Source vs Repository

I have a number of Record Offices/Archives in the UK which I class as sources with the Repository as being my means of access these same Sources are also in my list of Repositories because I access them directly usually for information about record holdings but other times using the Microfilm/Microfiche devices and that is where the physical records are deposited, was wondering if other people handle this differently.

1 Like

For me, the Record Offices/Archives is never a “source”. It is where the Sources are kept and maintained. A Repository.

The source is the spool of microfilm you accessed at the archive to find the bit of information, the citation. The source is the Book from the shelves at the archive, or the electronic database from the archive’s computers.


I agree. Record Offices/Archives should be repositories. The documents and
artifacts they contain are the sources.

A website such as is strictly speaking a publisher. I can understand why it could be convenient to class it as a repository.


I rethought how I classified sites like Ancestry and FamilySearch. I have them as repositories. I go there and search their catalogue of databases to find the bit of information.

The term to describe them that I think applies: Aggregator.

They are not Archives. If their computers were shut down, no information would be lost or destroyed. It would just no longer be easily identified.

1 Like

Thanks for the replies

I am beginning to think I am trying to adapt the real and changing world to the restrictions of GRAMPS (this is not a complaint, just a fact).

The Category structure of Repository, (Publisher), Source, Citation does not match the reality and would be much better replaced with a hierarchical structure similar to Places that can be user defined.

Best if I give examples.

Say I am looking for a birth in Manchester, Lancashire, in the 1840’s to order a certificate.

I would first go on one of the National Indexer sites (GRO FREEBMD etc) and get a year and a GRO Reference Number (Citation), I want to order from the local Register Office to whom this GRO Reference is meaningless so I use the another Indexer (UKBMD) and get the Local Reference Number (Citation).

This goes to the Register Office in Manchester who then have to go to the Archives of Manchester Central Library where all records over 100 years old are stored due to lack of space in the Register Office find the document and either scan or hand copy.

In order to make space for the Register Office Records which have to be close to the Register Office Manchester Archives have taken other materials (usually there will be film or fiche copies) and stored them in a “long term storage facility” outside Manchester (disused Salt Mine NW England).

Why do I do this, legally in the UK Birth, Death and Civil Marriage records cannot be published or made freely available to the public.

If I want a Church marriage in Manchester from the 1840’s I would first go on one of the National Indexer sites (GRO FREEBMD etc) and get a year and a GRO Reference Number (Citation), I want to order from the local Register Office to whom this GRO Reference is meaningless so I use the another Indexer (UKBMD) and get the Local Reference Number (Citation).

However I can get around this because I know that the Marriage Registers have been Indexed, Transcribed and Digitised by Ancestry (and others) so I can get the image of a certificate for the cost of my subscription.

Should I only want a transcription, I can go onto an Indexer/Transcriber website and get a transcription of the certificate for free… which as a Citation will include the FSO film number. This same film being available for me to go and see in Manchester Central Library where it would have a completely different Citation (Reference Number). I can also see Marriage films which have not been digitised by the commercial sites for legal reasons.

So you can see numerous routes to achieve the same end. With different Citations

Film and fiche are coming to the end of life not necessarily because they are deteriorating which some are but because film and fiche readers are getting harder and more expensive to maintain.

Another example would be:
The UK Probate Service

If I want a will, I go on the Indexer website locate the will and order online a hard copy at a cost. These are all held in one physical archive which is now full and it is an expensive process in terms of staff costs to send out hard copy. So the proposal is that hard copies of only the famous and wealthy will be kept as paper copies, the rest digitised and the paperwork destroyed (as part of the UK contribution to global warming) because of the cost of storage.

Apologies for the verbosity and giving UK only examples but is the only way I could explain my thought processes and I am sure the same or similar issues are occurring all over the world.


1 Like

But you do not have to create a Source/Citation that takes into account all possibilities. You create a S/C with the process you actually used. And if you use a second (or third) method they become their own S/C.

I can get many of the same records from Ancestry and FamilySearch. The S/C I add to the record depends on the method I used. In both cases it is usually a scan of the same record.

You mention the Register Office in Manchester. That the R/O may have some of its records stored by the central library or stored in a salt mine is irrelevant. To get the record you go to the R/O.

Gramps gives you the three components. Repository, Source and Citation each having various fields. Each user fills the fields so that another person has a possibility to find the exact same tidbit of information.

Remember, information put in the Repository does NOT get added to the S/C printed in reports. This is why I use the Repository to signify where I went to obtain the information.

You may find this thread of interest.



I am doing precisely what you are doing for simple practical reasons,
but trying to explain to people why I do it is difficult. Particularly
if you look at the thread you highlight and Nick’s second paragraph

“The repository is the physical location, such as an archive, where the
source can be found.”
and most dictionary definitions
“a place where things are stored and can be found”

Neither of which cover data bytes (which can be altered or corrupted or
cease to exist at the flick of a switch) floating around between
multiple hopefully secure server systems

For what it’s worth, just as a microfilm image is a facsimile of an original paper record, so also a digitization is a facsimile of a microfilm image. Each facsimile is it own derivative record, housed in a different repository of sorts. I don’t know the exact location of the digitized images I find on a given site (they’re probably on some third-party cloud storage, which means they could be scattered across any number of servers in various locations) so I just refer to the archive as FamilySearch for example. But I do store the original microfilm roll number as the Call Number in the repository reference. And in the source details I describe the original paper records that were microfilmed, even though I have never seen the originals myself.

My goal in doing it this way is to make it clear where I found the records, but also enable myself and other users to someday find the originals if possible.

I think the ability to have layered citations might also help with this.


Hello Phil,

As you can read in that other thread, our current model is largely based on the GEDCOM standard, and even if we’d implement version 7, you won’t see much improvement. That means that you will either need to find a way to deal with the features that Gramps gives you, or migrate to another program like Family Historian or RootsMagic, two programs for which I know that they offer citation templates inspired by Evidence Explained, written by Elizabeth Shown Mills. She once gave me a free copy of her book, but I lost that, and if you want to read more, you can check her web site at. You can find some free examples there too, if you want.

I run RootsMagic Essentials on Linux Mint using Wine, for its integration with Ancestry, FamilySearch, Find My Past, and My Heritage, and it’s quite stable, although some parts work better when you run it in Windows. It has hundreds of citation templates, and the Essentials version is free, and the full version comes with a 50 % RootsTech discount this weekend, so that you can get it for $ 20, if you decide before monday night.

Family Historian has similar features w.r.t. citations, but has no support for Ancestry or FamilySearch. It also costs more, and only has a 30 day free trial.

I don’t use either as my main program, because they don’t work in Dutch, and I really like to be in control, using open source, but if citations are important to you, they might have something to offer, if you accept some vendor lock-in.

The words that this Englishman in Leiden wrote in 2011 are still valid, so that’s what it is.

Hi Enno

Really started this thread to see what other people are doing I am happy
with GRAMPS so as they say I can either “like it or lump it”.
My problem is I have no real need for GEDCOM I use it maybe once/twice a
year I maintain my “Online tree” for use in the likes of Ancestry etc
offline it is simplified and with whole years for birth, marriage, death
and locations are at county or country level. So I then delete the
online tree and import the updated GEDCOM. Eliminating all the junk that
Ancestry might have inserted with me playing around.
Also I would never dream of using FSO for anything other than record
hunting, and that is the only thing I would download, I refuse to put a
tree on FSO or Wikitree because I am not in control everybody and their
brother can make alterations.

1 Like

Hi Phil,

Nice! I work pretty much the same, and use Ancestral Quest and RootsMagic as external add-ons. My main tree sits in Gramps, and I recently bought a one year subscription to My Heritage, to find relatives here on the continent.

I have data on Ancestry, FamilySearch, Geni, and My Heritage, and even spent some time on wikitree long ago, and everything that I find worth saving goes back into Gramps.

I’m also still looking for the best way to record the types of call numbers that Dutch archives use. They have collection and item numbers, which still play a role if you want to register where the real (church) book can be find. That system is still in use, even though most of these books were indexed, and are also available as films on FamilySearch.

I’m also still a pit puzzled about the proper citation of this document:

Can you help with that?

But seriously: Have you checked what we have about the subject on the Gramps wiki?


My head is shredded with the last few days wish I had not started
thinking about it, I have read the wiki but not recently

OK the Trye or de Trie there are 4 documents
Repository: Gloucestershire Heritage Hub or Gloucestershire Archives
(for you to choose)
Source: Trye family of Leckhampton
Citations: D303/F38, D1809/F5/1, D1809/F14, D303/F31
D1809/F14 is the Pedigree

Here are details
Orderable Yes
Date 19th cent
Finding reference D1809/F14
Held at Gloucestershire Archives
Number of items 1 roll (outsize)
Level Item

I just make a repository for each type I use and link them all to the source…
I give them each repository a type, use custom if I need to…

  • Digital Images (copies) Repository
  • Digital Index Repository
  • Original Source Repository

etc. etc.
When I have multiple repositories for a source, I create a note where I write where I actually found the document I am using.

I really wish for a multi-level hierarchy for both Sources and Repositories that can be linked together and that Repositories also could be linked to Places…

I mostly use Obsidian and/or Foam for VSC for all my research in addition to a network graph software… there I can easily link any note to any other notes and see a network graph where all the relations/links will add up, that way it is a lot easier to see where sources are stored and who I find in the different sources, and if there can be a connection between multiple people based on what sources they are recorded in… when the repositories and especially the sources start to add up, it is often easier to see these connections in a graph rather than in a table or list…
I also store all my sources in Zotero, that way I can create Collection hierarchies both for repositories and sources. I can also link documents/sources together or create a sub-item for a document to get a more detailed citation or when multiple objects “use” the same source, but at different pages in a document/book.

It is a lot easier to keep order when there is a possibility to use layers for this type of information…
I also add this information to a YAML header for the item so that not only can it be used to generate visual information in that software, but I can also export the notes to different formats if needed and use the YAML headers as information for how the Note shall be transcoded/translated to a new format.

I can also add those Notes to Gramps or any other software I use as media or media link and easily open them to update the software if needed…

Yes, I do use a few software but I use Markdown as a hub so that my information are as “synced” as possible between the different software I use…

This way I can store my information almost as I want it (still brushing the edges of the workflow) and use the same written information in multiple software without rewriting the lot.
I actually just copy/paste the citations- and bibliography- string from Zotero to Gramps or other software and then add the research notes to the subject/object (Gramps object) I am researching…

Thanks for that more to study but I think it proves the point about
GRAMPS (Sources/Citations) it seems everyone is going to extraordinary
lengths to get round the limitations of GRAMPS

I like the idea of a hierarchy of sources to support layered citations.


IMO, these are not really the limitations of Gramps, but the limitations of GEDCOM 5.5.1. and even 7.0 for that matter. Both commercial and open source developers know, that there is not much to gain when their competitors stick to the same standards, and they also know that noone adopted GedcomX, which already has support for layered citations. And when I write noone, I mean that noone adopted the format for imports and exports.

In reality, many desktop programs do use GedcomX, because that’s needed to exchange data with FamilySearch, meaning that Ancestral Quest and RootsMagic already speak GedcomX, to the site.

OTOH, when you look at FamilySearch now, you will see that they use a single layer, meaning that the whole distinction between repository, source, and citation has gone.

Hello Phil,

I can understand what’s going on in your head, and I have to admit that my example was a bit of a trap. It’s a document that’s been on my reading list since years, and I did not add it to Gramps, because I had no chance to read it, until today.

In this particular case, the document is a book, published in France, for which it seems that the Gloucestershire Archives have a copy, made by a cousin, in the 19th century. And because it’s a book, and the other 3 documents are all different, I will not treat them as one source. Using the higher level as a source only makes sense, when I see multiple volumes of the same type, like a collection of church books with numbered volumes. Most Dutch archives have such collections, and in those cases, it does make sense to cite the collection name as the source, and refer to the book number and page in the citation. And that also makes sense, because these books all reside in the same archive repository.

If I’d seen this book in the Gloucestershire Archives, I would put the title in the source, and its reference in the call number, because it is specific for this archive repository. And the reason to do that, is because it’s a book, that is most likely available in more archives than this. That too is a reason to not see it as part of another source, because it was never createt that way.

And in fact, a Google search that I did today, revealed that it was published in Paris in 1730, and has already been scanned by Google, so I can now download it in PDF format. I found it om this site

but I think that I will just cite it using the author and title as listed, probably with Google books as its publisher.

For other types of sources, the Evidence Explained site has a few free examples, and free articles, like this one:

I picked this example, because the citation model used by Ancestry is exactly the same as ours, and 100 % GEDCOM compatible.

Yeh, I think both that and Main-Sub Events fits very well into a tool like Gramps, even if many may say they not going to use it to use it, they most likely will if the feature is accessible.

As long as it is easy to transcode a collection to a flat comma separated string, it is no problem that it is not directly supported by the obsolete format of gedcom.
It will easily be possible to fit it in the Gramps xml format and any other xml or Json format and I think that is what’s important… and then it can just be made as a comma separated string when included in a gedcom, regardless of how deep the hierarchy is…

I hope you can get the rest of the team of core programmers with you on this…


No trap different thought processes I found a Gloucestershire Collection
I make no distinction between book, certificate, map, pedigree chart.
So I did the citation as presented.
If I later found evidence of an alternative or actually visited the
Archive to inspect the items I would maybe? adjust the citation to suit.
For the me the sole purpose of the citation is provide a
paper/electronic “trail” to where the supporting evidence lies,
currently Repository, Source, Citation.

Sticking to Standards is all very well but I have found they never keep
pace with the real world usually 10/20 years behind and are created by
organisations with vested interests, whilst it maybe convenient for some
for GRAMPS to match and support the latest GEDCOM, is it absolutely
essential for the majority of users? Is there a compromise in terms of
producing usable GEDCOM exports but moving GRAMPS onto hierarchical

Hello Phil,

Although I understand what you mean, I don’t believe that many people really want hierarchical citations at all. They’re hard to enter already, with our current GEDCOM based hierarchy, and even though the archives store documents in fonds, most users just want to register whatever reference(s) they find. There is no need to register that either, because in this case I’m only interested in the book, and I really don’t care about the fact that the refence number represents a hierarchy, with its numbers between slashes. And most users don’t care about the two level hierachy in Dutch archives either, as long as they have fields for the numbers that represent that hierarchy. One of the most advanced programs made here, in The Netherlands, has a single form where users can enter both numbers, and I see the same in the examples given for the template based entry methods in another one.

I joined the Better GEDCOM community in 2010, and it failed. And the organization that came after it, called FHISO, based on the fantasy that they could set standards like the real ISO, failed too, just like the real ISO failed with its 7 network layers. It failed, because most users and developers were happy with TCP/IP, and felt no need for more.

And as an engineer, and a scientist, I don’t care about hierarchical citations either, as long as I can enter the data that I need for a proper citation. And for that data, I don’t need more than a dozen fields, which can easily be created as part of a single object. I can process all known examples from Evidence Explained with that number of fields.

Moreover, when you look at the reference managers used by scientists, like Zotero, you will also see a flat model, an no hierarchy. And if they can deal with that, why would we need more?

Here in The Netherlands, the most popular programs, Aldfaer and Pro-Gen, have one single source line, that users can use to format their own citations, and they are happy with that, just like they’re happy with Geneanet, which offers the exact same thing.

And in fact, who are we, if even FamilySearch moved to a single layer, at least in their presentation to users?