I hope everyone will agree that the definition of a "family unit" is very subjective. Although many products, and the old GEDCOM format, offer some type of FAM record, I have never seen an agreed definition of it.

Most of us have found family groups where not all the relationships are full-blood ones, or simple parent-child ones, and these may include step children, second spouses, adopted children, in-laws, etc

One convoluted example goes as follows: father deserts household, youngest children put in industrial school, wife has relationship with new man, new man has existing children but new ones born to pair of them, deserted wife dies, partner starts relationship with new woman.

In order to cope with scenarios like this, I want to develop a framework that will accommodate custom definitions of a family-unit, including variants in the same data (as a simplistic example:. natural-family, extended-family). I may even extend this in order to accommodate any type of person grouping.

What I'm after is any information about how this is tackled elsewhere. Are there any products that accommodate different types of family unit? If so, are the terms fixed or can custom ones be defined?

  • I suppose your model should also record the temporal aspect of membership: foster children may come and go, children grow up and move out (or move back in!), spouses divorce and remarry, etc. Another aspect is geographic: must all members of a family-unit live together at the same time? Surely not... yet the boundaries here can get blurry. Nov 27, 2012 at 21:48
  • There's a fine line between "family unit" and "household". The latter could include staff, but is otherwise pretty close to "family living together" (on a given date). BTW, if you're looking for test data to model families and marriages, take a look at Richard Burton and Elizabeth Taylor. Using public information in for example Wikipedia you can find their marriages, all the people they married, all the people they married married (and so on), and the many resulting mixed families.
    – Rob Hoare
    Nov 27, 2012 at 22:32
  • While it doesn't include any descendants of any of the marriages, you might find this tree interesting. Would you consider all these people to be one "family unit"? trees.ancestry.com/tree/26095047/recent?pg=42 Nov 27, 2012 at 23:43
  • Membership will certainly have a temporaral aspect if I want to cope with, say, a club or society. Geography is implied in the case of a household but should not be mandated for the general case. I separate the concept of marriage (which is dependent upon both culture and life-style) so my own variants of family-unit are mainly based on social groupings such as living together.
    – ACProctor
    Nov 28, 2012 at 11:20
  • A genealogical definition of family? I've stated before that "A genealogist should record what a family was. A genealogist cannot dictate what a family is." (Family in Scientific Genealogy), observed that families change shape over time, and offered the solution that "Genealogy software should allow the genealogist to group individuals into families.", but all that is broad strokes. I'd love to see an implementation of a detailed approach to family. Dec 2, 2012 at 13:52

3 Answers 3


One possible representation of a family unit that could work quite well is one that is location and time based. That is, everyone who is living in the same residence at the same time is considered a family.

I'll try to illustrate this with this series of examples: A man and woman move in together (whether they get married or not). They are now a family. They have a child who lives with them and is now part of the family. Their parents get sick and move in with them, so they become part of the family. A neighbor's child becomes an orphan, and moves in with them, and becomes part of the family. Their child grows up and moves out on their own and leaves the family. A parent goes to a nursing home and leaves the family. The entire family moves to another home but stays as a family.

What is critical here are that events happen at certain dates, where people enter and leave the family. Families can merge. Families can split into two.

For your "convoluted" example: Father deserts household (leaves family), youngest children put in industrial school (leave family), wife has relationship with new man (he only becomes part of her family if and when he moves in with her), new man has existing children (only family if they moved in) but new ones born to pair of them (most likely with mother and part of family), deserted wife dies (she leaves family), partner starts relationship with new woman (does she move in? Is everyone else still with partner? whoever is, is part of his family)

The one really nice thing about this system is that you can precisely and unambiguously define who is in a family and who is not. There is no relationship required. If unrelated friends move in for a month because their home burnt down, then for that month, the families become one.

My reasoning why this method may be best is what I encountered with father's family. He grew up with his mother, sister and step-father. They were his family. His two brothers grew up in an orphanage. They were indeed full brothers from his real father and mother, but they were not the family he grew up with.

My wife's family is another example. She grew up with her grandparents living with her. They were always treated as part of her family while they lived together, and any definition of family is wrong if they are not included.

The one disadvantage of this definition is that a family will be quite dynamic, with frequent events that causing people to join and leave it. As a result, some family groups will only last days, and one person will be in many different families during their life. This may be difficult for a program to properly present and make useful, but it sure would be a heck of a thing if some smart programmer figured out a way to do it.

  • Thanks @lkessler. This is the general approach I was hoping to take. There will certainly be a dynamic aspect to people joining or leaving (defaulting to their death of course). I have a paper definition in mind but no software yet to support it. It will include some basic SET operations to ease manipulation and allow construction of groups from other groups. The group type would be a 'partially controlled vocabulary' implemented by a URI but I already use this approach for many of my types/sub-types/etc.
    – ACProctor
    Nov 28, 2012 at 11:31
  • In addition to, perhaps, being difficult for a program to properly represent, it's not how most people think of families. Are kids at summer camp and students away at school really no longer part of the family? Does Dad leave the family every time he goes to Chicago? If he's got another family in Chicago does he change families twice a week or is he really part of both simultaneously?
    – Tom Morris
    Nov 30, 2012 at 13:47
  • @TomMorris - Defining it this way is relatively easy. Families should be defined by permanent residences. Temporary residence changes or even armed forces service or extended vacations would not remove someone from a family. However, if a child goes off to camp for 3 months, then they temporarily have left their family at home and for the summer have a new extended family made up of their counsellors and campmates. And these temporary families could also be tracked. Doing so moves this analysis into the realm of relationship tracking which would be another very useful tool for genealogists.
    – lkessler
    Nov 30, 2012 at 14:50

Family - First referred to the servants of a household and then to both the servants and the descendants of a common ancestor. It comes from Latin familia, "household; household servants," which came from another Latin term, famulus, "servant." It was not until 1667 that the term was used specifically for parents and their children. Farlex Trivia Dictionary. © 2012 Farlex, Inc.

Common terms such as "natural family" (man, woman and their direct offspring) and "blended family" (man, woman and offspring not necessarily of said man and woman) try to differentiate between these family types.

Also missing from your examples is common-law couples that may be recognized to varying degrees by their country's legal authority. (Stieg Larsson comes to mind. Were he and Eva Gabrielsson a family? Would they be recognized as one if they had had children?)

I am all for customizing terms for family types that add clarity, but that do not imply judgement upon the individuals. The term "natural family" may be offensive to some. How about "chosen family"?

Are there any products that accommodate different types of family unit?

  • Family Tree Maker 2012 has a "Blended Family View" function.

  • Genograms visually show the relationship of all members of a family. These can be customized with the relationships that you think are important.

  • I mentioned above about the fact I already separate "marriage" from "family-unit", and for the reasons you give. Thanks for the derivation of the term "family". My own usage is for different types of social grouping in a household and so has a strong match with your suggestion here.
    – ACProctor
    Nov 28, 2012 at 11:23

Because of its religious background, the FAM record has a particular world view, but it's actually got the core of what you need. It links a father, a mother, and a set of children. It has some accommodation for things like adoption, but other than that, all the relationships are fixed.

All you would really need to do to accommodate what you want is to make the number and type of relationships variable (ie N parents, arbitrary family member relationships).

If you really wanted to get fancy you could model family changes on more than just an individual level (e.g. this family split to form these two families, these 3 family fragments joined to create this blended family), but there's a rapid diminishing rate of return. You need to think about how much work you would actually be willing to invest in modeling and recording this information. Simple and flexible is likely to trump detailed and comprehensive.

  • This scheme won't work for me. Biological relationships are fixed and finite, whereas all other types of relationship are time-dependent and possibly overlapping. I therefore model them very differently. A social grouping related to a household (my nearest equivalent to the FAM concept elsewhere) is also time-dependent whereas the FAM record has no time dependency at all (..another weakness of it).
    – ACProctor
    Nov 30, 2012 at 15:36
  • The FAM record does have dates. They're focused around marriage because that's what the designers though was important, but it's not much of an extension (from a technical point of view, anyway) to just make them generic start and end dates for the family unit. Similarly the 3 fixed relationship roles (husband, wife, child) could be made more general. homepages.rootsweb.ancestry.com/~pmcbride/gedcom/…
    – Tom Morris
    Nov 30, 2012 at 22:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.