Rice Family Y-DNA Project Highlighted technical terms are explained in the Glossary. Note: a term may be used many times but is highlighted only the first time within each section. Last updated: 2024 Sep 3

Project Summary

• This project seeks to discover the ancestry of the Rices, including those with name variants such as Royce and Reese, using patrilineal DNA testing as a supplement to conventional genealogy. The project is open to all males with the surname Rice or any variant and to males whose surname would have been one of these except for an adoption or other name change.
• Although DNA testing has been decisive in settling a number of genealogical questions, some amount of conventional genealogy is necessary to help identify a participant's place within one of the Rice families.
• The DNA testing is relatively inexpensive.
• The New England Rices belong predominantly to two large families. The Edmund Rice family (Group 1) is the larger of the two and also has the most members in our project of any group. The other is the Robert Royce family (Group 3). These two families have been, and still are, frequently confused, since the name variants "Rice" and "Royce" appear widely in both families. However, the two are genetically distinct and have no common genealogy.
• Several other New England Rice families have participated in the DNA project: Group 8 (the lineage of William Marsh Rice, founder of Rice University), Group 15 (Rices of Rhode Island), and Group 17 (a family radiating from Dedham, MA). The other New England Rices have not yet joined in sufficient numbers to be identified as separate groups.
• Southern Rices constitute the largest number of groups. Some eleven distinct Rice families of the South have been identified via matching Y DNA and varying levels of conventional genealogy. The conventional research can now focus on searching for the links within each group. The first of these groups to be identified was Group 4.
• A large number of results are in separate listings (Group 2 and Group 99) because they have no matches or only matches of close relatives such as brothers. From time to time some are matched and thus become a group. Only six samples from outside of North America have been analyzed, and most of these are unmatched. As more such samples are included, more matches should eventually be found.
• We have found numerous matches between participants named Rice and others named Royce, but we have not (yet) found any clear matches between Rices and Reeses or between Royces and Reeses.
• Twelve markers can tentatively place a sample when combined with conventional genealogy, but 25 markers and in a few cases 37 markers are necessary to increase reliable group identification.
• Most of the American Rice families have not been traced to their European sources. Family traditions often point to England, but some indicate Ireland, Germany, or Wales. DNA testing can sometimes shed light on origin questions. For example, the DNA patterns of Edmund Rice and Robert Royce were almost surely not of Welsh patrilineal origin, but other Rice lines could conceivably be Welsh. Indeed, both Edmund and Robert have been traced back to England as their immediate place of origin, but the DNA results go further and address their "deep" roots, albeit with some uncertainty.

Table of Contents

• Project Summary
• Introduction
  • Privacy Policy
  • Ancestors of Edmund Rice?
  • Descendants of Edmund Rice
  • Edmund Rice (1638) Association
• Origin of the Rice Name
• Contacts
• Results
  • Nomenclature
  • Data Format
  • Haplogroups
• Test Verification
• Other Relevant DNA Projects
  • Genographic Project
  • YHRD
• Notes and Discussion
  • Group 1
  • Group 2
  • Group 3
  • Group 4
  • Group 5
  • Group 6
  • Group 7
  • Group 7A
  • Group 8
  • Group 9
  • Group 10
  • Group 11
  • Group 12
  • Group 13
  • Group 14
  • Group 15
  • Group 16
  • Group 17
  • Group 18
  • Group 19
  • Group 20
  • Group 21
  • Group 22
  • Group 23
  • Group 24
  • Group 25
  • Group 26
  • Group 27
  • Group 28
  • Group 29
  • Group 30
  • Group 31
  • Group 32
  • Group 33
  • Group 34
  • Group 99
  • Mutation Rate
  • 111-Marker Testing
• DNA Data
  • Table 1 (1-25)
  • Table 2 (26-37, others)
  • Table 3 (38-67)
  • Table 4 (68-111)
• Glossary

Introduction

The Edmund Rice (1638) Association has a project underway seeking to discover the ancestors of Edmund Rice and other Rice and Royce families. (There was a tendency in past centuries to regard these two spellings as interchangeable.) Y-chromosome DNA analysis offers exciting opportunities to learn more about early family roots. With the assistance of three genetics testing labs, we compared the DNA of many male-line descendants of Edmund Rice of Sudbury and Marlborough and reconstructed the genetic "fingerprint" or haplotype of our immigrant ancestor (see Table 1). For the details of how we did this, see the article entitled "How We Obtained the Rice Haplotype" in our newsletter.

Knowing the haplotypes of Edmund Rice and several other progenitors, we can now invite all Rice/Royce males to compare your DNA against Edmund's and against each other's. For those of you who wonder whether you may be his descendants, such a comparison can help to investigate that possibility. A match with Edmund's haplotype will confirm that you are indeed related (though not necessarily a descendant) and will encourage and aid you in further genealogical research to discover your Rice ancestral line. We have on line a portion of the Rice family history (Edmund and five generations of his descendants) to help in finding the connection if you do match. On the other hand, a big difference from Edmund would indicate you are not related to him at all, but might reveal a similarity to other Rices who are related instead. For male Rices who already have reason to believe you are not Edmund Rice's descendants, the comparison may reveal whether or not the separate Rice families have some connection back in the British Isles. In either case, it should be clear that some conventional genealogical research will be needed in order to get the most out of the DNA results. Table 1, Table 2, Table 3, and Table 4 below have all the DNA results we have obtained to date.

You, too, can participate. We have arranged with FamilyTree DNA (FTDNA) to offer a 12-locus DNA analysis for a reduced rate of $59 to those who join our project. There is a similarly reduced rate for the expanded 25-, 37-, 67-, and 111-locus tests from FTDNA as well. However, this project is independent of FTDNA, and membership in it is neither contingent upon nor automatic with testing at FTDNA. We freely accept participants who test with other labs, and the only requirement is that the testing must cover the same Y-DNA STR markers that we use and follow reporting standards that we can convert as needed for comparison with our other results. We will need to know something of your male-line ancestry in order to make sense of any DNA matches or near-matches you may have. If your haplotype matches our reconstructed haplotype for Edmund Rice, we will be especially keen to learn more about your Rice/Royce ancestry.

For more information contact our project administrator/coordinator: email us.

Privacy Policy

When you join our project, we add your DNA results to the data tables shown below, identified only by an arbitrary code, not by your name. We do not share your name or contact information with others, unless you specifically ask to be put in touch with another participant. Your DNA test results remain yours, and you may at any time withdraw from the project by writing to us.
 

Origin of the Rice Name

The origin of the Rice surname is not known. Indeed, there are clearly many different origins. Click here to see a description of some of the most likely origins, and the relationships among them. It seems likely that the Rice name may blend into other spellings -- not just Royce, but also others, including Reece, Rease, Ris, Ries, and even Price. In other words, the participants of this project need to keep in mind the possibility that their closest genetic relatives may bear a different form of the name. Therefore, if you have no close matches within the Rice DNA project, you may wish to participate in other projects as well, particularly the Reece project and/or the Price project. Similarly, if you are a Reece or a Price with no close DNA matches within your own name, you may wish to join the Rice project.
 

Contacts

More information about this project: email us.

For information about the Edmund Rice (1638) Association (and how to join, should you care to do so), please visit our membership page.

There is also a Royce Family Association. You may visit their web page to learn more.

The Reece DNA project has a web page where you may see the test results. Some, but not all, of the members there are also in the Rice project.

The Price DNA project also has a web page.

Results

At the end of this page, we present the DNA test results we have obtained so far. As additional information becomes available, members of the "other" group may be separated out into new groups with identified common ancestors. In cases of ambiguous DNA results, we will depend in part on lineages supplied by the test subjects for determining how the groups should be constructed.

In Tables 1 through 4, each line begins with a unique ID. The 4-, 5-, and 6-digit ID's refer to FTDNA results. ID's consisting of "N" and a number refer to members of the Genographic Project (tested by FTDNA) who have also joined our project. (See also below under Relevant DNA Projects.) 4-character ID's beginning with the letter "S" refer to Sorenson Genomics/Relative Genetics results. 4-character ID's beginning with "sm" refer to haplotypes found in the Sorenson Molecular Genealogy Foundation Y-DNA database. All other participants are represented by 5-character alpha-numeric ID's designed not to conflict with the above conventions.

Nomenclature

The haplotypes are all presented in a consistent nomenclature. For example, pre-2003 FTDNA results for the DYS464 complex have been adjusted by -1, in accord with their change in reporting standards on 2003 May 19. Similarly, pre-2004 results from Relative Genetics have been adjusted by +1 at DYS461, in accord with their change on 2004 June 29. (Our DYS461 result for 1673 came from Relative Genetics in early 2002 and is the only one affected by this particular change.) ID 1673 in Group 1 is a case in point: the same person was tested at three labs (FTDNA, Relative Genetics, and Oxford Ancestors), and the tables include the results from all three. In addition, the pre-2008 SMGF haplotypes have been adjusted by -1 at DYS19, -3 at DYS448, +1 at DYS461, and -1 at Y-GATA H4. The topic of standardization is a thorny one for genealogical DNA testing, since all of the testing companies have changed their reporting standards at one time or another, and they all have disagreements with each other about these standards. At present, the haplotypes in this project are shown according to the reporting standards of FTDNA, since much of the testing was done there.

Data Format

In Table 1, we show only the results for loci included in the FTDNA 25-locus test. Other loci from various sources are shown in Table 2. 30 additional loci from the FTDNA 67-locus test are shown in Table 3, and 44 more from the 111-locus test in Table 4. If an entry in the tables is blank, that means the corresponding locus has not been tested. The only exception is the DYS464 complex, which normally has four values reported as a-d, but which may have more or fewer (we have seen 3, 4, 5, and 6 values in members of this project). Loci that have been tested and found to be "missing" are indicated by a minus sign (except DYS494 as noted above). Note that a "missing" locus need not signal a missing segment of DNA -- it may merely indicate a mutation in the flanking region such that the conventional primers no longer amplify that segment.

The reconstructed ancestral haplotype, if known, of each group is given as the first entry in the group, with the common ancestor's name (if known) as the ID. Individual mutations from the relevant ancestral haplotype are shown in red boldface against a gray background. For loci whose ancestral value is unclear, the background of the whole column is white. We recognize that mutations are inevitable, given enough test subjects and/or the passage of enough time since the progenitor. Nonetheless, these mutations are rare, as the table shows, and the appearance of any discrepancy between the haplotypes of putatively related individuals is cause for concern. (See the discussion of Group 2.) The question is always whether we are so "unlucky" that a rare-but-inevitable event occurred right here, or so "lucky" that a rare coincidence gave two unrelated persons very similar DNA. To resolve that question, we need conventional genealogy (as we do indeed have for the individuals assigned to Group 1).

Haplogroups

Besides the results presented here, some of the participants have tested for certain additional loci with much lower mutation rates. These loci are used in studies of population genetics to define categories known as haplogroups. Since haplogroups are distributed broadly on continental scales and date back to prehistoric times, these categories are not useful for genealogy, and we have avoided displaying them here. Indeed, they would be a hindrance, since they have a notable tendency to distract viewers and participants alike from genealogical pursuits.
 

Test Verification

Subjects 1668, 1669, 1670, 1672, 1673, 4188, 5128, and 5129 all were tested twice independently, once through BYU/Sorenson/Relative Genetics and once through FTDNA. We therefore have cross-checks for 73 of the numbers in Table 1. Unfortunately, there are systematic calibration questions that complicate the comparison for 25 of those numbers. However, 48 of the numbers are straightforwardly comparable, and we find that 46 of the 48 agree between the two labs. This is moderately encouraging, though by no means as good as we expected. In cases of discrepancy, we show the FTDNA result in the table.

Subjects 4765, 15824, and 9uxtm have also been tested twice independently, once through a commercial lab and once through SMGF. In addition, N3385 has been tested three times, twice through commercial labs and once through SMGF. These additional redundant tests give us cross-checks for 131 more numbers in Tables 1 and 2, and no discrepancies.

Subject B2341 has also been tested twice, but the overlap was only fifteen markers, since he took advantage of a transfer option to cover the first 12 FTDNA markers. Again, no discrepancies.
 

Other Relevant DNA Projects

Besides the many other surname-based DNA studies, there are some projects of global scope that include, or may include, Rices and other variants. First of all, there have been public databases of user-contributed test results, such as YSEARCH (which is no longer maintained and will probably be taken down in the near future) and YBASE (long since taken down). Each of these databases contained test results that members of our project uploaded. YSEARCH even included the reconstructed 42-marker haplotype of Edmund Rice. These databases were searchable by haplotype and by surname.

Another database that contained Rices was maintained by SMGF. This database was initially searchable only by haplotype, but it allowed surname searches as well starting in August of 2005. It also had a list of the surnames included and the number of each surname. Initially, there were 4 Rices included in SMGF, and our searches discovered one of them by his haplotype (sm04 in Group 7), in addition to three related men whose family surname is no longer Rice. When the search by surname was implemented, we found the remaining three Rices and eight others that have been added since then. Unfortunately, this database was also taken down, having been sold to Ancestry when the foundation ran out of money.

A fourth database known to contain Rices is that of the Genographic Project conducted under the auspices of the National Geographic Society. This project is not a genealogical one, but is aimed at population studies, such as tracing the paths of human migration over tens of thousands of years. The main thrust of this project is the analysis of isolated indigenous populations around the world, but another component is a study open to the public. Participants in this public study are given their choice of the FTDNA 12-marker Y-DNA test (males only) or the FTDNA mtDNA HVR1 test. In either case, the measured haplotype is used as a predictor of the participant's ancestors' migration routes. After these initial results are returned, the participants are routinely invited to follow up with the genealogical aspects of DNA testing by joining surname studies at FTDNA, and several such persons have joined the Rice project. (Note: the link to the Genographic Project may not work for some web browsers.)

A fifth database of global scope, called the YHRD database, includes anonymous samples from all over the world. This database is aimed at forensic DNA research, rather than genealogy, but it is genealogically useful nonetheless for indicating where a given haplotype can now be found.
 

Notes and Discussion

Here, we discuss each group of test subjects in turn. One clear message from all of this is that a correct understanding of the relationships requires some conventional genealogy to go with the DNA test results. See especially the discussion for Group 2. You may click on the heading for each group's discussion to view the data for that group in Table 1, just as clicking on the group number within Table 1 brings you back to the discussion in this section.

Group 1

As indicated already, Group 1 consists mainly of male-line descendants of Edmund Rice of Sudbury and Marlborough. This was the first group set up for study and remains the largest. Initially, we focussed on known descendants of Edmund, but, as the project has grown, we have tested more and more subjects who have not been able to trace their lineages back to Colonial times. Some of these have proven to match Group 1 while others have not. In some cases, the conventional genealogical proof of descent from Edmund has been found only after the DNA testing was done. At present, no member of Group 1 has found evidence that he is not descended from Edmund, but that remains a possibility, particularly if and when we find a Group 1 member outside of North America. The following members of Group 1 are still researching their ancestral lines: 5300, 6220, 13364, 28544, 39845, 40252, 42343, 86189, 96421, 103060, 118799, 141258, 143852, 156292, 158651, 178478, 182789, 190787, 192825, 210847, 238944, 241673, 295556, 311925, 314300, 319652, 337486, 403390, 431075, 501199, 845433, 859783, and 899681. Given the new motivation provided by the DNA evidence, it seems likely that some or all of them will eventually find solid conventional links to the rest of the group.

Three members, sm02, sm03, and sm06, were not tested as part of this project, but were instead discovered in the Sorenson Molecular Genealogy Foundation Y-DNA database by searching for the Edmund Rice haplotype. These are a father and two sons, and their surname is not Rice, but rather King. Their ancestry goes back to Edmund Rice via Edmund's grandson Samuel King alias Rice. (See the story of the latter's name change in the database of early descendants of Edmund accessible from our main page, where he is listed as "Lt. Samuel Rice King".) Subsequently, two more Kings were found there with this haplotype, both also turning out to descend from Samuel King alias Rice. Additionally, two other Kings from this line have come forward and been tested directly as part of the project.

Two other members, 13364 and 24143, have a rather unconventional link in addition to the DNA evidence. In 1704, four Rice boys descended from Edmund were captured by Mohawks at Marlborough (later Westborough), Massachusetts, and carried off to Canada. One was ransomed, but the other three remained and were adopted into the Mohawk tribe. Many years later, one returned to visit Westborough, but he no longer spoke English and had to talk to his relatives through an interpreter. The contact was not maintained, and so there is no collected record of the descendants of these expatriates. Nonetheless, Rice remains a relatively common surname among the Mohawks to this day, and both of these participants are Mohawk Rices. The two match each other 25/25 and match 24/25 with Group 1. 24143 traces back in the male line to one of the Rice captives and thus in turn to Edmund. (However, his lineage was not completed until after we had obtained his results for the first 12 markers. This research was a daunting task.) The fact that the two match each other and share the same surname within the same tribe is compelling evidence that 13364 is closely related and suggests that he may descend from the same captive, but his lineage has not yet been traced all the way. More recently, another Mohawk with a surname other than Rice has been discovered to share the same haplotype, but his lineage has been trace only back to the 19th century. Pending further research, we are assuming that he also descends from one of the captives. For comparison, we note that two other Mohawk Rices have also been tested and show a haplotype very much unlike Edmund's. These two form Group 29.

A tree diagram of the 25-locus haplotypes in Group 1 shows the majority haplotype at the center and thirteen mutated branches radiating from the center -- albeit with some questions about the order of mutations along branches proceeding two steps from the center. This is a typical "star-shaped" diagram indicative of evolution from a common ancestral type. Not surprisingly, the diagram becomes even more complicated with the inclusion of the extended haplotypes shown in Table 2, as 1671 and 1673 share a forked spoke with one mutation in common and one separate mutation each. Interestingly enough, the shared mutation (CDYb=38) can be pinpointed uniquely to Thomas Rice, son of Edmund. This is the first mutation in our project we have been able to tie to a specific individual.

As of 2019, we had upgraded 24 members of Group 1 to the 67-locus test introduced by FTDNA in the spring of 2006. These descend from at least five different sons of Edmund Rice, and they form a clear consensus. Thus, their predominant agreement on those 30 "new" loci gives us 30 additional loci of the reconstructed ancestral haplotype. Further, we have upgraded 18 of these members to the 111-locus test introduced in 2011, and the consensus has therefore now reached a total of 110 markers. One of the new markers, DYS712, appears to be particularly variable in Group 1, Twelve of these 20 testees have lineages known all the way back to Edmund, representing five different sons and eight different grandsons. These are distributed 1-4-3-2-2 among the five different results found. All four results having more than one instance each include descendants of at least two different sons of Edmund, and so all four of those results are each arguably the ancestral value for DYS712. At present, the best we can say is that result with the most instances also has descendants of three different sons of Edmund and is therefore the favorite candidate, but the statistics are too slim to make a final call.

This variability may stem from the values being at the high end of the observed range for DYS712, where we might expect to see a higher than usual mutation rate. In contrast, the eight tested members of Group 3 all agree on a single (much lower) value for DYS712.

Group 2

Group 2 was originally created on the basis of a single sample, mainly because of a lesson it had to teach. However, we have subsequently found other samples that teach the same lesson and have added them to this group, even though there is no evidence at present that they are related to each other. Consider the haplotype of sample 3109, the orignal member of this group (but now part of Group 23). It is sobering to compare this haplotype with that of Edmund Rice. In the first 12 loci (the basic DNA test), these two differ by just one step at one locus, but the additional 13 loci of the expanded test reveal three more differences, including two differences of two steps each. (The contrasting color code in Group 2 shows the loci that differ from Edmund Rice.) Looking only at the basic test, one would be tempted to conclude that the two lines are closely related, but the expanded results show that notion to be false. It is clear now that 3109 is unrelated to Group 1.

The same hazard is lurking for other testees, as we discovered. Therefore, it is often necessary to upgrade to the expanded test when the basic test indicates a possible match. Without the backing of conventional evidence, the DNA results can obviously be misleading. A similar problem (and the same solution) can be seen in the Chandler study.

The conclusion we must draw from Group 2 is that sheer coincidence may yield near-matches with the same or similar surnames, based on the existence of a large pool of vaguely similar haplotypes. This vague similarity does, in fact, indicate a relationship among the test subjects, but that relationship apparently goes back thousands of years and falls outside the realm of genealogy. We now discuss the rest of this disparate group.

Although Group 2 is much smaller in numbers than Group 99, it has the same "catch-all" nature and sometimes has clusters of individuals who match each other but cannot be considered a separate group because there is not yet any hint of a genealogical connection or because the known connection is trivially close (e.g., brothers). Such clusters are shown at the beginning of the group with a small gap separating each cluster from the next and from the rest of the group.

N1892 is a 11/12 match with Group 1, being only one step away at 12 markers. However, he shows three additional differences at 25 markers, including one two-step offset. With only five steps of difference at 25 markers, he falls into the "gray area" for 25-marker comparisons (more than two, but fewer than six steps of difference). However, there is another consideration in his case: he has traced his ancestry to a presumed German immigrant, whose Reiss surname was Americanized to Rice in 1840. Furthermore, he has five additional mismatched markers at 37. It is therefore clear that the seemingly shared surname is really a cross-language coincidence, and the connection probably dates back before surnames were used. N1892 has at least 26 exact 12-marker matches in the FTDNA customer database, including seven with Finnish email addresses and a like number with American email, but Finnish-sounding names. This, by itself, is far too small a sample to yield any statistical inferences, but the FTDNA ethnic origins database shows that he is an exact 12-marker match with nearly 8% of the samples tested with Finnish ancestry, plus 1% of those with Swedish ancestry and no other country-wide match even approaching 1%.

22348's results are only five steps away from some of the members of Group 1 at 25 markers (and only two steps at 12 markers). As such, he falls (barely) into the "gray area" with respect to those individuals. (For 12-marker comparisons, the "gray area" is more than one, but fewer than four steps of difference.) However, comparing a modern individual and a reconstructed group progenitor is much more stringent than comparing two modern individuals -- there are two separate lineages where mutations could accumulate in the latter case, but only one lineage in the former. Thus, we can be quite confident that 22348 is not a descendant of Edmund Rice, just as in the cases of the other members of Group 2.

A tree diagram of Group 2 would be a hopeless snarl, with duplicated parallel branches criscrossing each other and leaving many gaps unoccupied by actual samples. Such a pattern indicates a very much older and larger population that is only sparsely sampled. This point re-emphasizes that the members of this group are not related within a genealogical time frame.

Group 3

The members of Group 3 are identified as male-line descendants of Robert Royce (c1606-1676) of New London, CT. Five of these have been traced back to Robert Royce with reasonable certainty by conventional means, seven others tentatively, seven are not yet documented, and three others were previously thought to be descendants of Edmund Rice. However, the near-exact match of all the samples that have been tested on 25 loci makes it seem likely that at least these are closely related, and probably all of them. Since the testees with documented genealogies back to Robert Royce agree at least 24/25, and include exact-match descendants from two different sons of Robert, the group as a whole is now quite firmly established, even though the other members have not traced their lineage back to the founder. We have therefore reconstructed the ancestral Royce haplotype and included it in the table. In any case, this group shows the same sort of "star-shaped" tree diagram as Group 1: a central consensus haplotype with several adjacent ones showing one or two members each.

Three members of Group 3 have been extended to 37 loci, including two members who trace back to two different sons of Robert Royce and a third member who has tentatively traced back to yet another son. All three agree exactly on the last 12 loci.

Two members of this group, a father and son, submitted their samples and pedigrees to the SMGF project, where we found them and designated them sm18 and sm19. We have not yet verified their lineage from Robert Royce (and we therefore consider them "tentative" genealogically). They are also tentative genetically because they have four differences from the Group 3 haplotype out of the first 25 markers. Nonetheless, they match on all the rest of the markers (18 more in all), and it seems most likely that the four differences are simply a statistical fluke.

More recently, a participant (100323) has been found who matches them on all but one marker among the first 25, including three of their four mutations. Two of these three mutations are shared with one or more other members of the group, but not both at the same time. In any case, 100323 is closely related to sm18 and sm19 according to the stated lineages, and the fact that he shares only three of the four mutations with them improves the fit of all of them into Group 3.

Shared mutations

Just as in Group 1, where 1673 and 3111 share both a common line of descent and a mutation from the ancestral haplotype, we see three shared mutations in Group 3: DYS19=15->16, DYS385b=16->17, and DYS393=14->13. In principle, the established lineage of one member of such a cluster can serve as a useful clue for any others lacking a clear lineage.

However, three of the four in the first cluster have traced back to Robert Royce with no intermediate common ancestors, and therefore this clue is contradictory for the remaining member. Of course, such clues do not always bear fruit, even without contradictory indications, since duplicate mutations can arise independently, unlikely though that is.

The second cluster led to a search for the exact timing of the mutation. In the case of 12451 and 15824 (and 15824's near kin 21622, 33906, and 33908), we now have evidence that the mutation at DYS393 did indeed occur twice independently, since 15824's fourth cousin 21032 (and third cousin 33907) lack the mutation. By recruiting all of these cousins for testing, 15824 has now demonstrated that the DYS393=14->13 mutation occurred in his great grandfather independently of the duplicate mutation seen in 12451.

The third cluster is also problematic, since two of the testees belong to two such clusters, thus assuring that at least one mutation has been repeated independently.

The second cluster is of interest for another reason. It includes a striking example of an unusual process that has only recently been found in the Y chromosome -- known variously as "gene conversion" or "recombinant loss of heterozygosity" or "doubling." Comparison of the results for 15824 and his great-grandson 33906 shows a change on DYS459a,b from 8,10 to 10,10 as well as a change on DYS464a,b,c,d from 11,14,14,15 to 14,14,14,14 and on CDYa,b from 33,38 to 38,38. We see an apparently simultaneous replacement of an "8" by its partner "10" and of "11,15" by their partners "14,14" and of "33" by its partner "38." Since all three of these markers occur on the same palindrome of the Y chromosome, it seems virtually certain that one of two possible events has taken place between the two individuals: (A) a large-scale replacement, such that a long segment on one arm of the palindrome has been converted into a copy of the corresponding segment on the other arm, or (B) the long segment was simply lost. Unfortunately, the lab tests are not well enough calibrated to detect the difference between duplicate markers and missing markers, and the lab reports always indicate duplicates in such cases. We are therefore unable to choose between the two possibilities, despite the appearance of certainty in the lab reports.

Group 4

Group 4 was created based on the exact matches at 12 loci for three samples. Since then, many more have joined, and most of them have by now extended to at least 37 loci. Quite a few members of the group have one or more discrepancies among the first 12 loci, but the overall agreement is clear. For the most part, the discrepancies overall are consistent with sporadic, random mutations, but there is a complication at DYS464: although all have two 15's, and most have at least one 17 among the four copies of this locus, there is no consensus on the fourth copy. A majority have another 17, but eight have a 14 and two have a 16. It seems likely that the 14-vs-17 discrepancy is due to a single mutation, but it is uncertain which is the ancestral value. 17 is favored by the fact that it is most common, but it is easier to postulate a 14 mutating into 17 in this family than a 17 to 14. We could resolve this question if we had clear lineages for all testees, but the relationships have not been tied down. We have therefore left DYS464c blank in the reconstructed ancestral haplotype for this group.

In any case, it seems likely that the 14-17 mutation in DYS464 occurred relatively early and that it therefore splits the group into two subgroups which could be distinguished genealogically. In fact, we have already used this splitting to characterize the discrepancies we see in Table 2 for the members of this group who have upgraded to 37 markers. Some of the discrepancies appear to be individual line mutations, but one (CDYa split 37-vs-38) appears to split the group in the same way as DYS464. That is, all but one of the members with DYS464=14,15,15,17 have CDYa=37, while nearly all of the others have CDYa=38.

Although nine of the donors have tentatively been traced back to Rices of Virginia, and eight others to Rices of Kentucky or the Carolinas (and probably to Virginia ultimately), the identification of the progenitor remains uncertain. Indeed, some other testees who do not match seem to be contending for the same progenitor (Thomas Rice of Gloucester Co, VA, c1650 - c1716). It will be necessary to test more descendants to firm up this group.

One apparent member of the group is sm09, discovered in the SMGF database. He is in the odd position of differing on three markers among the first 12 but agreeing on all the rest, being 29/34 with 4131 and 56745 and 28/34 with 17939. sm09 has been traced back to 1825 in North Carolina and thus fits in geographically with the group but cannot yet be tied in genealogically.

Another apparent member is 187671, whose Rice lineage is unknown, but who has a 65/67 match with the DYS464c=17 subgroup.

24 members of Group 4 have upgraded to the 67-locus test, and their results agree perfectly for the 30 "new" loci, except for three discrepancies of one step each. This agreement reaffirms the consistency of the group as a whole. 16 have further upgraded to 111 loci, again with near-perfect agreement.

Group 5

This second Virginia Rice group is a similar collection of shared and almost-shared ancestry, bound together by the matching DNA. Until recently, the arguable common ancestor of the whole group was one William Rice born c1756, thought to be the father of three men known to be the ancestors of five of the members of this group. No other documented ancestors of the group were early enough to challenge him. Now, however, other lines have been discovered going back to contemporaries of William who may have been his brothers. If so, the new candidate common ancestor would be their father, possibly a certain James Rice who died in Ohio in 1844 at a very advanced age (said to be 120 years old!).

In the process of constructing this group, the DNA evidence was at first inconclusive, since there was one discrepancy among the first 12 loci. However, extending the tests revealed a perfect match among the 13 additional loci for the two subgroups. This overall 24/25 match is a fairly convincing demonstration of relatedness.

13040 has been traced only as far back as 1845, but matches the main haplotype 110/111. 17265, N2333, and N46489 also have not been traced all the way back but match 12/12. We therefore include all of these in this group pending further research.

Note: we are assuming that the subjects 5894, 6838, 48417, 149267, 158245, 178728, 200203, 202932, 207974, and so on represent a side branch of this group, since their lineages have not been traced, as most of the others have. Therefore, we show the discrepancies as mutations in these haplotypes. However, given the small number of testees in the group, we have to entertain the possibility that these are the "main" branch, with the rest being a "side" branch. This issue would be resolved if we knew how the two subgroups are related.

Additional testing could also help clarify the picture. Eventually, as Rices of unknown lineage are randomly added to the project, the most common haplotype in Group 5 is more and more likely to be the ancestral, "main" branch. The originally discovered haplotype in Group 5 is still the most common, but the selection of test subjects has not been predominantly random, and so that line of argument is only theoretical.

Group 6

This third Virginia group has only six representatives so far, and only two have tested beyond 25 loci, but the group seems to be well documented. Four members have been traced back to a common ancestor in Loudon County, VA. The fifth member has been traced back to a Rice in Kentucky who married in Loudon County, and his Y DNA matches 25/25 with the other members.

Group 7

This fourth southern Rice group was designated on the strength of its growing numbers. Reports of the earliest known ancestors range from Tennessee and Kentucky to North Carolina. Most members of this group match each other 23/25 to 25/25. These test subjects thus appear to constitute another group, even though there is as yet no agreement on a common ancestor.

Originally, three other testees were tentatively included in Group 7, but extending the tests to 37 markers has shown that the relationship is too distant to fit all into one group. (See Group 7A.)

Based only on the first two panels of markers (as shown in Table 1), subjects 44461, 115795, and sm04 appear to partly bridge the gap between Groups 7 and 7A, since they share a mutation at DYS454 with 7A and lack a mutation at DYS439. 44461 also shares DYS464c=16 with 7A, though 115795 and sm04 do not.

However, all of them have the normal four copies of DYS464, unlike the members of 7A. Moreover, the extended results shown in Table 2 give a different picture. Thirteen of the sixteen members of Group 7 who have tested 37 markers (including 44461) agree 12/12 with each other on the third panel and 7/7 with sm04 and v4pyw. Thus, 44461 is definitely not a bridge between the groups. sm04 is slightly less clear. On the seven extended loci in common between FTDNA and SMGF, sm04 matches the extended Group 7 members perfectly, but only 6/7 with Group 7A. However, he has no test results for the other five extended loci, where four of the discrepancies between 7 and 7A are found, and it is therefore difficult to place sm04 with complete assurance. sm04 traces his line to William Littleton Rice, born 1836 in Arkansas, and probably further to Tennessee and Virginia. However, he remains anonymous, and the details have not been verified. v4pyw has been traced back only as far as 1905. Another route to the proper placing of sm04 (and further investigation of the relationship between the groups) is to obtain test results on other members of Groups 7 and 7A for the markers on the right-hand side of Table 2 by upgrading to 111 markers. A consensus has now been established for these eleven loci for Group 7, and sm04 agrees with it exactly, while the only members of 7A tested to this level differ at three loci.

Group 7A

This group was originally thought to be part of Group 7, but further testing has shown the two groups to be separate.

5796 was the first member of this group. He differs from Group 7 by two ordinary one-step mutations among the first 21 markers, some anomalous results for the DYS464 markers (he has six copies instead of the usual four), and six or seven one- or two-step mutations in the third panel. The second member to be added was 6977, a third cousin once removed of 5796. 6977 also has six copies of DYS464, but one of these copies differs from 5796. Other than that, they match exactly.

In the original test reports, however, they differed in a startling manner: 5796 was reported with only three copies of DYS464 while 6977 was reported with five. At the time, we questioned these results because they were so extraordinarily different for people so closely related, but we were assured that the results had been double-checked (though not actually retested). More recently, we raised the question again in light of the results reported for 134428 (see below), and this time FTDNA chose to resolve the issue by an independent test. This additional test (called DYS464X) has the advantage of splitting DYS464 into two batches for most men in Group 7A's haplogroup, thus making the analysis much less subject to uncertainty. The results are tagged with either a "c" or a "g" to indicate which type, and we have shown these tags in Table 1. As it happens, though, these three men are atypical in more ways than one. Where most men in their haplogroup have three copies with "c" and one with "g", the members of 7A have all copies with "c". In any case, the final conclusion is that the two cousins do indeed differ by one step on one copy, but not by having different numbers or different types of copies.

Test subject 37139 also appears to be closely related to 5796. He shares both of the distinguishing mutations (DYS439=13 and DYS454=11) that set 5796 and 6977 apart on the first 21 markers, and he also has an unusual number of copies of DYS464. Oddly enough, despite having an extra copy of DYS464, he has significant differences on these markers. Still, the mere fact of having an extra copy of DYS464 is very suggestive, and the comparison between 37139 and 5796 shows 31/33 agreement on the remaining loci. It is possible that a retest for him would also reveal that the original test was incorrect, but his report should have been extremely reliable because it showed all five copies of DYS464 with different lengths.

Similarly, 134428 appears to belong to this group on the basis of the DNA, but also without a conventional link as yet. Indeed, the testing of 134428 has shown a new standard in the reporting of DYS464. Where 5796 was reported originally as having three copies of this marker (see above) on the basis of three appoximately equal peaks on the electropherogram trace, 134428 was reported as having six copies on the same basis, but with the additional judgment that the three peaks were more intense than expected. DYS464 has long been a troublesome marker to analyze because there are usually (though not always) four copies, and therefore often coincident electropherogram peaks due to two or more copies with the same length. Because of this discrepant reporting method, we asked for a re-examination of 5796 under the new standard, and the response was to conduct additional tests on both men, as well as on 6977 for good measure, to get an independent assessment of the number of copies for each. The independent testing confirms that 5796 exactly matches 134428 on DYS464, even to the extent of having all six copies of type "c".

Group 8

This group descends from the Rices of Weymouth, MA, who are now confirmed to be unrelated to all the other known Rice/Royce lines. Note: it was a member of this family who founded Rice University.

There are now 23/25 and 24/25 near-matches with two other Rices, 78290 and 7v3qc, who also have very near matches with each other, but who have so far found no indication of a connection with the Weymouth line. Such a match is neither close enough nor distant enough to give an unequivocal answer to the question of whether these two are paternally related to Group 8 in a meaningful time frame. Perhaps, if one or more members of the group were extended to a 37-marker test, the expanded comparison would give a clearer answer, one way or the other.

Group 9

Although testee 14746 was initially placed in Group 5 based on a 12-marker comparison, the 25-marker results showed him to be quite remote from that group. Since then, many other 12-marker results matched 14746 exactly or 11/12, and it became clear that these form a separate group. Of course, we cannot assign permanent membership without convincing 25-marker tests or documentary evidence, but we moved these results into a tentative group by themselves in the expectation that further testing would confirm the relationship. All members have now been extended to 25 markers, and the results are convincing. Two subgroups of three or four haplotypes each match among themselves exactly. Six haplotypes differ from what seems to be the ancestral pattern only at one locus in DYS464, one being an ordinary mutation of one step, four being offsets of two steps probably due to recombination, and one being apparently a three-step mutation. The latter could also be interpreted as three separate one-step mutations, but such a compound mutation is very unlikely, even more so than a single three-step. These subjects match each other far more closely at 25 markers than they do anyone else in the project, and their differences are consistent with a common ancestor within a genealogical time frame.

One participant, 36743, differs by two steps from the consensus, and both of these differences match the members of Group 5. Thus, although 36743 is closer to Group 9 than to any other group, it turns out that Group 5 is a close second. This participant is also notable for bearing the surname Reece, instead of Rice. As noted above, a surname variation at this level should not be taken too seriously, but it does stand as an additional factor in the group assignment. Thus, this haplotype's membership in the group remains a bit more tentative than the others. Finding a conventional genealogical link or a point where Reece changed to Rice, or Rice to Reece, would help to solidify the connection, but it remains possible that the link is an ancient one dating back before the adoption of surnames. Another member, 320029, is in a similar situation, in that he is almost as close to Group 5 as to Group 9, but his surname is Rice.

Another group that comes moderatly close to some members here is Group 34, which is primarily Reese/Reece. Still, even the members of Group 9 with ancestral names like Reese are definitely closer to the consensus of Group 9 than to Group 34.

Group 10

Although the first two members of this group match each other only 24/25, they have family lore saying that their great great great grandfathers were brothers.

Group 11

This group has a perfect 25/25 match between two members and genealogies pointing to a common ancestor in Tennessee, Joseph Rice. A third member, 139670, has been traced back as far as Arkansas, where Joseph Rice settled. It is therefore possible that 139670 descends through a parallel branch from a common ancestor much further back in time. In that case, the two discrepancies on markers 1-25 between 139670 and the others could be considered as an unresolved difference of opinion on the ancestral values. However, pending further research, we are assuming for the time being that 139670 is also a descendant of Joseph, implying that these two discrepancies represent mutations somewhere along the line from Joseph to 139670.

Group 12

This group, like Group 11 above, also has a perfect 25/25 match and genealogies pointing to a (different) common ancestor in Tennessee. This was the first real group to spring from the cluster of unmatched, but similar, haplotypes we designate Group 2. As such, the members have some near-matches in Group 2, and even in other real groups, as well. For example, there is a member of Group 17 who is only three steps away from Group 12 in a 25-marker comparison. They could conceivably be related, since a difference of three steps falls in what we would normally call the "gray area". However, such a relationship is not suggested by conventional research (nor helped by the fact that the other member of Group 17 is four steps away). Again, the fact that so many haplotypes lurk in the vicinity of Group 2 means that the standard of proof must be higher for designating groups among them. In any case, one member has been added to this group on the strength of his exact genetic match, despite having no obvious genealogical connection (traced to Germany by way of Wisconsin, and no link to Tennessee).

Group 13

This is yet another group with roots in Virginia. The progenitor, Samuel R. Rice, was born there in 1744, believed to be the son of another Samuel. Descendants of two of his sons have been tested, and most of the descendants have results for at least 25 loci. From these results, we can reconstruct the ancestral haplotype for those 25 loci, but, unfortunately, two loci in the 26-37 panel are uncertain.

Group 14

Like Group 12, this group has a DNA pattern very similar to that of Group 1, and the first two members of Group 14 were initially placed in the catch-all "Group 2" until a third match was found. The members' relationship to each other is based on tracing their lines back to Pennsylvania. The documentation is still lacking to connect them all to one common ancestor, but the 25/25 match among them definitely supports the connection. Even so, this group was not deemed "solid" until there were three members, because of the other (apparently unrelated) members of Group 2 who were also genetically similar.

Test subject 14295 (the first member of this group) has no fewer than three near-matches in the "gray area" of 12-marker comparisons, as well as one 11/12 match, in addition to his matches with the rest of Group 14. Needless to say, the others in the group share the same near-matches.

First, they differ from Group 1 at two loci, by one and two steps, respectively. Even factoring in the possibility that the two-step difference stems from a two-step mutation, the most likely time of the common ancestor is about 14 centuries ago, long before the Rice surname was adopted. Extending the test to 25 loci revealed one further two-step difference (at DYS448, where all members of Groups 2, 12, and 14 differ from Group 1). The most likely estimate of the common ancestor, based on a 25-marker comparison, is about 10 centuries, though a separation of only four centuries is still within the realm of possibility.

Second, they differ from 7648 in Group 17 at two loci, again by one and two steps. The 12-marker-based common-ancestor estimates are the same for this case, but the 25-marker comparison gives an estimate of about nine centuries.

Third, they differ from Group 6 by only one step at each of three loci. The 12-marker common-ancestor estimates for Group 14 and Group 6 are similar to the above, but a century or so more remote. Indeed, since no member of Group 6 has been extended to 25 loci, we would need several extensions in order to probe this possible link.

Finally, they differ from 17633 (in Group 2) by only one step on 12 markers, or three steps on 25 markers. Indeed, one member of the group has extended to 37 markers, as has 17633, and the difference there is six steps. These distances all fall into the "gray area". In any case, there is (as yet) no conventional evidence pointing to a link. Without such evidence, we have no grounds to suppose a link exists within genealogical time, even though such a link is possible.

The latest addition to the group (558761) adds a new and interesting interpretation to the whole group. This testee shows clear evidence of two things: (1) a duplication of part of the P1/P2 palindrome, and (2) a subsequent mutation of one of the duplicated STR markers. Since, without the subsequent mutation, the evidence of the duplication would not be at all clear, it seems almost certain that the other members of the group have also undergone the same duplication, but failed to suffer any addition mutation to the affected markers. Consequently, we now consider those markers to be actually duplicated, even though the duplication was not detected. Thus, the markers are considered to match, aside from the subsequent mutation to DYS464h in 558761.

Group 15

The members of this group have a haplotype that belongs to a cluster commonly found in northwestern Ireland, but there is no conventional evidence pointing beyond the immediate origins of this group in Rhode Island. Indeed, they do not yet have a documented link between them, but their 36/37 match, coupled with the fact that both 4507 and 86858 trace their lines back to Rhode Island, provides convincing evidence of a link.

There are several other members of the project who come close to this group, but not close enough to be convincing and not backed up by any conventional evidence.

Group 16

All but one of the members of this group are a very close match (36/37) and have a genealogical connection as well, albeit a connection based in part on undocumented sources. They also have a very close match (36/37 for one of them and 63/67 for two others) with a group of Fletchers from the same area of Virginia. It is not known what connection there may be between these Rices and Fletchers. Since the genetic match is not perfect, the separation between the two families could date back as much as five centuries, i.e., long before the earliest known ancestors of either group, though the odds favor a more recent connection.

Group 17

This is yet another group that is close to the pool of similar, but evidently unrelated, haplotypes in Group 2. Indeed, subject 7648 makes an interesting case study. He was one of the fairly early participants in the project, and he came close genetically to many others along the way, but it was five years before a real match came along. The long delay was especially surprising because his male-line ancestry had been traced back to early Colonial Massachusetts, and so there should (in principle) be many of his lineage to be tested. Suffice it to say that a match eventually did appear. The common ancestor of Group 17, Samuel Rice, is believed to be the son of Massachusetts immigrant John Rice of Dedham, who married there in 1649. Since John Rice had another son who raised a family, there is a good possibility that we can extend the group back one generation by testing a descendant of the other son. This would have the added benefit of producing the first real evidence of a connection between John and Samuel (beyond the mere fact of living in the same town). Let us hope it won't take five more years!

This haplotype falls just short of matching Groups 1 and 30. In the first 12 markers, 7648 differs by just two steps from Group 1 and by three from Group 30, roughly splitting the difference between the two extremes. Such a result is difficult to assess, since it falls into the "gray area". A 12-marker difference of three steps is ordinarily enough to show two subjects are unrelated on a genealogical time scale, and even two steps of difference would be very unusual within the same family. However, a large enough population will inevitably include a few outliers, and the best way to detect such outliers as such is to test more markers. We therefore encouraged 7648 to extend from 12 to 25 markers. With this broader base, we found that 7648 differs by five steps (including a two-step difference on one marker) from Group 1 and by six steps from Group 30, thus giving a much more statistically secure indication that no close relationship exists among them. Two more close-but-not-close-enough matches for 7648 also came to light. First was a near-match with 21119 and 41527 in Group 12, manifested as a one-step difference at 12 markers. This is the same level of difference as seen at 12 markers between N1892 and Group 1, and the same warning applies here. Without explicit conventional research and/or additional DNA testing, the 12-marker near-match is merely a clue, and possibly a misleading one. Extending the comparison to 25 markers revealed a total of three one-step differences. This result was still in the "gray area," and we could neither rule out a connection by the DNA comparison, nor find any evidence of a connection by conventional means. The other near-match was two-step, 10/12 with 21748. Extending this comparison to 25 markers showed a difference of 8 steps in all -- no longer of any interest.

When the real match finally appeared, it was 12/12 and 24/25. Just as importantly, though, was the fact that 111320 had also been traced back to the same immigrant Rice. This concurrence of the two types of evidence makes a convincing case that both types are correct.

Group 18

This group rests on the tests of two cousins. One of them was originally discovered in the SMGF database, but later transferred his results to FTDNA and upgraded to a 76-marker composite test.

Group 19

This group is based on a four-way match. All have tested beyond 25 markers, but not all at the same lab, and the overlap beyond 25 is not complete: two have tested 37 at FTDNA, and two 43 at Sorenson. Thus, we have 32 markers tested for all four, plus five more tested for two, and 11 more tested for two others. Of the comparable markers, only three are discrepant (one for each of three members). Two discrepancies are just one step, while the other is three steps, apparently due to recombination affecting a multi-copy marker (DYS464). Because recombination is a one-way process, we can determine with confidence which version of DYS464 is ancestral and which is the mutation in this case. The other two discrepancies are also each divided as three-against-one, and we have tentatively adopted the majority value in each case as ancestral. In principle, we should wait until we know exactly how the participants are related in order to determine the ancestral values, but a three-to-one margin is probably enough to make the call. Although all four members have links to North or South Carolina, the details of their ancestry are still sketchy.

Group 20

This group's members include three clear-cut and two tentative. Nicholas Rice, the earliest known ancestor of the three sure members, lived in Kentucky, but was probably born in Pennsylvania or possibly in Virginia or even in Germany. The names used in the family are consistent with the hypothesized German origin. Another member, N3385, has been traced to a contemporary of Nicholas who came from Wurttemberg, by the name of Caspar Rice. Clearly, this member is not a descendant of Nicholas, but it seems likely that Caspar and Nicholas were related, perhaps very closely. Originally, based on limited testing, N3385 appeared to have no matches at all, largely because of what now appears to be a single recombination event affecting the multi-copy marker DYS464. The effect of this event was to replace two of the four copies with duplicates of the other two, giving the appearance of many steps of difference between the results. With more extensive testing, we find that he differs by only one step from the group consensus on the 72 tested markers other than DYS464. There is as yet no confirmation of the tentative link, but research is ongoing. A fifth member is also tentative and was originally included because of the DNA match, but has more recently been connected to a family named Reis or Reiss coming ultimately from Aschaffenburg in Germany. Again, he is clearly not a descendant of Nicholas Rice, but Aschaffenburg is close to Wurttemberg, and there is undoubtedly a connection.

Group 21

This group has two members who are first cousins and very close on 37 markers, but the situation is complicated by the fact that the tests were done at widely separated times. One man was reported with an extra copy of DYS464, while the other was reported with a "microallele" on one of his copies of DYS464 (a feature introduced by FTDNA late in 2012). The second member is also reported with a null result for DYS425, which often reflects a deletion or recombination on the same Y palindrome that includes DYS464, even though his report for DYS464 shows only the usual four copies, all distinct.

Group 22

This group has two members whose connection is not yet known, but both trace back to County Mayo in Ireland.

Group 23

This group was discovered from a close match within Group 2, which at first was not especially encouraging. The second member of the group (53053) differs by just one step from the first member (3109) at one locus among the first 12 (just as 3109 differs by one step from Group 1). However, 53053 goes on to be an exact match with 3109 on the next 13, and now differs on only four of the total 111 loci tested. One of these differences is by 3 steps (on DYS716), but the close agreement elsewhere suggests strongly that the 3-step difference was brought about by a 3-step mutation.

There is, as yet, no known common ancestor between them, but both lines have been traced back to North Carolina, and these two therefore have a presumable link based on the close genetic match. A third member now added to the group shares a common ancestor with the first, born in 1801, and agrees with the first on both loci where the first and second members disagree. It therefore remains uncertain which is the ancestral value for all four loci where the first two members disagree.

Group 24

This group was originally expected to match a member of what became Group 14, both having connections to Kentucky. However, the DNA comparison shows them to be distinct groups.

Group 25

This group represents a trans-Atlantic collaboration between Norwegian and American descendants of a common ancestor, Rasmus Endresen Riise (1825-1924), whose son Johan emigrated to America and became known as John Rice. DNA comparison between the Norwegian and American cousins shows a 35/37 agreement, consistent with the established conventional genealogy.

Group 26

This group was formed on the basis of its growing numbers, rather than on a determination of a common ancestor. The common factor appears to be an Irish origin, but the members do not trace back to a single county in Ireland, and one member even has a discrepant surname (Reese). Nonetheless, the members are genetically close. All have tested at least 37 markers, and some have tested 67 or 111. Those at 67 or more all share a special feature, namely, a lack of marker DYS425, and only five one-step discrepancies can be seen relative to the consensus of the 30 additional markers. The situation at 111 is less clear because three of the 44 extra markers do not yet have a consensus. More research is needed to find links among the members.

Group 27

This group is based on two men without a known common ancestor, but both traced back to the same Irish county. Research is ongoing.

Group 28

This group is based on two men without a known common ancestor, but both traced back to the English county of Suffolk. The match is only 11/12, but the haplotype is rare, and the geographic link supports the genetic one. Interestingly enough, both men currently live in former British colonies, but not in North America (one in Australia and one in South Africa).

Note that Suffolk is the same county that produced Edmund Rice, the earliest known ancestor of Group 1, but there is no possibility of a patrilineal link in historical times between Groups 1 and 28, since their haplotypes indicate a separation of many thousands, if not tens of thousands, of years.

Group 29

This group is based on two Rice men who share patrilineal descent from a 19th-century chief of the Kahnawake Mohawks and have a close match on their Y DNA, which is unlike that of the Mohawk Rices in Group 1. (The latter have been traced patrilineally back to Edmund Rice through one of the Rice captives of 1704.) Instead, the Y DNA of Group 29 is typical Amerindian, not European. Thus, although their surname is European in origin and has been transmitted from father to son in the European fashion in recent centuries (beginning at least two generations before the common ancestor), it is apparent that their patrilineal ancestry is ultimately Mohawk. Still, their surname suggests they are descended from Edmund Rice in some way, and that is certainly plausible because of the traditional Mohawk system of matrilineal inheritance. Indeed, there may be many different Mohawk Rice Y lines in the present day, even though we have as yet found only two.

Group 30

This group is based on two Royce men whose lineage was formerly thought to connect to Robert Royce of Connecticut, but whose DNA differs from that of Group 3. Their lines have been traced to a common ancestor Lemuel Royce born 1757 in Connecticut, but whose 1853 gravestone claims an unusual age of 104. Census, military, and court documents support the Connecticut connection despite the gravestone, and this link to Connecticut was the basis for believing this Lemuel to be a descendant of Robert Royce, but the DNA evidence tells us that the lines are separate.

Group 31

This group is based on three Rices who trace back to a common ancestor who lived and died in Somerset County, PA. This man is both the earliest known male-line ancestor and the most recent common ancestor of the group.

Group 32

The members of this group were originally placed in Group 2 because of their distant-but-discernible similarity to Edmund Rice. Later, the close match that they have with each other was confirmed by conventional lineages connecting them all to a common ancestor in Colonial Virginia.

Group 33

At least six of the seven members of this group trace back to the same common ancestor born in 1755.

Group 34

This group includes two known descendants of a common Reece ancestor born in 1802 plus one other Reece (Rees) and three tentative members with other surnames who are reasonably close genetic matches. The two known relatives match each other 104/111, and the consensus haplotype is taken to be those 104 matching markers plus two others where all four tentative members agree with one or the other relative.

Group 99

In the "other" group, we are finding considerable diversity, with about 100 more distinct haplotypes identified already, and only 11 matches among them. These matches are shown as clusters at the beginning of Group 99 with a small gap separating each cluster from the next and from the rest of the section. Within each cluster, the columns of the discrepant markers, if any, are displayed on a white background.

One match, an identical match between 4090 and 5070, is no surprise at all because those two are brothers.

The second match is three-way, but at different levels for different pairs within the cluster. 3145 matchess 11/12 with 21802, but extends to 21/25 and thus falls into the "gray area". The latest addition, 7312l, matches 11/12 and 12/12 with the other two, or 20/25 and 24/25. Thus, 21082 and 7312l appear to be a good match, though they are only 34/37. 21802 has now extended to 67 and 7312l to 111, and they match 60/66 and share a moderately common mutation in the flanking region of DYS425 that makes it undetectable to the normal primers. This level of matching is consistent with, but does not guarantee, a connection within the genealogical time frame. An upgrade to 21802 and/or 3145 might further clarify the relationship(s). In any case, all three have Irish roots.

The third match started as a two-way exact match at 12 markers between anugt and 88027, who share the Reece variant of the surname. They would do well to compare at 25 or higher, since they may find convincing evidence of a recent link. One has already upgraded to 37, but the other has not. Meanwhile, a third and fourth match have appeared and agree 36/37 and 35/37 with 88027 and 66/67 with each other. No obvious genealogical connection has yet been found.

The fourth match is two pairs of two-way matches at 12 markers, and one of the four has tested only 12. The pair who match 12/12 and extended to 25 markers (87439 and 100521) also match each other on 23 of the 25, with three steps of difference in all (thus putting them in the "gray area" on the available tests). Indeed, other participants are nearly as close on 25 markers to one or the other of these two as the two of them are to each other. Moreover, these four are all very close to the most common DNA type in western Europe, and it is necessary to extend all of them at least to 25 markers, or preferably 37, before the match could become convincing evidence of a link. The potential for a near relationship is there, but we would need some conventional evidence to confirm that possibility. Indeed, one member of each pair has now extended to 67, and the two of them do not match very well on the additional markers.

The fifth match is three-way. Two of the men agree 25/25, and two (but not the same two) have links to Connecticut, and two (but again not the same two) have tested 37 markers. It would be helpful to extend the comparison to 37 markers all around.

The sixth match is two-way and 37/37 between 173356 and 263777. This match is brand-new, and no link is yet known.

The seventh match is another pair of close relatives.

The eighth match is yet another pair of close relatives.

The ninth match is a pair whose relationship is not known and needs investigation.

The tenth match is another pair whose relationship is not known and needs investigation.

The eleventh match is another pair of close relatives.

The twelfth match is another pair whose relationship is not known and needs investigation.

There was briefly just a hint of a possible larger grouping of subject 4283 with what subsequently became Group 7 and Group 16. At the time, only one member of each of those groups had been found, and those two, along with 4283, appeared to fall into the "gray area" between one and four steps of difference. In that case, there was no pair as close together as two steps apart, but it was still intriguingly close. Here again, we relied upon an expanded test to cast further light on the situation. The three candidates were all extended, and they showed large differences between every pair of them. Thus, the suggestion of a connection fizzled here as well. Nonetheless, we subsequently found matches for two of the three and thus established groups 7 and 16.

There was yet another hint of a possible distant connection between subjects 4798 and 6510, who differ by one step at each of three loci in the 12-marker test. However, now that both have been extended to 25, we find three more differences and conclude there is no genealogical connection. Indeed, 4798 had a second possible connection, being only two steps away from 25686 on the 12-marker test. Again, the 25-marker gave a more precise reading; the two steps of difference turned into five steps, still within the "gray area," but just barely. Without specific evidence of a relationship, such a wide gap is probably not worth pursuing.

During most of the year 2003, subject 7897 was tentatively assigned to Group 7 on the basis of a 12-marker test, even though his 12-marker haplotype differs by two steps from the consensus of that group. The reason for this assignment was the presence of two other tentative members of the group who shared one of the two apparent mutations in 7897. Now, however, the 25-marker results show that 7897 is not closely related to anyone in Group 7, and he has been moved to Group 99. For the same reason, another test subject, 59548, once tentatively placed in Group 7A based on the first 12 markers, was later moved to Group 99 after an upgrade to 25 markers. Yet another test subject, 14746, was initially assigned to Group 5 based on the same type of similarity to "atypical" members of that group, and the extension to 25 markers had the same outcome: 14746 was also moved to Group 99. However, others have now been found who match him, and these matches have now been moved, yet again, to a new group by themselves (Group 9).

17791 at first appeared to have a possible connection to Group 15. Of the first 12 markers, he matched on 11. However, extending their test results produced a match of only 21/25, thus falling into the 25-marker "gray area" (more than two, but fewer than six steps of difference). At 37 markers, the distance is 10 or 11 steps, making the connection seem very unlikely. By the same token, 36732 has a 12/12 match with Group 15, but here the 25-marker comparison is even worse: six steps (same as the 25-marker comparison between 17791 and 36732). These test subjects are all at or near the most common haplotype of northwestern Ireland, and there are therefore many apparently close haplotypes with only a distant genetic connection. Thus, it is necessary in these cases to compare at least 25, and perhaps 37, markers to get convincing evidence of a recent connection, unless some conventional genealogical evidence can be found.

A haplotype that comes moderately close to Group 7A is not fully filled in. sm27, found in the SMGF database, has only 21 of the first 25 markers. If the omission of DYS464 signifies that sm27 has five or more copies of this marker, then he may indeed belong to 7A. Until we have DYS464, we cannot place him properly, and we are leaving him in Group 99 for the time being.

Mutation Rate

This study, although too small to provide a statistically significant measurement of the mutation rate in Y-DNA STR markers, does at least provide an estimate of that poorly-known parameter. We have expanded the mutation study to include all confirmed lineages in Groups 1, 3, 5, and 6 as of 2004. Samples from Group 1 (1670-3, 4188, 5128-9), Group 3 (3156, 6061, 7242, 7628), and Group 5 (3869, 4765, 4808, 5022, 8850) represent 125 separate father-to-son transmission events with 25 loci measured; Group 1 (3111-3113, 5300, 6336) and Group 6 (4046, 4631, 5032) have 44 more with 12 loci; and S002-S009 have 40 more with 9 loci. There are thus 125x25 + 44x12 + 40x9 = 4013 mutation opportunities. The mutation seen in both 1673 and 3111 is undoubtedly one and the same mutation, since these two men share several generations of common ancestry. (However, note for comparison that the same mutation has also occurred independently at least twice in Group 1 - in 19847 and 87949. The fact that the marker in question is DYS439, which has a relatively high mutation rate, increases the chances of parallel mutations.) The mutation in 5129 is an astonishing three steps. There is no way to be sure whether this is one mutation of three steps all at once or three separate events of just one step each that coincidentally happened to fall on the same locus and in the same sense, but the latter possibility is so unlikely (five-in-a-million) as to be scarcely worth considering. In contrast, the chance of a 3-step mutation somewhere in this line is probably on the order of one in a thousand. Pending further investigation, then, we will treat this as a single event. Thus, we derive a mutation rate of 6/4013 = 0.0015, with a broad uncertainty.

Note: it may eventually be possible to include the results from other groups in this analysis of the mutation rate, but their ancestral haplotypes are still too uncertain at present.

A combination of many such studies based on deep-rooted pedigrees could yield a significantly more precise value. Preliminary results from a combination of 13 studies shows an average rate of 0.0023 mutation per generation per marker, with a statistical standard deviation of 0.0003. Ultimately, of course, each locus has its own individual mutation rate, which should be taken into account, but the average over all loci is the really important statistic. Indeed, for our purposes the only question that matters is: Does the likely separation allow for a connection in the time since surnames were adopted, or not?

111-Marker Testing

It is remarkable that the number of markers tested and compared has increased by almost an order of magnitude since this project began. Our first tests, at Sorenson Genomics, included 15 markers, but some of those proved to be duplicates or unreliable. In switching to FTDNA as the primary testing lab, we settled down to 12 markers for comparison. In those days, we had to regard any discrepancy at all between purported cousins as cause for concern about the accuracy of the conventional genealogy. Of course, mutations do happen, and they happen suddenly between father and son, so that allowances have to be made for the random nature of the process. Still, seeing more than one discrepancy would be serious cause for concern in a 12-marker comparison, albeit not necessarily ruling out a connection altogether.

With that as a background, it is startling to see the results of comparing members of Group 1 at 111 markers. The most extreme cases show 9 steps of difference between pairs of known distant cousins, and even the best case has 3 steps of difference. Clearly, it is important to keep a sense of proportion and to make wider and wider allowances for randomness as the number of markers grows.
 

DNA Data