New error idea: LNAB different than parents LNAB

+12 votes

I am looking into new error LNAB not same as any parents last name (Father's or Mother's LNAB or CLM).

I did check it and we would have 400K errors.

When I looked it in details, There are a lot of *dotter and *son lastnames and parents *son and *dotter. Example Nilsdotter-503

If I skip *dotter and *son we have 350K errors.

When we get new database dump from Chris, I can check it also against other last names. Until then it will be more false errors.

in The Tree House by Aleš Trtnik G2G6 Pilot (479k points)
retagged by Maggie N.
Wouldn't a lot of profiles be on the list due to spelling variation in the actual records rather than our spelling mistakes? Or is that what your looking for?

E.g. my 2nd great grandfather (and great grandmother) spelled their surname Rushfirth which is the spelling on birth certificates/christening records and later records, but earlier ancestors used Rushforth. Which would come up as an error report
You could maybe remove the Welsh ones that start with ab or ap or ferch or verch or Fychan or Vychan.

Also maybe anything starting with mac followed by a space.

You won't get a huge reduction in numbers though.

Can't do much else with Mac and Fitz names because there's no clear difference between the ones that are hereditary and the ones that aren't.
I think *dotter and *son would take care of most of the Norwegian/Swedish/Danish patronyms, but all the Dutch patronyms (*sz, *s, *se, *sen, *sdr, *ze,  etc.) would still show up.

I wouldn't do this unless you can filter out patronymics.
Thank you Joke :)
This is not going to work for Arabic naming conventions (where a child's "Last Name" is the paternal grandfather's "First Name").  The child's LNAB is ALMOST ALWAYS different from the parent's LNABs

15 Answers

+19 votes
Best answer
This gets way too complicated, so i will not add this error for now.

Maybe we can make just a few variations for scandinavian pre 1800,...
by Aleš Trtnik G2G6 Pilot (479k points)
selected by Dale Byers
Or even better idea, to create test errors, rhat would show only on demend, until error is acceptable for general public. That would be usefull for all new errors.
that brings up an interesting idea... how difficult would be be to allow someone to define and search for their own error on demand?
+11 votes
Please make sure to exclude ather variations of patronymic names, -sen, -datter, -søn, -dottir.

And my knowledge is limited to the nordic countries, other nationalities will have other variations
by Torben Friberg Sørensen G2G6 Mach 1 (12k points)
+7 votes
If most of our entrances depend on censuses for the spelling to add children (like I have done) you are going to run into a lot of false errors due to the listings.  Although not as big as the patronymic names.  I did one large family of French immigrants where the father changed his name three times and another man that had three different spellings of the LNAB for his children.
by Beulah Cramer G2G6 Pilot (237k points)
I think this will be a lot like the current errors for unique first name spelling, yes, there will be a lot of false errors, but, we will probably find some significant "other" issues that are worth looking at.   I say, with only 350 errors, it is worth going through them.
But if father changed his lastname 3x 1 should be in current last name and 2 in other last names.

Robin 350K is 350000,
Nobody ever had 3 different spellings for the LNABs of his children.  This is random noise, not genealogy.
Usually spelling of the name changed in case of migration or government changes. But in such case also parent's CLN should change and then child is born to parent with same spelling of the name.
OOPs, 350,, let's not do that!
Yes, fathers did have 3 different spellings for the last name. Sometimes 4 or more. The vicars recording the baptisms, marriages and burials spelled it as they thought fit. Father and/or mother signed with an X. Later generations adopted one of the many variations.

e.g. Firns, used to be spelled Ferns, Fearns, Fern, Fearn, Firn and all in one generation of a family.
I give you the baptismal records of my 7xggf's children, which include the spellings Kronhise, Kroneiss, and Kroneisen. In a couple generations the spelling had stabilized as Cronise, but for a while the LNAB varied greatly.
Both of these were from gedcoms so I was not involved with the original profile naming, but Mr. Pitre changed his name from Pitre to Peete after migrating from Canada to the United States and his family and siblings had about six variations of that.  The Corssen, and Corsens were in the same family and children also used Corrson in the next generation.  I would have had a hard time determining the original spelling when it varied in the censuses. Especially when there might be a missing census between the ones available and new children's names had been added with different spellings between the censuses.
Like I said, this is random noise, not genealogy.

One of my names goes back locally to 1200.  It's been spelt umpteen different ways, but it's always been the same name.

There's no value in tracking the spellings.  They don't signify anything.

Written language isn't the primary form of language.  Spelling isn't intrinsic.
Incidentally, I never saw a census where the surname spelling was different for different siblings in the same household.

Strange how millions of families were enumerated over decades and nobody ever respected the fact that their kids had different surnames according to official government records.
Parish registers are a different kettle of fish. Census only goes to 1841 or 1901 for the countries I work with. Registration of a birth has whatever the registrar wrote down when the informant used X as a signature.
I recently saw different surname spellings in the census record for a family whose entries were split between two pages. (That won't happen when the whole family is on the same page, since the surname is entered only for the first person, and the rest are "dittos.")
+16 votes
Not a good idea, in my opinion. As others have noted, there are myriad good reasons for a child's LNAB to be different from the parent's name. And since most genealogical software packages have a built-in default assumption that children will have the same last name as their fathers, I believe we have more trouble with people forcing LNABs to match when they shouldn't match than we have with people accidentally changing the spelling of a family name. (And WikiTree already has a default assumption about matching of last names. When a new profile is created as an immediate relative of someone who could be presumed to share the same last name, WikiTree already suggests an LNAB based on that match -- and the user has to override that suggestion if the suggested name is wrong)
by Ellen Smith G2G Astronaut (1.1m points)
+7 votes
I have a step son as well as his son who would generate that error  that I know of without even checking and they would both be false errors due to the way Ohio law requires birth records to be handled under certain situations, for example parents unmarried. I also just thought about one of my grand daughters that would show up.
by Dale Byers G2G Astronaut (1.3m points)
Can you be more specific? What is the child lastname with unmarried parents? In Slovenia parents decide for child to have one, other or both last names.
I do not feel like getting involved in this debate. I noted that I would have a few false errors that I know about without spending time looking things up and that was my point. I do believe that Ellen has the best answer, the way the software is set up we have to make a deliberate effort to have the LNAB not match one of the parents names and for that reason I feel that most, if not all, of the "errors" found this way would be false.
+11 votes
I think not a good idea. Especially because the Dutch patronymics that occur frequently in Pre 1800 profiles.

For example a child of Jan Adriaansen can get lastnames like Jansz, Jans, Jansen, Janse independent of the clerk who wrote the note or local customs.
by Joop van Belzen G2G6 Mach 9 (95.5k points)
edited by Joop van Belzen
So If Child LastName is like Fathers name + S + anything, that is not an error if he was born in holland before 1800.

If we define it like that, we can add another error if child is from Holland before 1800 his LNAB must be Fathers name + S + anything.

Same goes for Scandinavian and other patronyms.
Its not always a S,  if fathers firstname ends with a S then sometimes only the Z is added, for example firstname  Claas, the Lastname for the child good be Claasz.

Sometimes the X is used, child of Heyndrick : Tanneke Heyndrickx

An other problem is that sometimes during lifetime (in my family between 1600 and 1700) a kind of extra  lastname /nickname was added.

For example de National naval hero Michiel Adriaansz de Ruyter. Fathers name Adriaen Michielszoon, LNAB Michiel Adriaensz .  In liftime the name "de Ruyter" is been added. His children were given the surname"de Ruyter"

Also Lastnames can change. In my case of fathers line: Meeuwsz, Danielse, Danielse van Bilsen, van Bilsen, van Belsen and at least van Belzen. This in a period of 400 years.

So I think its not a good idea because its to complicated.
+14 votes
I don't like this idea because it would have to exclude so many Russian gender-oriented names, Scandinavian Patronymics  and Spanish / Portuguese constructions.

Second it would give rise to many and various solutions depending on how the inconsistency arose.  Indeed I suspect it would generate more errors than it solved.
by Sir William Arbuthnot of Kittybrewster G2G6 Pilot (166k points)

Can you explain Russian gender-oriented names, and Spanish / Portuguese constructions. I don't know anything about that.

I found this explanation of Hispanic last names, which also explains that they do not use the term "Last Name".  It is likely that there are many other cultural issues with last names, and we only know the ones that we are familiar with.  The more comments I read, the more concerned I am that we could potentially make things worse by checking this, and offend people who have taken great care to list names correctly for families with whom they are familiar.

And the endings of Russian names can vary according to gender.  For instance with a name like Romanov, all the females in the family will use Romanovna (I think that is how it is spelled). Even if you are using the Cyrillic alphabet, the ending of the name is different for males and females in the same family.

I found this blog post on 7 naming conventions around the world.  To add to the list of issues already presented by others, Korean names are written with the family name first.  Afghan names  traditionally consist of only a first name.  Last names are added later, if needed, and often result in people within the same family having different last names.  Somalis have three personal names and no family name. 

Another article on naming conventions explains that Muslim citizens of the Ottoman Empire didn’t have surnames, and used their father's name instead, such as Ahmet oğlu Mehmet, (Ahmet son of Mehmet).  A surname law was  adopted in Turkey in 1934.

This post on Baby Names explains that in the South Indian naming tradition, there is no family surname. Once a man marries, his wife and children all take his first name as their surname.

+4 votes
So are the following profiles errors or not? I just pick a few random ones.



Lepelle dit Desmarais-1


St Denis-25

De Ribas-2


by Aleš Trtnik G2G6 Pilot (479k points)
Speaking for Wilcox-1428 vs. Willcox - They are both legitimate spellings.

Kitchen vs Kitchin-26 is probably also ok
How? Why? Did father Willcox-167 changed his lastname?
That's the way we do it on Wikitree. Last Name at Birth uses the spelling on government documents. Government documents early on were spelled phonetically by different clerks. He may not have even known how to write. So spellings changed all the time. One clerk spells it Willcox, another Willcocks, and another Wilcox.

Take a look at to see an example of lots of different spellings, all used during this man's and his father's lifetime
You cannot conclude who made the spelling error, without seeing the birth-certificate. If the clerk made the error then the name is legitimate, for the simple fact it shows so on the certificate. Transcriptions of those certificates will also show the same "wrong" spelling. .

Stclair-27 WRONG
Naftziger-2 RIGHT
Lepelle dit Desmarais-1 NOT WRONG
Kitchin-26 NOT WRONG
St Denis-25 RIGHT
Belshe-45 RIGHT
Wilcox-1428 RIGHT

+4 votes
If John Smyth's daughter is birth-registered as Mary Smith, obviously it's likely that John was spelling his own name Smith at the time.  I never heard of anybody saying to the registrar "I spell it Smyth, but she's going to spell it Smith."

This would make a case for showing Smith as John's CLN, and then of course no error would be flagged.

However, this isn't commonly done.  Should it be?

Putting it another way, there's a question whether this calls for a fix rather than a false error.
by RJ Horace G2G6 Pilot (561k points)
+13 votes

I suggest that we hold off on checking LNAB as a potential error.  There are many reasons why the child's LNAB is different than the father's name, and some of those reasons have already been listed by others here.  It is easy to change the LNAB, but it can have significant implications.  Though I am all for bold changes, this is one field where I believe we should not send the profile managers a note unless we have researched the LNAB and can provide an authoritative source along with the suggested request for review.

by Star Kline G2G6 Pilot (535k points)
You don't have to change an LNAB.  You can "fix" it by making the father's CLN the same as the child's LNAB.
Good point, but that works only if that information is correct.  If we cannot provide a source that indicates the father ever used that last name, we should not add incorrect, unsourced information to a profile.

My great grandfather's name is Παναγακος.  He had two sons who emigrated to the United States - one changed his last name to Panagakos and the other changed his name to Pannick.  My great grandfather never used either Americanized name, so adding either Pannick or Panagakos as a current last name would be incorrect.  To paraphrase your earlier comment, this would add random noise, not genealogically accurate information, to the profile.
But the Americanized names of the sons aren't their LNABs.  The name of a new-born baby is usually the name its father is using at the time, and is itself the evidence of that.

RJ, The key word in your comment is Usually. I know of a man with the surname of Marn who has a son that the birth certificate list's his name as Kellner. The reason is both his parents were married to other people at the time of his birth and the state said that the child could not have that name. Before you even go there his mothers CLN at that time was Hancock and this happened in Ohio, United States about 14 years ago so I do have a personal knowledge of this particular case.

We also have other last name field fort that.
Still would not fit. The state does it this way because there have been too many times that after DNA testing the presumed father was able to have the child support removed and it then caused other legal problems. Lets face the facts, it is not just my opinion on this one but at least 7 others agree that it would be a bad idea.There are too many legal and religious reasons why a child would not share a surname with his parents, and even in countries where the normal practice is for the son to share his LNAB with his father other factors would make that not happen sometimes. In the example I used the father was born Corbe and had his name changed by the courts to Marn so Kellner is not a name he ever had or used.

Yes, the key word is usually.  It's not like anybody's planning to write restrictions into the software that would enforce these rules.  The aim is only to make a checklist of things that are usually errors, so somebody can check them.  If they turn out to be correct, no harm is done.


+8 votes
As this gets more and more advanced maybe we could have this error defined but you get a checkbox or a listbox if you would like to show it or not in the reports.
by C S G2G6 Pilot (273k points)
Mangus, with about 80% of the answers saying it would be a bad idea I would say that the majority are against having this error even being checked.
Good idea, checkboxes for the field you want to check for errors.

And Dale you are against 100% of all new errors.... I think a check box solution displaying just "your" errors would be good for you and then we don't have to repeat the same discussion in every thread if you like the error or not ;-) instead focus to make the criteria better...

One dimension in Swedish genealogy is that after abt. 1900 it was forbidden to use patronymics but this year 2016 we got a law that it's ok again...

Interesting is to see EM in soccer and the Icelandic team  most players have still patronymics.... In Iceland the custom of using patronymics is prescribed by law, so that every generation must follow the old custom. Thus the most important name of an Icelander is his given or baptismal name and he or she is listed by this name in the telephone book, the patronymic coming last.(source)

==> if father Anders and before 1900 and Sweden then Andersson surname on child is ok

If father Anders and after 1900 and Sweden then Andersson is ok but it could also be Nilsson Eriksson or .... But Andersson could also be a bad guess that the son got a surname based on the fathers name....

Another non resolved issue is if the patronymic should be stored in the last name at birth or not. Some part of Sweden and Norway you also used the farm name in the surname.... 

Mangus , There is no need to put down my opinion. In this case I am only one of  8 who think this LNAB error would be a bad idea while you are in the minority of those who think this error should go forward, As for your checkbox idea, we already have it set up so that if I do not want to check for errors I do not have to even look at the list. Unless there was a box that allowed me to stop others from checking my watchlist for errors there would be no change but  my "negative" attitude for this project will not allow me to say that preventing others from looking at errors would also be a bad idea, so just how would a check box be an improvement.
Also in your reply to me you included another reason that this error would be bad, there are just too many reasons why a child would not have a surname that is shared by his parents.
Dale maybe to many parameters for a person not used to Swedish genealogy but a no brainier for a computer...

Doing genealogy is not easy. Ancestry green leaf approach has proven not the best way to do genealogy. WikiTree suffers a lot of unsourced imported Gedcom files and need all possible tools to great a family tree with good quality.... The Database Error project could help people beeing more critical to the unsourced facts they find....
Let me put this in simple terms. The green leaf approach by Ancestry is using a computer program to find matches. You would have a computer  program find things that should be changed, in your opinion, to match better. I agree the solution is to have better sources in profiles and that there have been a lot of problems from GEDCOM uploads but this LNAB error would not be something that could be solved by a computer program, there are just too many cultural and religious reasons why the child would not share the same LNAB or even the CLN of one of their parents for it to have better results than the Ancestry green leaf.

Good links about patronymics

+4 votes
Another, more modern invention is hyphenated last names of children. These include both the LNAB of the mother and the last name of the father.

I'm sure it doesn't represent a significant number, but this would also show up as a false positive error.
by Eric Weddington G2G6 Pilot (209k points)
It colld be ignored automatically.
+1 vote

Aleš, I think there are way too many good reasons for there to be a difference to ever make this a valid error. I wouldn't want to try to count the number of times a spelling change occurred when a family emigrated/immigrated to a new country. And, as mentioned, all the Scandinavian patronymics, Arabic conventions, Welsh conventions, etc. And what about adoptions where the child did not change their name? That was fairly common back in the day when the adopting couple knew the original parents.

And in my family, I have a branch where the spelling changed occasionally just because someone was tired of correcting people. My original (as far as I've been able to trace) many-times great-grandfather spelled his name Linch. The branch I'm on kept it that way, even thru the officials who kept writing land purchases & vital records using Lynch. I've even seen records where the father's name was spelled one way & the child's the other. Some of the siblings generations back gave up & used Lynch from then on. And some changed back to Linch; that occurred when someone met a cousin from my branch & realized the relationship. It's a real mess trying to keep track of who spelled it how.

by Kitty Linch G2G6 Mach 4 (40.6k points)
+1 vote
I encountered a colonial Watertown family with the head of the family was Browne and all the children were Brown. We use LNAB. If the record is spelled differently between parent's LNAB recorded in town A and the child's LNAB in town B, which is "right"? I have a copy of a letter written by my 2ggf, in which he states his last name is spelled Kennaston, but his enlistment papers spell it differently. Additionally, some of his siblings have different spellings. Calling the spelling of a child's LNAB different from the father's LNAB an error would not work out too well.
by Bob Keniston G2G6 Pilot (200k points)
I think those would be ok, if both last names are entered in generation where change occured. I would compare all lastnames against all lastnames and any word matching would make it OK.

But the error is on hold for now. Maybe in the future.
In the future then, would the error apply to families that are, say 100 years old or less.
+2 votes
How about an optional run that you could do on your own tree for a certain number of generations. It would generate name synonyms. This would tie into my previous interest to generate alternate spellings of surnames. Somewhere we could have a table of name variations. I suppose if someone wanted to know how prevalent certain variations were a run on a small tree could be interesting.
by Sue Hall G2G6 Pilot (133k points)

Related questions

+11 votes
4 answers
133 views asked Sep 5, 2017 in WikiTree Tech by Karen Harbert G2G6 Mach 2 (26.7k points)
+12 votes
1 answer
220 views asked Jul 24, 2016 in The Tree House by Robin Lee G2G6 Pilot (638k points)
+4 votes
1 answer
312 views asked Mar 7, 2017 in Genealogy Help by Jim Parish G2G6 Pilot (157k points)
+11 votes
2 answers
+9 votes
1 answer
114 views asked Jun 17, 2019 in WikiTree Tech by SJ Baty G2G Astronaut (1m points)

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright