Genetic Groups on MyHeritage: 5 What Next?

In previous posts I've put down some First Thoughts on MyHeritage's new Genetic groups, discussed their Ethnicity Estimates and Historical Maps and found them both wanting, and had a first look at how the Groups themselves shape up in my own family, and found them very promising indeed. 

Here's a few ideas for next steps.

1 Match info
First off, MyH should add the Genetic Groups to the information it gives us about all our matches. This information is already quite substantial, but the GGs have the potential to add even more value to it. Up front, in the Match List, just below the Estimated Relationships, there's plenty of room:

2 Match filter
Next, add a Filter for 'Your Genetic Groups' at the top of the Match List, so that we can see in one list all matches who share a particular GG with us. Once again, there is plenty of room. This could be very revealing, and lead to immediate progress in tracing family connections.

3 Group info on segments
Let us see the segment information on which the Genetic Groups are based. This could be via an option to display a GG label on segments in the Chromosome Browser, for instance. This would immediately help us to distinguish maternal matches from paternal ones, and could even help us to further narrow down how these matches are related to us.

4 Maternal/paternal labels on segments
Once we have identified our closest maternal and paternal matches, MyH could possibly even automate the allocation of maternal and paternal sides to individual segments, thus taking the 'bucket' procedure used by FTDNA to a whole new level.

5 Pile-up info
MyH say they have taken segments of all sizes into account when compiling the Genetic Groups. I presume this includes the pile-up regions that are the bane of Ashkenazi Jewish genetic genealogy. Could it even be that the pile-ups themselves are pointing towards the Genetic Groups? For instance, could membership of a given GG be determined by a combination of specific pile-ups on specific chromosomes? If this is what is happening, MyH has all the info needed to identify this association for us.

6 Timeline
Following on from (5), might it then be possible to suggest a timeframe for the formation of these Genetic Groups? This could contribute towards our understanding of the patterns of migration of the AJ community, in particular the movement from Western Europe towards the East during medieval times, and even the origins of the community itself.

7 Time to MRCA
In turn, could (6) even lead us towards an estimation of Time to Most Recent Common Ancestor(s), for any given pile-up region? Where and when did a particular pile-up segment originate?

8 Location, location, location
How local can we get? In her original article - all of two weeks ago! - Roberta Estes said that MyH had "absolutely nailed" her Dutch ancestry, pinning one of her Genetic Groups down to an area 20 miles square.  I don't expect quite this level of precision for AJs - but ..... ???

9 Third Parties
I would love to see what our friends at DNA Painter and Genetic Affairs can dream up. I have in the past asked Jonny Perl at DNA Painter if they could find a way for AJs to identify our individual pile-up regions. Maybe he could offer us the option of colourising our Genetic Groups in the Chromosome Browser display? And Genetic Affairs' AutoClusters do not seem to be working for AJs as well as they do for non-endogamous groups. Can Evert-Jan Blom find a way of including MyH's Genetic Groups in the cluster information he gives us? We would then be able to see to what extent there is a correlation between the clusters, which are based on shared matches, and the GGs, which are based on shared segments - ie, actual DNA.

That's probably enough to be going on with.

Genetic Groups on MyHeritage: 4 The Tales They Tell

 

The Genetic Groups assigned to myself and my brother by MyHeritageDNA
NB: Confidence levels: High - Medium - Low

Genetic Groups (First Thoughts) are the most significant new feature of the recent upgrade to MyHeritageDNA. They are doing something that I don't think is even on the radar of any of the other consumer-oriented DNA companies. They compare the autosomal DNA of thousands of people at the segment-level, and where they find sufficient similarities, they assign us to specific Genetic Groups. I think this approach has huge potential for those of us who are looking to uncover our genetic background, at both individual and community levels, and also for those seeking to discover new - or long-lost - family connections.

Not ours, but theirs
But first, let's dispose of the Ethnicity Estimates and Historical Maps presented to us by MyH as part of this "upgrade".

In previous posts (Who I Am and Who I Am Not, and Mapping the Past, or Not) I have been highly critical of both features. They are based on faulty logic, and are consequently highly misleading. Both purport to tell us about our ancestry, but in fact both are based not on our own genetic history but on that of the people we have DNA matches with. The "ethnicities" and maps are not ours, but theirs.

This pseudo-science culminates in the absurdity of a map of Europe in the period 1600-1650, which appears to portray Britain as being home to a substantial proportion of the continent's Ashkenazi Jews. At this time, of course, there weren't any Jews in Britain at all. Jews - all Jews - had been excluded from the country for over 300 years.

Both features should be consigned to the bin marked "In need of urgent attention". At the very least they both need to display a Genetic Genealogy Health Warning, prominently and immediately. Maybe something along the lines of: "These Ethnicities and Maps can seriously damage your understanding of your genetic origins. Remember: they are not yours, they are those of your matches."

Genetic Groups
The Genetic Groups, on the other hand, appear to be built on a sound scientific basis. I shall look at what I think these Groups can bring to our family history research, and suggest ways in which the concept could be developed.

The chart at the top shows the Groups that have been assigned to myself and my brother. Both our parents were from Ashkenazi Jewish families, our mother's side from Poland and our father's from Belarus. I'm hoping that we can make some sort of sense out of these Groups. Are some of them paternal side, and some maternal? Is there a geographical logic to them? Or are they all one indistinguishable mess, as most of AJ genetic genealogy seems to be?

Note that MyH gives a us 'Confidence level' for each of these Groups, of High, Medium or Low. In the charts I have shown these in bold, normal, and italic script respectively.

I have been classified in 5 Genetic Groups, my brother Brian in 6 - the same 5 as I have, with one extra. For GG5215 and 5052, we are both High Confidence; I am High for GG5043 and Medium for 5073, whilst Brian is the other way round. We are both Medium for GG5032. Meanwhile Brian has a Medium for GG5227, which I don't seem to have at all.

So far, not so far. Brian and I are very similar, but not identical.

The cousins show up
However, I think there are indeed some interesting things to discover when we look at how some of our cousins show up.

Genetic Groups for myself, Brian and some of our cousins

NB: blue - paternal side, pink - maternal side

Brian and I have 14 known cousins on MyH-DNA, 7 on our paternal side (blue in the chart, from various places in Belarus), and 7 maternal (pink, from Central Poland 100km West of Warsaw, on or near the Vistula River). In the blue corner they range from Katy, a 1C1R (first cousin once-removed) to Gerald, a 3C1R. Katy shares all our ancestral families on the paternal side, the 2Cs represented here all share Ilyutovich and Levin, but not Schreibman, and the 3Cs share only Levin.

In the pink corner the cousins we have on MyH at the moment are a little more distant. The 4Cs share only Frankenstein, whilst the 3C-4C*2s are all closely related to each other - Arlene and Sandra are sisters, Daniel K and Laurie are also siblings, children of Sandra. This line is related to us twice over, via two brother/sister Frankenstein - sister/brother Zegelman marriages over 150 years ago.

As far as we know, there is no connection between our paternal and maternal sides, until our parents got together, of course. However, FTDNA does insist on telling us that a number of our matches there match us on both sides. We presume this is down to general AJ endogamy, especially given the geographical separation between the two sides over at least the past 200 years.

Intrigue
The resulting charts are intriguing; please note I am not necessarily looking at the Groups in order from left to right:

GG5215: nobody in the pink corner (maternal side) is in this Group, and there are only 2 blues; only one of these blues - Katy - has a High Confidence rating. She is our only Schreibman match on MyH at the moment. We can therefore confidently classify GG5215 as our Schreibman Genetic Group.

GG5073: has no pinks, but all the blues are there. GG5073 is therefore paternal, and associated with Belarus. Furthermore, the only High-rated cousins are the Ilyutoviches, who were from Lida in NW Belarus. Katy, who shares Ilyutovich, is Medium-rated, as are the 2 Levin cousins, who do not. Brian is High, I am Medium. This is our strongest Ilyutovich Group.

GG5052: is present across both blues and pinks - but all the pinks are High-rated; amongst the blues, only the Levins are High, whilst some of the Ilyutoviches are Low. This is clearly our main Frankenstein/Zegelman Group, and is based in West Central Poland. It could be that our Levins have a connection with this area that we do not (yet) know about. Or it could just be more general AJ endogamy creeping in.

GG5043: all cousins bar Katy share this Group. Only Gerald of the blues is rated High; on the pink side, only the double Frankenstein/Zegelman cousins are High. I am High, Brian is Medium. I don't think we yet in a position to definitively allocate this Group to paternal or maternal, to Belarus or to Poland.

GG5032: Brian and I are both Medium. No-one in either blues or pinks is rated High. All the pinks are in this group, but only half are Medium, the others are Low. On the blue side it is a bit more nuanced. All the Ilyutoviches have it, whereas of the others, only one Levin does, and he is Low. There are 2 Highs, Daniel and his nephew Jeffrey, one of whose lines is Rothstein from Miedrzyziec Podlaski in Eastern Poland, not far from the border with Belarus. So, more endogamy, but leaning towards E Poland.

GG5227: Brian is Medium, whilst I don't have this Group at all. The only others who have this Group are the two Rothsteins, and three of the Frankenstein/Zegelmans, but they are all weak. I'm guessing Central Poland for this one, maybe tending towards the East.

Strangeness and Litvaks
GG5031: this one is strange, as whilst neither Brian nor I have this group at all, most of our cousins on both sides do. Of the blues, Katy is High, and all the Ilyutoviches are Medium, as is Beatrice of the Levins. Of the pinks, all the Frankensteins and Zegelmans, bar Angela, appear, but at a lower rating than the blues. Katy's High rating is intriguing; this Group is clearly not a Schreibman Group - Brian and I would be there if it were. Katy's father's family is from Lithuania, and we have no known connection there, although Lida, our shared Ilyutovich area, is close to the Lithuanian border, and often came under Lithuanian influence. Maybe some of the other cousins have Litvak connections; we know little of their Trees outside how they connect to us.

As you would expect, there are further Groups like this last one that have some cousinly representation, but that Brian and I are absent from. However, there are not many of them, and our cousins' presence in them is sparser than in the Groups that we do share with them.

OK. That's a first analysis of our Genetic Groups. Fascinating, although at the moment I don't see much more that we can do with them. 

What next?

Genetic Groups on MyHeritage: 3 Mapping the Past, or Not

 

MyHeritage AJ Genetic Group 5215, distribution 1850-1900

Ashkenazi Genetic Groups
MyHeritage has discerned 24 distinct genetic groups for Ashkenazi Jews, based on their analysis of the autosomal DNA of the people who have tested with the company. I must say before we even start, that I don't think this has ever been attempted before. It is a huge advance on the vague, confusing and inaccurate "Ethinicity Estimate" approach that they and other companies have been using up till now.

I must also say that unfortunately, the most visually appealing element of the initial roll-out of Genetic Groups, the historical distribution maps, are presented in a highly misleading way. This post attempts to explain how this happens.

My 5 GGs
MyH has allocated me to 5 of the 24 AJ groups, which in itself is of interest. It suggests that I do not have much, if any, genetic affinity to the members of any of the other 19 groups. In other words, AJs are not all related to each other. It suggests that our endogamy is within our own genetic groups, and not across them. Which makes sense, but also raises an intriguing issue: many of these AJ Genetic Groups appear to cover quite wide geographical areas (see the map in the post First Thoughts), and many of these Groups appear to be in the same areas at the same time. How did they manage to keep the groups genetically distnct? Did they not inter-marry?

Also of interest, to me at least, is the fact that I do not appear to connect to any of the Sephardic Groups. That knocks that family myth on the head.

From Trees to Maps
One of my AJ groups is numbered 5215. About half the people in Genetic Group 5215 have posted family Trees on MyH. The company has used the location information included in these Trees to map the geographical distribution of the ancestors of group members over time. 

The map above purports to show where the ancestors of the members of Genetic Group 5215 were living in the period 1850-1900. There are other maps for each of the 50-year periods between the years 1600 and 2000, and it is highly instructive to see how the distribution of population shown on these maps evolves over the years.

A bit of a puzzle
The families of all 4 of my grandparents are documented in Eastern Europe in the period shown on this map - my mother's side in Central Poland, and my father's in Belarus. The map above shows a just a light smattering in Belarus, some in Central Poland, some in the Austria-Hungary area, and stronger concentrations in the Netherlands and in Britain. Were my 5215s really more concentrated in England than in Belarus or Poland in the late 19C? This is a bit puzzling, because we know that the mass immigration of Jews from Eastern Europe to the UK only began after 1880, and the bulk of it was after 1900. So why does the map suggest otherwise?

A little look at Belarus
Before we delve further into this, let's have a closer look at Belarus.

When we zoom in on Belarus in the 1850-1900 map, we can see that the 5215s are indeed represented in my three paternal-side areas - my Schreibmans and Zaturenskys are from Pinsk (bottom left), my Levins from Gomel (Homyel, bottom right), and my Ilyutoviches from Lida (upper left). But there are some wide open spaces in between. Some of my other Genetic Groups have even less presence in Belarus, so this might still be an indicator that this is indeed where my families are from. Unless, of course the pinpointing of these three areas has arisen at least in part from the information I have put into my own MyH Tree - in which case, we're going round in circles.

Pause for thought
At this point, we should pause, and remember where the information that has gone into these maps comes from. MyH has found that amongst the people who have tested with them, there are 1330 whose autosomal DNA is sufficiently aligned to justify the creation of a Genetic Group specifically for them, GG5215. Just under half of these - 623 - have Trees on MyH. The company has combed those Trees for whatever ancestral date and place information it can find, in particular births, marriages and deaths. It then maps this material in 50-year periods. So far, so good.

However, what they have not done, and cannot do, is DNA-test all the ancestors that appear on those Trees. They may not all be GG5215s. In fact, we know that many of them are not; for a start, they have allocated me to 4 other AJ Groups. The descendants who have tested may show sufficient genetic characteristics to warrant being allocated to this Group, but many or most of them will also have ancestors from other AJ Groups, or who are not AJ at all.

For instance
A couple of examples from my own case will illustrate this point. As I mentioned before, half my ancestors are AJs from Poland, half AJs from Belarus. I have both sides documented in those places back to +/- 1800. So I would be happy to accept that GG5215 covers both areas, and both sides of my family. But none of my genetic ancestors were in western Europe or the UK in that period, and I very much doubt that any of the AJ ancestors of other members of the group were there either, for the reasons outlined above. So what is the genetic connection between GG5215 members and these places? It can only be that some of them have ancestors who were not AJ. So I'm expecting to see evidence of non-AJ ancestors in the Trees of at least some of my fellow GG5215 group members, and specifically of ancestors from Western Europe and Britain.

Close to home
And I don't have far to look. Consider my 1C1R (first cousin once-removed), Katy. We share my paternal line - her grandfather was my Uncle Mick, my father's brother. Mick married Margaret, a non-Jewish English woman, so although Katy got enough of the characteristic GG5215 DNA from Mick to qualify for the Group, she also has 25% English ancestry through her grandmother Margaret. MyH have picked her English ancestors up from her Tree, and put them on the 5215 map. Where they don't belong.

And this is all exacerbated by the fact that people with English ancestry can generally take their lines much further back that AJs can, with the exception maybe of a handful of rabbinical lines, and this applies to places as well as names. So when you look at the earlier maps for GG5215, you find that the places where records are available show up much more strongly than those where they're not. And the further back you go, the greater the disparity, and the more skewed the results will be. 

Reductio ad Absurdum
And you end up with this absurdity:

MyHeritage AJ Genetic Group 5215, distribution 1600-1650

But there weren't any Jews in England in 1600-1650.

They had been expelled in 1290 and were not re-admitted until 1656. (See From Expulsion to Readmission, by Ariel Hessayon, for the background to both of these episodes). And even then they were mostly Sephardim, originating from Spain and Portugal, not AJs - including no doubt my GG5215 ancestors - who to the best of our historical knowledge, originated in Germany, moved towards the east in the late Middle Ages, and were by 1600 well established in Poland and probably settling in Belarus as well.

So whilst these maps do represent my GG5215 ancestors, to the best of our limited collective knowledge, they also represent the ancestors of all members of 5215 who have Trees on MyH. And this includes those of their ancestors who were not 5215ers, or not Jews at all, and could have come from anywhere, especially, it seems, from Southern England and the Netherlands.

So it's not my ancestors on these maps, it's theirs.

Genetic Groups on MyHeritage: 2 Who I Am and Who I Am Not

 

Who I am not

Well, that looks interesting. Under the heading 'Your ethnicity results', MyHeritage appears to be informing me that in addition to being 91% Ashkenazi Jewish, I am 2.9% Celtic Fringe, 2.8% Middle Eastern, 1.9% Mizrahi Jewish, and 1.1% West Asian. I must say that this comes as a bit of a surprise, as we have always considered ourselves to be 100% AJ, my mother's side coming from Poland and my father's from Belarus.

At first glance the 2.9% figure seems to be suggesting that one of my great-great-great-grandparents might have been from one of the Celtic areas of Britain. However, I have researched all 4 of my grandparents' lines back to this level - to 1800 or earlier - and as far as I can tell there's no sign of an Irishman anywhere. There's Leweks and Jankels, Movshas and Chaims, but absolutely no Paddys. There's no indication that anyone from Ireland, Scotland or Wales migrated to Poland or Belarus, married a local Jew, and forwarded their DNA down through the generations. And the same goes for the other "ethnicities" that I seem to have been allocated.

Something's not quite right.

Who they are

The list is wrongly labelled. Look at the subheading under 'Ethnicities' at the top. This is not the distribution of my own ethnicity. It's the distribution of ethnicities across all of the people who match me. It's not who I am, it's who they are.

Which could be of interest in its own right, of course. How come I share DNA with over 2000 people whose DNA has a strong element of Irish, Scottish and Welsh?

How it comes about

Let's do a mind exercise.

This is what my Ancestor Chart, back to my 16 great-great-grandparents, looks like, with me at the bottom. Yours will be similar.

Now here's the Descendant Tree for one of these couples, Shmuilo Gronim Ilyutovich b 1825 and his wife Tauba Belagratsky (they are at the top here, and this chart also shows their parents). They're from Lida in NW Belarus. I've taken it down as far as my own generation.

The chart is illegible because there are 83 of us in my generation. At least those are the ones I have been able to trace, there may be a few more yet to be discovered. There are also a number of people down the generations who I have been unable to follow through, and there will certainly be some who have slipped through the net entirely, especially in 19C Belarus where the records can be patchy to say the least. But I do have names and places for the intervening generations of all the 3Cs shown here. None of these cousins, nor any of their Ilyutovich ancestors, were in Ireland, by the way. However, one or two of them may well have married people with Irish ancestry, once they reached countries where there was Irish immigration as well as Jewish, such as the UK or the USA.

Now consider that this is the Tree for only one of my g-g-g'parent couples. There are seven more couples at this level, each of which may well build up through the generations as this one does. Even if they "only" come down to half the size of this family, say to an average of around 40 in each family in my generation, that is potentially another 300 or so 3Cs.

I have developed some of these families to a similar level to this one, and the same arguments apply. However there are a few families I know little about. Those 40 potential 3Cs may well exist in these families, I just don't know anything about them. I have no idea who their ancestors married, or who they had children with. Did they move elsewhere in Poland or Belarus? Did they emigrate? Were they killed in the Holocaust? Might one of them have married an Irishman or woman?

What about his brothers and sisters?

But there's more. Take Shmuilo Gronim, for instance. He was one of seven children. I have information on just two of his brothers, and that dries up around 1900. I have no idea what happened to them or any of their families. Some of them may have had descendant charts as fully populated as that of Shmuilo. Their descendants would be 4Cs to me, and there could be another 400 potential cousins out there. I have no idea if they even exist, let alone whether any of them might carry any Irish DNA.

And of course, we can repeat this exercise for the siblings of each of my 15 other g-g-g'parents. There could be another 400 unknown 4Cs in each of those families. That's 6000 more cousins. As if I didn't have enough already.

Celtic Connections

And there's a good chance that somewhere along the line, an ancestor or two of at least a few of these unknown thousands might have married a Celt. And when a descendant of theirs in the 2010s or 2020s takes a DNA test with MyHeritage, their Celtic DNA will show up in their results, along with whatever AJ DNA has come down to them.

It's not me that has Celtic DNA, or Middle Eastern, or Mizrachi, or West Asian - it's my matches.

What about my brother?

My brother also has his DNA on MyH. Here's his Ethnicity results:

Oh? Wot, no Irish?? My brother has a similar number of matches to me with Celtic ethnicity, over 2000 of them. However, he doesn't seem to share any of their DNA. He also has more West Asian, plus a trace of Central Asian that I don't have, but none of my Mizrachi.

Of course, although my brother and I received our DNA from the same two parents, we only actually coincide on 36% of it. The other 64% of my DNA consists of segments that he doesn't have, and vice-versa. This explains why our match lists can seem so wildly different, outside of known cousins. Even with known cousins we share significantly different amounts with many of them, and often share on different segments.

So it shouldn't be surprising that we see a difference in the ethnicities of some of our respective matches. I have segments, some of which my brother doesn't have, that I share with matches who have Celtic, or Mizrachi, origins. Conversely, he has segments that I don't have, that he shares with matches who have Central Asian origins.

Again, it's not us, it's them.

Genetic Groups on MyHeritage - 1 First Thoughts

 

MyHeritageDNA have introduced 'Genetic Groups', a new feature which I think could have a huge impact on our understanding of Ashkenazi Jewish (AJ) history in general, and of our own families' histories in particular. The same will undoubtedly apply to other groups across the world.

MyH has published an account of how they have put these Groups together. Briefly, they base them on the segments of DNA, including very small ones, that you share with other people. These patterns of shared segments can be regarded as the signature of a Genetic Group. These shared segments must have come from a common ancestor, who may well date from long before any of the ancestors that most of us can trace by traditional genealogical methods. MyH then combines this shared segment information with the historical and geographical information contained in those people's Family Trees.

This technique will not point directly to a specific common ancestor, but the fact that you share a Genetic Group with someone should help narrow down the field, and may suggest likely areas for further research. It should eventually provide a more reliable, and potentially much more informative, indicator of ancestral origins than merely looking at shared matches, or vague "ethnicity estimates". 

My Genetic Groups

The map above shows the geographic reach of the 5 AJ Genetic Groups that they have found for me, based on shared segments and the date and place information my matches have included in their Trees. Of course, not everyone has a full Tree, but when the numbers of people providing this information are high enough, MyH can begin to see patterns emerging. These patterns show that people who match me on particular segments tend to have ancestors that come from particular geographic areas.

You can see immediately from the map that these Genetic Group areas cover distinct territories. There are two Groups in smaller regions that have a definite western bias, running from the Netherlands across to Poland; one of these reaches a bit further north than the other. Another Group has an eastern bias, reaching from Poland to central Russia. One of the Groups seems to cover much of the map, stretching most of the way across from Germany to Ukraine. The fifth Group is centred in the East, across Belarus and Lithuania. My matches in each of these Genetic Groups share a DNA signature with me that corresponds with these areas of origin. You can almost see the Litvaks and the Polacks peeking through.

My mother's parents both came from central Poland, and my father's from various places in Belarus. Jews came into these areas from Western Europe during the Late Middle Ages, moving gradually eastward over a period of several centuries. You can sense the pattern of this migration even in the roughly-drawn areas on this map. I have documented Trees back to 1800 or so for the male lines of the families of each of my four grandparents, but I do not as yet know where any of these families were living before they moved east, nor when they moved. Any source that can suggest answers to this is worth looking at.

My initial thoughts are that a common ancestor for me, with someone who is in one of the more western groups, will probably date from medieval times, and will be correspondingly hard to trace. Matches from the two groups that cover the eastern areas may well trace back to more recent times. My Polish matches could come from any of the five groups, as the map shows that all five of them include the central part of Poland.

I'll be looking more closely at my five Genetic Groups, from the geographical, historical and genetic perspectives. What can they tell me about the stories of my ancestral families?

NB: Roberta Estes has posted an excellent guide to using this feature, on her DNAeXplained site: Introducing Genetic Groups at MyHeritage

Moshe Chaim, Czar of Pinsk: #10 There's more to Dora than meets the eye

No docs for Dora
Key moments of Dora Zaturensky's life appear to be almost totally undocumented, so what follows is an attempt to piece together fragments of disconnected information into what I hope is at least a half-way coherent story.

Some things are certain:
1) I have a strong DNA match with one of her great-grandchildren, Cousin Paul, and a weaker one with another, Cousin David.
2) Her father has the same name as my great-great-grandfather, Movsha Zaturensky, he's from the same town, Pinsk, and was having children at about the same time. I deduce from this and the DNA connection that her father *is* my gg-g'f, that Dora is the sister of my g-g'm Shprintsa, and that Paul and David are 3C to me. There are virtually no documents openly available from Pinsk from this period, and none that mention any members of this family.
3) Dora left Pinsk around 1887 and emigrated to the USA. She came to live in Peoria, Illinois, and married Joseph Kawin either just before or just after emigrating.
4) She had two sons with Joseph, in Peoria: Abraham b 1888 and Samuel b 1890.
5) Joseph died around 1897.

There appear to be no documents available for any of the events described in (3), (4) or (5) - no Passenger Manifests that I can identify as being for either Dora or Joseph, no Naturalisation documents, the 1890 US Census has been destroyed, there's no marriage record, no birth records for the sons, and no identifiable death record for Joseph. All the information presented here is either culled or inferred from later documents.

Dora and the Kawins
The first time I can find Dora on paper is in the Peoria Directory for 1898, as the widow of Joseph, living at 108 Gallatin Street:

As you can see, there is one other Kawin family in town, running a china store. They must be related to Joseph, but I have found no documentation to verify this. They are in Peoria in the 1880 Census, where the head of the family is Max.

Who is Joseph?
A second look at the 1880 Census reveals that one of Max Kawin's sons is a Jacob b 1862.

Now Joseph appears in Peoria Directory listings in 1890 and 1896, and Jacob does not, although several other members of the Kawin family do (see below). Jacob dies in 1897, and as we have seen, Joseph has also died by then - Dora is the "widow of Joseph" in the 1898 Directory.

Are you thinking what I'm thinking?

Joseph is Jacob? Jacob is Joseph?

Or is he??


Not forgetting George
Thinking on this question, and browsing through the Directory listings for Peoria on MyHeritage, I stumbled across a feature I had not noticed there before. You can do a "who else lived at this address" search. It returns matches for the address across all years, but could be very useful for answering questions for the years between censuses, when you have to be a very lucky researcher indeed to locate that information.

So I checked Joseph's entry for 1890:

And then I did the "who else ... " search on his address, 303 Gallatin. I found this in 1891:

George Kawin? Who is he? He's not in Max Kawin's family. He's a peddler, like Joseph the year before. Come to that, where's Joseph? Hmm.

So I followed up on George. There is only one other entry for him on the whole of the internet (trust me), this one for 1892:

He's moved down the street to number 414. And still no Joseph. Hmm again.

Let's try the "who else .. " search again on MyH, and see who turns up in 414:

Hello, hello! It's our old friend Simon Moses! Or Shmuel Zaturensky, as we prefer to call him. And this entry is also for 1892 - so this time we've got two people in the same house at the same time. And they are two people that we know quite well: Shmuel is of course the brother of Dora, and Dora, as we know, is married to Joseph/Jacob Kawin.

Are you thinking what I'm thinking (again)? Is George just another name for Joseph? And are George, Joseph and Jacob all the same person?

Why Peoria?
Once I had reached that, albeit tentative, conclusion, I decided to follow this Kawin family through. I found that Jacob had arrived in the USA aged 8 with his mother Hinda and 4 siblings in 1871. They came from Volkovisk, in what is now western Belarus, about 100 miles NW of Pinsk, and settled in Peoria. Dora was born around that date, in Pinsk, and seems to have emigrated, on her own, around 1887, aged about 16, although as yet I have not found a Passenger Manifest for her. She must have married Jacob/Joseph/George soon after arriving, as Abraham appears to have been born around 1888-89, in Peoria. Bear with me, these dates are crucial.

All this raises more questions. Why did she come to Peoria, specifically? The most likely explanation is that her brother Shmuel was already there - he had obtained US Naturalisation as Simon Moses in 1886, in Peoria, so must have been in the USA for a number of years prior to that. He would have known the Kawin family, who were well established in the town by then, and was a similar age to Jacob/Joseph/George. Did Shmuel suggest she come over to marry his friend? As we have just seen, Shmuel was living with Dora and Joseph/Jacob/George a few years later.

Dora's daughter
In the 1900 Census, 3 years after the death of Jacob/Joseph/George, Dora and her two sons Abraham and Samuel are in Peoria, at 108 Gallatin Street, the same house as in the 1898 Directory.

I couldn't find Dora at first in the 1910 Census (more on this later), but from 1913 onwards they all start showing up in Los Angeles. In several Directories there, she is listed again as "widow of Joseph".

And then, in the 1920 Census, a strange thing happens:

Suddenly, Dora has a daughter, Sarah, aged 18, born in Iowa. 

Oh dear. This causes us all sorts of problems.

If Sarah is 18 in 1920, she would have been born in 1901 or 1902. But Joseph/Jacob/George had died by 1898. If Sarah is Joseph's daughter, she must have been born by 1899 at the latest - so how come she doesn't appear in the 1900 Census?

We seem to have two options. Either:

a) Sarah is Joseph's daughter, was born in 1898, was left off the 1900 Census by accident, and has been falsifying her date of birth ever since.

or

b) Sarah was indeed born in 1901/2, and so is not Joseph's daughter. In which case, whose daughter is she? And why can't I find them in the 1910 Census?

At this point, I am afraid we shall have to hold this part of the investigation for a moment or two, as other matters are about to arise that will complicate Dora's story still further.

Something in the air

There must be something in the air ...
There I was, whiling away the time on the top deck of the Number 7 bus, scrolling through the emails on my phone. There was one from MyHeritage, so I opened it to see what news there was on the family history front. It was a list of 'record matches' they'd found for people in the Family Tree I've got on their site. One of them was for the family of Joseph Szapira of Brighton. In the 1891 Census they were living at 3 St James's Street.As I read this, the bus pulled up at a stop, and I glanced out of the window., as you do.
There it was, just behind me - Number 3, the yellow building currently offering organic Belgian chips. In 1891 it was a fruiterer's. Joseph Szapira, formerly a jeweller, seems to have taken over the business after the death of his father-in-law Samuel Joel. Apart from the shop-fronts, the row of buildings, with their bow-windows on the first floor, are more or less as they were in 1891.So was this email from MyHeritage primed to open at just this time, and just this place? Are they in cahoots with the bus driver?
Or is it maybe Kate Joel's way of reminding me that I haven't yet managed to link her family with that of Jacob Joel, who lived at 89 St James's Street? Or is it even Isaac Frankenstein nudging me again, from across the Atlantic - I haven't managed to link him to anyone at all.
Or .... ?