for YSEQ SNP results to show up on FTDNA project screens the kit owner would need to enter their YSEQ result in their FTDNA kit's 'Most Distant Ancestor' box on their
'Personal Profile' page. Only NGS tests submitted to the DNA Data Warehouse get you on the Big Tree. One-at-a-time SNP testing based on clues from existing NGS results does not identify one's own family SNPs. Any testing approach depends on individual goals. If happy to simply find a link to existing kits from a few hundred years ago, YSEQ-style one-at-a-time tests can do that. If the aim is to identify one's own unique family SNP trail, with the potential to find more closely related NGS matches in the future, NGS testing is the route to go down.

(click image to enlarge)


A 'lead' SNP is a SNP used at the start of the phylogenetically equivalent block to represent the subclade. It is not the first, last or 'terminal' SNP, it's just one someone chose out of the equivalents to be at the start of their list. Unfortunately many different names for the same SNPs have arisen since 2014 and different equivalent SNPs also get chosen as lead SNPs by different people. YFull use Y8397 (FGC name FGC11374) as Rox2's lead SNP in their treeYFull have also given their YFS names to some of the other phylogenetically equivalent Rox2 SNPs already named by FGC, eg. FGC11395 is known as Y8842. Alex Williamson has FGC11388 (YFull's Y8837) as the lead phylogenetically equivalent SNP in his Big Tree. YSEQ lead with FGC11369 (YFull's Y8407) in their DF27 Panel test. The above SNPs are currently phylogenetically equivalent to the SNP originally chosen by members of the subclade to represent their subclade, FGC11397. FGC11397 is also available to test individually at YSEQ.net.

The kits on the Big Tree and YFull's tree are those who have taken expensive NGS tests and submitted results for analysis. They represent only a very small proportion of the total number of Rox2 matches distributed across several FTDNA projects. Most have no SNP results, NGS or otherwise. Comprehensive NGS tests, like FGC or Big Y, taken in conjunction with 111 STRs, are the ultimate hi-res DNA genealogy combo at the moment. Those using this combination of tests are taken as near to their most recent Rox2 SNPs and matches as is currently possible. As more matches with NGS tests share their results and appear on the phylogenetic trees like the Big Tree, much more complex SNP-defined branching under Rox2 will be revealed.


If you have a NGS test, there is currently no need for further SNP tests. If you know you are a Rox2 STR match, there is no need for a 'backbone' SNP pack. SNP pack tests are less expensive than Big Y/NGS. However, when SNP testing for the first time in the dark (i.e. unaware of clues from off-modal STR markers that might point you to Rox2), at least two FTDNA SNP packs are needed just to eventually get to a basic Rox2 SNP identification. The cost of SNP packs begin to approach the cost of getting a much better and higher resolution Big Y/NGS test in the first place.

There is more branching yet to be identified below the subclade's shared block - NGS testing is not yet extensive. The more limited DF27 pack chips require SNP information to be added and regularly updated and there is no guarantee that one's own relevant SNPs will be on an existing pack or panel test.

In contrast, NGS tests like Big Y highlight a Rox2 match's own dozen-or-so variously named 'family', 'private', 'novel' or 'singleton' SNPs, i.e. ones specific to your own particular lineage stretching back from you. NGS results have the potential to fill out new or existing branches from your own particular lineage when shared on phylogenetic trees. It is their presence on these trees that might lead to new SNPs being represented on any future pack tests. With a NGS test you are investing in, and adding to, further understanding of the subclade. With SNP packs you are relying on the possibility that previously tested lines might match yours.


Rox2 matches should join the FTDNA R1b-DF27 and Subclades Project. Recent changes include the upgrade of Big Y information and improved ZZ12 subclade grouping. Some group names are said to be temporary.

Group 2fc. contains Z34609+ results. Z34609 is an ancient SNP below ZZ19. Group 2fcb., further down the results page, contains Z34609>BY21094 kits - BY21094 (hg19 7811558 C>T) is an ancient subclade just below Z34609, parallel with Rox2's ancient subclade, Z2571.

Group 2fca. (for Z2571) contains kit 617836, who is a clear Rox2 111 STR match. I assume this is a basic SNP result from a 'backbone' SNP pack that does not contain Rox2 SNPs below Z2571.

Group 2fcab. contains kit 10479 (YF06513) who is Z34609>Z2571>FGC11380* (positive for FGC11380/Y8841 and negative for FGC11369/FGC11397 and all other SNPs in Rox2's SNP block). FGC11380* is parallel with Rox2. Another NGS match with kit 10479 is needed to form a branch with shared SNPs in common. YFull: R-Y8841: FGC11385 * FGC11384 * FGC11380/Y8841.

Group 2fcaba. contains Rox2 matches who have tested positive at FTDNA for one of the equivalent block's defining SNPs, like FGC11397, FGC11394 or FGC11369. Some who were once in this group but had their DF27+ or FGC11397+ etc. results from YSEQ, BritainsDNA or FGC appear to have been removed and are now elsewhere in the results pages. If you have downstream subclade-defining SNP results from a source external to FTDNA, enter the most recent SNP in your FTDNA info panel
(in Account Settings>Earliest Known Ancestor) and let the admin know. Rox2 matches not under 2fcaba are: Page 1, E14895*, 228365, U2408, 60417*, 203123, 583270*, N131933, 645879*, 379979, 449235*, 267927. Rox2 kits on page 4 include: 202611, 241356*, 438361*, B106439, 429667, 246072, 122148*, 533969, 92381*, 433304, B221420, H1145, 279300, 300626*, 440, 134765*, 328073*, 726794. (* = 111 STR markers). Do not worry about being moved to different areas, you are still Rox2 matches. If you are unsure your kit is a Rox2 match, get in touch here via the contact page.

For clarity, all Rox2 STR matches could be together. It is possible to group Rox2 STR matches and predict SNP-defined subclades but FTDNA do not currently allow the creation of new subclade projects below DF27. Things have become more SNP oriented but there is still little doubt of a 67 STR marker DF27+ Rox2 match and practically no doubt whatsoever about one with 111 STR markers.

Many more Rox2 STR matches exist but have yet to realise they match, test SNPs, or find and join the DF27 project. If you are a Rox2 match let your close matches know about Rox2, its SNPs, this page - and join the DF27 project. For many it's the only way they get to hear about the subclade they match. You can belong to as many FTDNA projects as you like. To get an email address, c
lick the name of your match on your FTDNA Dashboard's 'Matches' page.

Below group 2fcaba. the kits are being organized into their respective subclades below the shared Rox2 phylogenetically equivalent block. Here the Big Tree, FTDNA 'Haplotree & SNPs' and YFull's tree may differ.

Group 2fcaa. is for CTS11567 (also known as Z2572). It is an ancient parallel 'brother' subclade to FGC11380, below Z2571. With the addition of Big Y-tested kits 370005 and 46496 to the Big Tree in November 2016, a deep and ancient subclade just below CTS11567 and parallel with DF84 formed, named BY3865, group 2fcaab. (hg19 position 14878305 T>C). DF84 kits are now below 2fcaaa. Downstream of BY3865 is BY3873 and these kits are in group 2fcaaba.. A third subclade, hg19 18777343 G>C/hg38 16665463 G>C, appeared on the Big Tree in December 2017.

So far, BY3865 looks to have a northern European distribution and DF84 has a southern European distribution. A 'north/south' geographical split appears to have formed at a very early point in DF27's existence. Many more NGS results are needed to see the distribution below Z2571. In February 2016 new sample, ERS257013, with ancestry in the south of Sardinia, to the YFull tree helped form a new branch (Z2569, containing the two South American 1000 Genomes kits) below Z2568, under DF84. This Sardinian kit comes from the Francalacci et al 2013 study of almost 2000 men with roots in Sardinia, in the Mediterranean. It seems to indicate DF84 is an ancient SNP, parallel with BY3865, with two branches (Z2568* and Z2569) springing from it so far. Kit 319215 is CTS11567*, indicating more branches are yet to be discovered downstream. DF84 is not in the Geno 2.0 test but it can be ordered separately. The DF84 kits, parallel with BY3865, below CTS11567, are from the 1000 Genomes project. A couple of Big Y or NGS-tested DF84 kits uploaded on the Big Tree would help develop branching below that subclade of CTS11567. The 2015 'backbone' R1b-M343 SNP pack omitted DF84.

An anonymized spreadsheet containing 2000 Chromo2 results was released by BritainsDNA in 2014. I found two kits (1525 and 1903) were CTS11567+ and DF84- but one (1525) was CTS11567+ and S25893+. Therefore, S25893 possibly indicates another subclade below CTS11567/Z2572, parallel with DF84. S25893's hg19 position is 23843638 G>C.

I made a draft SNP tree in order to visualize Z2571 in relation to other subclades of DF27. NGS testing is identifying many interesting new subclades below DF27>ZZ12 that were omitted/undetected by all previously available (pre-2016) chip-based tests, like 'backbone' SNP packs, Geno 2.0 and Chromo2.


See Distribution Maps. There is no firm evidence of a place of origin for Rox2 so far. Future ancient yDNA studies might shed some light on the matter or at least give better clues about the early geographical locations of various DF27>ZZ12 subclades in the Late Neolithic/Early Bronze Age. Specifically, it might be hoped that archaeologists will one day study a site that is found to contain the graves of Rox2 relations from the Neolithic, Bronze Age and Iron Age.

Recently, burial I2416 (Olalde et al. February 2018), one of the Boscombe Bowmen, near Amesbury, Wiltshire, England was found to be L151+/P310+, and FGC11381+. FGC11381 (hg19 6720487 C>T/hg38 6852446 C>T) is one of Rox2's dozens of equivalent SNPs in the shared block. The ancient sample was of very low quality compared with others in the study, so there is caution about the result. At the moment there is nothing in the data to rule out I2416 being an early DF27+ Rox2 match but there's not enough to say for certain either. The match could be simply ruled out if one of the SNPs (P312>DF27>ZZ12>ZZ19>Z23409) currently missing from I2416's .bam file between P310 and FGC11381 are found to be negative in further testing. Rox2's phylogenetically equivalent block, of which FGC11381 might be one of the earliest SNPs, formed 4300 years before present (+/- a big margin of error: YFull). The dates fit.

Rox2 is ZZ11>DF27. The first, and earliest, ancient ZZ11>
DF27 man to be identified so far was buried in what is now the middle of Germany
c. 2300 BC (2431-2150 BC). Mr. I0806 had a Corded Ware shaft-hole axe and a bell beaker and his autosomal DNA suggests origins to the east/Baltic. GEDmatch Kit Number: T253390. Corded Ware and Bell Beaker cultures coexisted side by side in Germany and I0806's mixed cultural possessions prompted researchers to test his DNA. He was alive not long after the founder of all ZZ11>DF27. ZZ11 is also the 'father' of the large more easily identified ancient subclade, U152 and therefore 'brothers' ZZ11>DF27 and ZZ11>U152's places of origin are going to be close by. The oldest currently known ancient R1b-P312>ZZ11 remains are RISE563, who is U152+. RISE563, found in what is now Osterhofen-Altenmarkt, Germany, belongs to 'Bell Beaker East' and dates to c. 2542 BC (2572-2512 calBCE). Isotope analysis shows he was a migrant to that area. RISE563's autosomal DNA plots with Corded Ware and modern day eastern Ukrainians, Kargopol Russians and Mordovians. Therefore, ancient DNA currently suggests the earliest origins of U152 and DF27 could be in the prehistoric eastern European Corded Ware and Bell Beaker cultures, not the west. It is thought those peoples previously came from the Yamnaya Culture even further to the east, on the Steppe.

There is modern DF27 from the east of Europe and in Armenia and DF27>ZZ12>ZZ19 is a large and extremely widespread subclade. ZZ19, the branch Rox2 descends from, appears to have a presence all across Europe, from the north, east, south and west. Recent (July 2017) NGS kit from Armenia, via Iran, 657355 Aviet, belongs to the same ancient branch as Rox2 (DF27>ZZ12>ZZ19>Z34609). Kit 657355's new subclade, also containing N123269 Serra from Spain, at the opposite end of Europe, is led by hg19 7605936 C>G/hg38
7737895 C>G from a phylogenetically equivalent block of seven SNPs. Z34609* kits come from several different countries, including Czech Republic, Italy, Iberia and Germany. Given Rox2's lengthy phylogenetically equivalent block (bottleneck) and incongruity in western European yDNA databases, its ancestry could ultimately trace back to an early central or eastern DF27>ZZ12 lineage before its arrival in northern Europe, perhaps in Late Antiquity.

Today, going by higher resolution Rox2 STR and SNP matches, the Rox2 subclade has an exclusively northern European distribution and is found in England, Scotland, Northern Ireland, the Republic of Ireland, Isle of Man, Orkney, Sweden and northern France (Normandy).
The locations of most earliest known ancestors' places of birth, when known, can be approximately split into thirds; Scotland (mainly central/south west/Borders), England (mainly north and east) and Ireland (mainly the north). Matches also turn up in numbers in southern England. The proportion of tests of Ireland/British Isles origin is high in public hobbyist yDNA databases and this affects attempts to accurately map a subclade's geographical origin, frequency and distribution using those databases. '
Heat maps' made with modern hobbyist data, not ancient yDNA, give little indication of the origin or distribution of subclades in ancient times.

Geographical holes are especially pronounced for the large number of 'invisible' ZZ12 subclades, i.e. ones chip-based tests can't see. ISOGG (used by ancient yDNA studies) has much lower resolution for downstream DF27 SNPs and other non-L21 P312 subclades. In contrast to those earlier chip-based (pre-2017/NGS) studies that produced undifferentiated P312* results for DF27, the recent (October 2017) study, Defining Y-SNP variation among the Flemish population (Western Europe) by full genome sequencing by Maarten H.D.Larmuseau et al uses NGS testing. As a consequence, that study accurately reports all DF27. Hopefully this is a sign of better things to come. R1b made up 61.5% of the Flemish study. There are 22 DF27+ samples out of the 270, or 8.14% of all Flemish yDNA. 12 of those are probably ZZ12. Of particular interest to Rox2, three of those DF27>ZZ12 Flemish men are R1b R-Z34609. This is Rox2's own deep (ancient) subclade immediately downstream of ZZ19, although none of the three are Rox2 matches.

If we were to pick an individual modern European haplogroup or subclade at random, it might have highest present-day frequency in Britain/Ireland because that is where the most DNA hobbyists (in America) trace their ancestry. It does not necessarily mean the ancient haplogroups were born in Britain/Ireland. Forthcoming ancient yDNA studies 'on the ground' are preferable when addressing population movement in prehistory. In a couple of years there should be tens of thousands of ancient genome sequences, so age and direction of movement of early DF27 will eventually become apparent. In studies of FTDNA results it looks like DF27 makes up around 1 in 10, give or take, of the R1b population of northern Europe, i.e. around 10% of all R1b in Sweden, Finland, Denmark, Germany, Poland, Baltic States, Low Countries, and England. It appears to be found in lower numbers in Ireland, Scotland and Norway. Of course, these figures are based on skewed hobbyist data and DF27 is not always clearly identified within the wider P312 population.

The distinctive yet incongruous characteristics of the subclade, combined with its estimated age and distribution, might suggest a relatively recent founding event/s, following migration by one man's descendants in northern Europe. TMRCA estimates, bearing in mind the generous margin of error, indicate this probably took place some time between the Migration Period (300- 500 AD) and the Norman Conquest at the end of the eleventh century - so there is a good likelihood that the founder was alive some time in the Early Medieval period. 

Prior to the ending of the c. 3000 year genetic bottleneck in the Early Middle Ages, the DF27 Rox2 subclade's ancestors might have lived anywhere but the bottleneck suggests Britain is less likely - otherwise we might expect to see many early branches like we see in British Isles/Ireland-heavy L21>DF13. Most Bell Beaker/Bronze Age remains in Britain have been found to be L21 in studies so far, none are identified as DF27 yet. Rox2's lengthy bottleneck begins around 2350 BC +/- a wide margin of error (YFull).  Bell Beaker was at that time expanding rapidly and to a great extent throughout the British Isles, so if DF27 Rox2 was involved in this Bronze Age event we'd expect their yDNA to leave an impression too - the very long bottleneck, evident in the British Isles/Ireland-heavy databases from the Bronze Age up to Early Medieval times, would not exist.

If, for the sake of argument, I2416 (Boscombe Bowman from c. 2350 BC) really is FGC11381+, then we would know that a Rox2 relative was in Britain at the beginning of the Bell Beaker period (Late Neolithic/Early Bronze Age). The Boscombe Bowman and the nearby Amesbury Archer stand out from others at the Boscombe Down site in that they look to have travelled there from a great distance, possibly directly from mainland Europe, as indicated by isotopic analysis of their teeth and the style of their grave goods. They also stand out as the ones with ambiguously low quality or non-existent yDNA results respectively. The peculiar few thousand year Rox2 bottleneck (no surviving early descendants) after the Bowman's purported FGC11381 mutation may support the possibility of a distant origin for I2416. Perhaps the Bowman's family, being widely-travelled non-local visitors to Stonehenge, left no descendants - or if they did leave descendants maybe they left Stonehenge for somewhere more distant and quiet i.e. somewhere 'up north'. The wider clan/family that I2416 came from might have originally lived some distance from Britain in a place that hasn't been well tested - there could have been a surviving family line there whose descendants later reappeared and flourished after around 750 AD. That place might not have any surviving lineages there today. What we can see in the data is that a distant surviving relation of the prehistoric bowman (if he is indeed FGC11381+) suddenly reappeared and reproduced in large numbers some time much later, around the Early Medieval Period/Migration Period.

The Migration Period and the later Norman Conquest in England bookend the Viking expansion and diaspora across northern and western Europe. The people of that diaspora settled in many of the same areas of Britain that Rox2 matches are now found and the timing looks to coincide with Rox2's large founding event. By the 1100s AD the subclade would have already become established in several independent family lines (SNP branches) in many different locations.


So a subclade founder appears to have been active in northern Europe in the Early Middle Ages. Of course, the Rox2 yDNA lineage existed before then but the similarity of many high resolution haplotypes today, combined with the sudden branching of several SNP-defined subclades immediately downstream of Rox2's singular and lengthy shared SNP block, suggest just one man and/or some of his sons produced a large number of descendants relatively recently. Time to most recent common ancestor (TMRCA) estimates suggest the founder might have lived about 1250 years ago +/- a margin of error. Possibly due to his circumstances, offspring of the Rox2 founder appear to have flourished in Early Medieval society - with a sudden and rapid expansion of several 'brother' branches in the ninth and tenth centuries AD. Subclades downstream of Rox2 appear to be the result of an Early Medieval Period dynastic founding event in northern Europe.

In 2014 NGS testing by two Rox2 matches at last identified the large but invisible (to commercial testing) block of SNPs below DF27. After further extensive NGS testing by many more people, that block remains intact. This lengthy list of shared equivalent SNPs, over three quarters of the total for a Rox2 match, indicates a prolonged 'genetic bottleneck'. Bottlenecks can result from a migration of an individual (or small related family group) from an original homeland, followed by a founding event some distance away. They can occur if male-line branches are 'pruned' to only one surviving individual who then goes on to have several surviving offspring. Rox2 might be imagined in gardening terms as an extremely long and thin tree trunk with uncertain roots that suddenly bursts into a very bushy canopy. It is possible that older and geographically/genealogically distant subclade 'cousins' have not been picked up yet in the mainly Ireland and British Isles-heavy American hobbyist databases. They might be found, if they still exist, in a country far from Ireland and Britain. The chances of a lineage only just surviving in its place of origin from generation to generation (one male per generation) and remaining 'bottlenecked' for around three thousand years must be quite slim. Population numbers were comparatively low 1000+ years ago and the times could be tough (violence, natural disaster, disease etc.). If Rox2 came from a small family, the other male family members could have died out in their ancestral homeland. Without the big founding event around 1250 years ago, no one would have been aware of Rox2's existence and I wouldn't be writing this.

Due to the long bottleneck and the sudden and relatively recent appearance of several downstream Rox2 subclades all at once, it is hard to say precisely where the Rox2 individual's lineage was before the Early Medieval Period - there are no definite geographical traces for thousands of years before the birth of the Rox2 founder. That might come with ancient yDNA analysis. 
This apparent rarity/incongruity contrasts with subclades downstream of 'brother' P312 subclade, L21>DF13, who do know - and are surrounded by - many of their ancient related 'cousin' branches in Britain and especially in Ireland. Most old clans/families with early Celtic/Ancient Briton origins are L21>DF13. In contrast, Rox2's shared equivalent SNP block (i.e. the few-thousand-year-bottleneck) appears to stretch from the earliest times of DF27 to about the eighth century AD, with no evident branching of cousin lines. Many more 'Bell Beaker' L21s have been found in the British Isles in latest studies. Their steppe-like autosomal DNA differs from earlier indigenous Neolithic people in Ireland but no ancient L21 has been found outside the British Isles and Ireland so far. The massive study by Olalde et al. 2018, published in Nature found no Bell Beaker L21 in mainland Europe. No ancient DF27 has been identified with certainty inside the British Isles or Ireland yet either.

There are well over one hundred different surnames at 67 STR marker resolution and above represented in the cluster. Generally surnames are thought to have only become fixed and hereditary by around the fifteenth century, long after the days of the Rox2 subclade founder. Rox2 subclade matching families usually have a name-type that is locally familiar in the region they lived (in rural England, Ireland, Scotland, Isle of Man or Sweden). Many surnames in Northern Ireland appear to have earlier Scottish roots. Surnames are not usually a reliable way of tracing back over 1000 years but can hint at the general geographical locations and origins of earliest recorded ancestors (in parish records, mostly beginning in around the sixteenth century).
The Industrial Revolution of the late eighteenth century saw populations become more mobile, resulting in movement around the British Isles and overseas to British colonies. Modern hobbyist DNA databases represent where some of a subclade's descendants were in relatively recent times. Modern subclade distributions can be quite different from their earliest origin. Archaeologists are now beginning to study ancient DNA in depth and in the next few years many thousands of ancient genomes will be tested.

The broad distribution of different, usually locally specific surnames and the subclade's relatively young age suggests Rox2 expanded quickly over a wide area (a rapid demographic expansion). The STR haplotypes of all Rox2 matches are very similar and trace back to one narrow point in time. The time and place of that expansion coincides with the diaspora of Scandinavians around the northern British Isles and Ireland. The people who were moving widely and rapidly in and around the northern half of the British Isles, including the Isle of Man and Ireland, at that time were Vikings. At this stage I would tentatively suggest the evidence may point to a connection (some time after the ending of Rox2's genetic bottleneck) with the Uí Ímair in the ninth and tenth centuries - although there are hints of potential Frankish/Flemish/Norman connections too. The earlier story and location of the subclade during the long bottleneck period is unknown. The study of ancient DNA in archaeology is not yet extensive and new data might change things.

I am from an old rural farming and sea faring North Riding of Yorkshire family that traces back to the start of parish records here (sixteenth century). The family were in the same area for centuries before then. Other Yorkshire Rox2 matches have similarly deep roots in the county. The increasing numbers of Swedish Rox2 matches are also old rural families and trace back to the beginning of parish records there, in the sixteenth century, in and around Bureå, Skellefteå Municipality in northern Sweden. As with my Yorkshire ancestors, Sweden Rox2 are likely to have been in the same locality (northern Sweden) centuries before parish records began.

Central and Eastern Europe, Western Asia and Scandinavia are not as well covered by NGS yDNA testing. There has been poor understanding of DF27>ZZ12+ subclades in academic studies, partly due to the inability of the chip-based SNP testing technology to 'see' DF27 and partly due to the DF27 subclades that are known not being on ISOGG's list. For example, the abovementioned Bell Beaker/Corded Ware burial (I0806 from Quedlinburg, Germany) was initially found to be P312+ and dated to c. 2300 BC. They were unable to recognize any downstream subclades at the time but on closer inspection in September, 2016 it was discovered that I0806 is DF27+). (link) (link2) (link3).


All results used in the TMRCA estimates are a minimum resolution of 111 STR markers. A match is based on a framework of eleven important R1b-P312 off-modal markers across 111 markers and not just on low genetic distance between the haplotypes. The off-modal pattern across 111 markers is very strong and can be used to predict SNP results. The key off-modal pattern is crucial in the identification of a Rox2 match because matches can occasionally happen between unrelated R1b haplotypes simply through coincidence (convergence) at low resolution, i.e. 37 STR markers or less. It is not unusual for a haplotype to occasionally differ on a few of the Rox2 ancestral key off-modal signature markers but the general pattern remains clear across 111 markers.

Unrelated subclades of R1b-P312 can match Rox2 STRs through pure chance at only 37 marker resolution. For example, the unrelated L21>DF13>L1335>L1065 'Scots Modal' has similar key off-modal markers to Rox2's at low resolution. Until 2008 it was thought there might be some relationship between Scots Modal and Rox2 because of the similar off-modal STR markers but the advent of L21 testing showed that Rox2 was L21-. Even with 67 markers, further DF27 SNP tests and/or an upgrade to the full 111 FTDNA markers are useful to confirm a Rox2 match beyond all doubt. N.B. As they belong to two parallel subclades of P312, there is no connection whatsoever between DF27 Rox2 and L21>DF13>L1335>L1065 'Scots Modal' in around 5000 years.

Many more potential matches exist at 37 marker resolution but their relationship to Rox2 is impossible to confirm for many of them at such low resolution with no SNP tests.

As the bar chart here illustrates, plotting Rox2 matches' genetic distances from the modal/base haplotype produced a clear bell curve. This indicates the off-modal STR markers work well in identifying common descent from a founder, even without the SNP confirmation we have now. Rox2 is a monophyletic clade.

KEY ROX2 OFF-MODAL MARKERS OVER 67 MARKERS:    DYS391=10 (P312 modal=11)    DYS389ii=30 (29)    DYS449=30 (29)    DYS607=14 (15)    DYS534=14 (15). The 68-111 marker section in the FTDNA test holds another six important key off-modal STRs for Rox2, including: 540=13 (12), 717=20 (19), 589=11 (12), 636=11 (12), 532=14 (13), 504=16 (17).

An early key defining STR marker was found to be DYS717=20It is included in the 111 marker upgrade. Some who matched Rox2 at 67 markers have ordered this individual marker from FTDNA under 'Advanced Orders'. DYS717 is strongly off-modal in Rox2 and is considered a stable, slow mutating marker. For reference, R1b modal is DYS717=19, L21+ 'Scots Modal' is 717=21. Results matching the above extra key markers (and the 67 marker ones) make identification of a Rox2 haplotype quite unambiguous.

The numerous Rox2 haplotypes are recently related enough to have retained identifiable similarities with other distant relatives in the yDNA database. Rox2 can be recognized by the solid off-modal 'signature' STR pattern - one common to all matches over 111 markers. This relationship can be confirmed with SNP tests.

Less uniform key off-modal markers in the 68-111 panels: DYS712 <=20 (21), DYS714 >=26 (25).

The Rox2 modal/base haplotype has 6 out of 67 off-modal differences from the average for all R1b-P312. Rox2 has 16 differences over 111 markers from the R1b-P312 modal. It is a genetic distance (GD) of 9/93 from Ysearch ID: XQJ7H (R1b-P312 (S116) and all Subclades Modal).


Join the FTDNA DF27 and Subclades Project and encourage your 67 and 111 marker matches to do so too - even if you have not tested SNPs yet, they are likely to be DF27+ (the DF27 project admin encourages this). When Big Y/SNP tests are taken your kit will be placed in its specific subclade group.

In order to get maximum information out of Big Y or FGC tests it is important to compare with other results on phylogenetic trees, like the Big Tree (free) or YFull's tree. This not only benefits one's own research, it also advances understanding of the subclade. Little can be learned without comparison with the results of others. If your particular subclade below Rox2 has few members, write to suggest Big Y/NGS testing to your closest STR matches - they have a chance of being near you on the Big Tree. YFull will analyse NGS test results for you and add them to their tree for $49. Full Genomes (FGC) will analyse BAM files for $50. Make sure you know the subject, what to expect, and your goals before testing; i.e. chose the type of test that can help achieve that goal. Unfortunately DNA genealogy is not a 'one test answers all' process and for best results it requires as much active research as traditional genealogy.

Don't be put off by the large number of close matches with different surnames. The relatively high number of such matches is common for all Rox2, the subclade appears to have been prolific in the Middle Ages. In most cases the yDNA connection is many centuries ago - often before surnames became fixed. Write to close matches on your FTDNA myDashboard 'Matches' page using the email addresses there. Let them know about this subclade, its SNPs and Big Y/Big Tree. If you test SNPs at YSEQ, enter the most recent SNP in the information panel for your FTDNA kit (in Account Settings edit the Earliest Known Ancestor box to include the YSEQ SNP number). Unlike the one-at-a-time SNP testing approach with YSEQ, NGS testing, like Big Y, finds the handful of important 'family' SNPs leading to the present day.

It's useful for matches outside a surname project to also be able to view and compare results but default privacy settings for new FTDNA kits mean new STR results (since March 2015) are invisible unless one's privacy settings are changed manually. Matches within a subclade like Rox2 span a number of surname projects with origins in different countries. If you don't mind sharing STR information so that results might be seen by matches that are not in your specific surname project: sign in to your FTDNA account and at the top right of your FTDNA 'myDashboard', click your name. Then click 'Privacy Settings' next to the padlock symbol. Next to 'My DNA Results' and 'Who can view my DNA results in group projects?' click 'Make my mtDNA & Y-DNA data public.' In May 2018 EU legislation on data privacy will be adopted by FTDNA. The customer must balance any privacy concerns with genealogy goals. If results are left at 'default' privacy settings, they're invisible to the outside world. However, DNA genealogy works by those involved comparing and sharing information.

If you have a website or a blog with details about your specific Rox2 family, send a link. If you don't have a genealogy website or blog, consider creating one (e.g. Wordpress or Blogger) to share family history and subclade knowledge and to help potential SNP relatives find you in the future. It is interesting for other subclade matches to read about different parts of the Rox2 tree.

More matches can be found by uploading STR results to YSearch.org. Check these pages regularly. Further understanding will be gained with wider engagement in higher resolution STR and NGS testing and ancient DNA testing in the coming years. As the technology improves, prices should fall. Future archaeological yDNA studies from the Bronze Age, Iron Age and Medieval Period could eventually provide vital clues to the subclade's origins.


The cluster was named 'Rox2' by Jim Turner, who created the YSearch ID: 3QNM8. Jim and I yDNA tested in 2005 and found we had fairly closely matching STRs along with an off-modal 'signature' pattern shared with a few other kits in the early yDNA databases. Most kits at that time, including ours, had only 25 or 37 STR markers and no SNPs. STR and SNP resolution has increased a lot since then. The term Rox2 has stuck after initial discussions. It was, and is, a useful shorthand way to describe a subclade currently known by several different equivalent SNPs. Some of those SNPs have multiple but synonymous names due to different labs adding their own nomenclature to the same SNPs.

STR results extant between 2005 and 2012 indicated to some of us that Rox2 was a monophyletic clade but the deep R1b-P312 subclade it descends from (DF27) was unknown at that time. We tested negative for L21 in 2008 and all subsequent SNP discoveries that were made available for testing below P312 also came back negative. Many more STR matches turned up over those years, however. Proud Scot, Angus Horatio Stewart, kit 142928, was the first Rox2 STR cluster match to test positive (derived) for DF27 when that SNP was identified and became available to test in May 2012. This finally proved beyond doubt that Rox2 is a monophyletic clade. Chip-based SNP tests were still not able to see most of the emerging ZZ12 section of DF27, let alone DF27 itself, so from 2012 until March 2014 Rox2 remained a R1b-DF27** subclade, i.e. negative for all then known SNPs below DF27. It was clear to those of us who remained DF27+ and Z195- that there was a lot more to DF27 than just Z195. Despite Rox2 matches taking tests like Geno 2.0 and Chromo2, no relevant SNPs below DF27 were identified. Of the 144 DF27 kits I looked at in an early Chromo2 spreadsheet (2014), 61 were DF27*, i.e. 61/144 were positive for DF27 and negative for all other then known subclades. 

These indications, among others, showed that many subclades (ones that later turned out to be below ZZ12) were being missed by the chips used for such testsEven current tests, like LivingDNA, using the Illumina GSA chip, cannot read DF27. However, post-2017 things are improving. The FTDNA DF27 project and Haplotree at last has coverage of ZZ12 and groups kits in a way that agrees, in most cases, with what the Big Tree and YFull had discovered for NGS customers. What a difference three years of NGS testing and analysis made. On the Big Tree, in August 2017, there were more branches and children below DF27>ZZ12 than any other subclade of R1b-P312. At the Big Tree (March 2018) there are twenty-three 'sons' of DF27>ZZ12 and several more waiting to be discovered. Parallel branch DF27>Z195, identified long before ZZ12, has two 'sons'.

C. Corner. April 2018.