How should we evaluate photo recognition software as proof of paternity?

+6 votes
367 views
Family stories about descent from an illegitimate child of a royal father may be hard to prove by documentation - there often isn't any. They may also be hard to prove by DNA, since the royal houses are notoriously unwilling to submit DNA samples.

Portrait likeness has been the basis for numerous claims of royal paternity among the common folk, in all variants from a general agreement in the village that someone is "the spittin' image" of the old king to actually displaying a portrait of grandfather on their living room wall next to a portrait of the king.

Today we have facial recognition systems. Does anyone have experience of using such systems for genealogical purposes?
WikiTree profile: Carl Lind
in Policy and Style by Eva Ekeblad G2G6 Pilot (573k points)

2 Answers

+11 votes
 
Best answer
Seems like it would be very weak evidence. Facial recognition apps will just tell you the percent of similarity between two faces. Even a high percent of similarity wouldn't mean much. Unrelated dopplegangers are very common and, unlike DNA, facial recognition can't distinguish between people who are related and people who look related.
by Chase Ashley G2G6 Pilot (312k points)
selected by Andrew Payzant

My impression from reading the legitimization document and Malvina's later admission is that the former is genuine and Malvina was browbeaten into signing the second.

The legitimization document explicitly mentions that the boys would be considered as of "marital bed" and that they would have rights to inherit. I find it hard to believe that John Lind would not have understood what he was signing.

And yes, the fact that the boys - well, young men by that time - sent a proxy to the execution of the estate inventory, strongly suggests that they still believed Johan Lind the elder to be their natural grandfather. Which means that Malvina did not tell them anything before her death, in spite of them being "big boys" when she died. The alternative, that they knew they were the sons of a king and yet turned up to get a share after the old Johan Lind, paints a less flattering image of them as persons.

As for the issue about the use and evaluation of facial recognition software in genealogy, I had hoped for more discussion. As it is now, there have been two responses against its use, plus one more against over the Sweden project group.

I think that if we are going to use facial recognition software for genealogical purposes in WikiTree, there should be a policy and instructions. I'm not sure how to get input from someone with experience from a neutral perspective. It may have been a mistake to link the question to a specific case, as I did.
In the light of Malvina's two letters about the father of her two boys, it should be mentioned that the birth record of Henrik Casimir Lind has a notation mentioning John Lind. This record is at Klara Parish, where Johan W. A. Lind was also born, whereas Carl Henrik's birth was recorded in Adolf Fredrik parish. Whether there is any significance to this, I don't know, except that I have often wondered why John Lind is mentioned in Henrik Casimir's birth record. My thought was he could have been the father of Henrik Casimir, even if not of Carl Henrik.

Other tidbits about Malvina, which could be pertinent, are that, on April 28,1863, she took out an ad in a Stockholm publication for a cafe in the Djurgarden with which she had some association. The Djurgården was owned by the royal family.  

And she is shown in some Swedish Household records as a Sommerska, which I'm told was a private dressmaker, often to an important woman, such as a lady-in-waiting.

Yes, I see now that the vicar of Klara parish added a note to Henrik Casimir's birth record when HC was acknowledged by his parents, while the vicar of Adolf Fredrik did not add a similar note to Carl Henrik's birth record. I think I have mentioned the note from Klara before, possibly in the email discussions. The note refers to "legitimis. Nr 133", which is the same document there is a copy of in the estate inventory proceedings, signed by the vicar, indicating that the copy is correct.

The digitized version of the birth book for Klara does not include this type of appendices, so there is no digitized image of the original note, but I think we can trust the vicar and the magistrates court: there was such a note, concerning both boys, signed by both parents, shortly after their marriage.

There are legitimization notes included for some of the other children in the Adolf Fredrik book. I just looked at a few - and it seems the usual format is a lot less explicit than the note for the Lind boys; the mention of inheritance rights seems to be a rare ingredient.

Also, yes, I have seen that Malvina is registered as "sömmerska" or "syjungfru" in some records. It does mean seamstress or dressmaker. It does not imply anything in particular about her being employed privately by some specially important woman - although her customers would have to have enough money to to pay for her services.

I have also seen that from some time in 1862 to January 1863 Malvina lived in Uppsala, where she is listed as a "Mamsell" among the serving girls at the restaurant in Rådhuset. So it's no big surprise that she was associated with a café soon after her return to Stockholm.
Being Swedish certainly gives an advantage in being able to read text!  I can only read names, dates, and places, but text only when the Google Swedish to English translation works, which it often doesn't with the Swedish of old. Which brings up a question.  Is there a better way to translate old Swedish than Google?
As far as I know there isn't a better way than Google - but I have not really looked for any other online translation service. Sometimes a translation of single words works better than phrase translation - but sometimes the other way around.

For specific stuff like occupations or titles I often look it up in Wikipedia and see if there is a counterpart in English. I also still, sometimes, use my old paper dictionaries, between Swedish and English and an English-English dictionary (which has some etymology).

But of course it is a lot more difficult (or impossible) to read old handwriting when it's in a language you don't speak - you have to be able to get something out of the scribbles before you can put it into Google.
While I can understand that the use of a single photo to declare relationship to another person could probably only happen if the people were twins, that is not what I did.

To begin with, my grandmother's family believed that Carl XV was the father of Carl Henrik Lind.

So, I used the 6 photos of Carl Henrik Lind that I inherited,  from about age 6, to adolescent, to late teenager, to fully grown young man, and as an elderly man. I compared these photos on a one-to-one basis with a variety of individual 19th c. Bernadotte males and took the average match for the whole group of photo comparisons.  I did the same with comparing a variety of the same Bernadotte males to each other and took the average of the whole group.

Then I did the same with Carl Lind and his adult offspring (mostly women) and made an average of that. The average matches of each group were remarkably close, though the family group with more females than males was a lower percentage average match than those of both of the all-male matches.

It is based on this sort of matching that I came to think that Carl XV might well be the father of Carl Henrik Lind, especially given family lore and evidence which we have.
I was trying to reply to the comment by Chase Ashley and Andrew Payzant, which was based on the idea of single individual photo comparisons.
I happen to have two photos of Johan Andersson Lind. I know this because one of them was in an Aftonbladet appreciation obit. shortly after Johan Lind died in 1897.  My 2 photo comparisons between him and Carl Henrik Lind were not particularly at an average of 33%. Carl Henrik's several photos (I have 4 photos of Carl Henrik taken between about 1870 and 1890 + 2 more in old age) show an average match of 71% with many and various members of the Bernadotte family. The same 12 Bernadotte photos average 73% among themselves only. As for eyeballing resemblances, I thought of using facial recognition software because several visual comparisons had enough similarities to make me very curious.
+3 votes
No experience with facial recognition software, but it's an interesting topic. My long-deceased paternal grandmother told a lot of stories, shall we say, that cast my father's paternity into doubt. And then we found a picture of the man he had always known as his father. They could have been twins. No DNA test (or facial recognition software beyond the Mark 1 human eyeball) needed to determine paternity there.
by Carolyn Comings G2G6 Mach 5 (52.6k points)
This is interesting too, but I think it's different from what I've been trying to say. For one thing, I'm not trying to use one photo to claim that Carl Lind was a particular person. I've used several photos of Carl Henrik and compared  them with several photos of a few Bernadotte males, and have found an average result which suggests that Carl Lind was likely to have been sired by a member of the Bernadotte family, in this case Carl XV, as his family believed.

This quest came about because, in the mid-1980's, my grandmother (who lived in America) got a letter from Sweden (presumably from a relative) which stated as fact that "it can now be told" that Carl XV was her father's father, as if a series of family members had known, passed down, and agreed to cooperate with the secret for a set period of time. Other family lore and the evidence of her own eyes suggested to my grandmother that something like this was probably true.

In addition, a WikiTree volunteer once told me that if I can find no paternal-side ancestors of Carl Henrik Lind, it becomes much more likely that he was sired by an important man who wanted to keep it a secret. The DNA matches for my mother have turned up many people related to her other three grandparents or their descendants and ancestors, but not a single one (beyond his own offspring and their very few offspring) who is related to Carl Henrik Lind. It's a dead end.

In any case, I wouldn't expect WikiTree to change their standard of proof based on one possible case, but I do think it is arguable as evidence and provides a possible link with the past that could be interesting to pursue in the future.
Emily, you did say somewhere that you had contact with a male descendant of Carl Henrik Lind in Sweden? Or was it a descendant of his brother, Henrik Casimir?

Did they ever check their DNA against Jan Bernadotte? https://www.wikitree.com/wiki/Bernadotte-69

Jan was the "Bernadotte black sheep" (according to himself). He died last September, so his profile is now public. He had, for a time, a company that offered comparison against his own DNA.

Related questions

+21 votes
3 answers
+4 votes
1 answer
221 views asked Dec 9, 2017 in Policy and Style by Tim Perry G2G6 Mach 3 (35.3k points)
+18 votes
0 answers
280 views asked Dec 17, 2016 in The Tree House by Mags Gaulden G2G6 Pilot (642k points)
+6 votes
2 answers
759 views asked May 28, 2017 in The Tree House by Geoffrey Hamilton G2G1 (1.5k points)
+10 votes
4 answers
+8 votes
10 answers
435 views asked Sep 15, 2023 in The Tree House by Richard Fuggle G2G5 (5.2k points)
+6 votes
2 answers
+8 votes
3 answers
325 views asked Feb 27, 2021 in WikiTree Tech by Phil Kennedy G2G6 (6.9k points)
+10 votes
6 answers
206 views asked Dec 29, 2020 in Genealogy Help by Ryan Ross G2G6 Mach 3 (39.6k points)

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

disclaimer - terms - copyright

...