* On hold until 23 Oct 2023: Viewing and editing citations

[AdrianBruce]

... Where in GEDCOM v5.5.1 is a definition that states the WHERE_WITHIN_SOURCE can be written in the Source_Record for an article or a census? ...

It doesn't have to be - the Title can contain absolutely anything you like when it's not a published, book-like source. It has to be a Description, not a published title. Yes, the example might indeed be leading us down one route but it wouldn't be the first GEDCOM definition ignored in the breach by many of us (e.g. Places aren't just jurisdictions, Addresses needn't contain the full postal address - whatever that means when you're recording stuff from before postal services )

[AdrianBruce]

The ideal system is where each distinct data item is recorded once, and only once.
...
Splitter or lumper is then driven largely by the capabilities of the storage system used and how far you are prepared to deviate from that ideal.

Totally agree. Though I'd say it's not just the storage system, but the User Interface as a whole. That's one reason (of several) why I don't like pushing stuff from the Source-Record down into the Citation - it's just not as visible to me. Although citation level visibility might have improved somewhat with v7....

[tatewise]

Where in GEDCOM v5.5.1 is a definition that states the WHERE_WITHIN_SOURCE can be written in the Source_Record for an article or a census?

Splitters typically put the full details of the specific document in the SOURCE_DESCRIPTIVE_TITLE (FH Title box) or the SOURCE_PUBLICATION_FACTS (FH Publication Info box).
That combines the source "collection" details with the "Where Within Source" details.
Alternatively, in FH V7, they use Source Templates that define custom fields to hold "Where Within Source" details.

[davidf]

although the various providers seem a touch inconsistent in their naming of what a "collection" is. Collection? Dataset? etc.

If by this you mean the sort of "meta-collections" that FMP are creating; they are a PITA!

I think "Yorkshire Marriages" is a good example - you have to be very careful to check: (1) Is this a record of marriage or of banns, (2) Is this the parish register or some bishop's transcript? Sometimes the "transcript" (in truth more an index page) will tell you, sometimes not (or even tell you wrongly - parish registers are not usually written up in a consistent hand month after month!)

So what would be my "source"? I don't think "Yorkshire Marriages" cuts it - you cannot begin to form a view of the general reliability of the source - and Banns are not evidence of a marriage. So with these meta-collections, I try and identify the sub collections.

Sometimes if you go to early pages of a set of images (and "Yorkshire Marriages" has a large number of sets of images - you can't say "Image nnn of Yorkshire Marriages"), you can discover the identity of a sub-collection which is meaningful "Deanery of Craven, Bishops Transcripts, Skipton 1600-1845" or "PE165/32 Scarborough St Mary Parish Marriages 1885-1889" - but I suspect that you have to be careful that a set of images does not cover more than one say parish, in which case the first images can be misleading.

You have to go to the images and work out what physically they are part of - which to me under lines the need to associate the original physical reality with your source. It gets messy - particularly if you want to refind the record either in another providers' "collection" or possibly even in FMP's collections should they subsequently recognise that they are creating monstrosities and re-organise them!

[KFN]

I’m writing this from my phone with one finger.

Adrian Bruce said:

It doesn't have to be - the Title can contain absolutely anything you like when it's not a published, book-like source. It has to be a Description, not a published title.

Yes I agree to a point, but when the Source is a “Collection” of individual items such as collection of BMD certificates, be they physically bound or logically grouped together in a named collection (such as a fiche, image collection) these are considered in the library setting collectively as a source because they are stored together. In cases outlined in the GEDCOM definition for Title, some of the examples for unpublished works such as a Family Bible, the individual information for a fact could be on one of many pages, for a letter it could have several pages or paragraphs, unless you have an original copy of a property title transfer, most of these items still need or could make use of the Source_Citation PAGE tag rather than put it in the Source_Record TITL tag.

I agree that the definition of Source_Citation in GEDCOM does not agree with the actual definition of a “citation”, but that is a different discussion outside of the intent of v5.5.1 GEDCOM.

Tatewise said:

Splitters typically put the full details of the specific document in the SOURCE_DESCRIPTIVE_TITLE (FH Title box) or the SOURCE_PUBLICATION_FACTS (FH Publication Info box).
That combines the source "collection" details with the "Where Within Source" details.
Alternatively, in FH V7, they use Source Templates that define custom fields to hold "Where Within Source" details.

Yes they do, and when I move via GEDCOM from FH to another application the sourcing data is lost or not usable because the new application does not support FH view of data or source templates. This is an issue we are discussing with GEDCOM beyond v7.

The current Data Dictionary for GEDCOM is clearly for “Lumping”, be that right or wrong when used by the general public and application.

[AdrianBruce]

... the need to associate the original physical reality with your source. It gets messy - particularly if you want to refind the record either in another providers' "collection" or possibly even in FMP's collections should they subsequently recognise that they are creating monstrosities and re-organise them!

Totally agree. I remember reading someone saying that the mark of a good citation was that you should be able to go somewhere else and obtain the appropriate image or whatever. In particular, if I was citing something in FamilySearch's images, you should be able to go to the correct Record Office and get the right physical bit of paper / parchment / whatever. We might even have said "Even if they've recatalogued the item".

(The closest I've seen to recataloging is at Chester Record Office where some registers appear at first glance to be doubly catalogued - e.g. parish registers are always P ???/1 in the browseable hierarchy but the item in question appears to have a different accession reference - e.g. P 284/1/1/6? Or P 284/5063/6? )

[KFN]

although the various providers seem a touch inconsistent in their naming of what a "collection" is. Collection? Dataset? etc.

If by this you mean the sort of "meta-collections" that FMP are creating; they are a PITA!

I think "Yorkshire Marriages" is a good example - you have to be very careful to check: (1) Is this a record of marriage or of banns, (2) Is this the parish register or some bishop's transcript? Sometimes the "transcript" (in truth more an index page) will tell you, sometimes not (or even tell you wrongly - parish registers are not usually written up in a consistent hand month after month!)

So what would be my "source"? I don't think "Yorkshire Marriages" cuts it - you cannot begin to form a view of the general reliability of the source - and Banns are not evidence of a marriage. So with these meta-collections, I try and identify the sub collections.

Sometimes if you go to early pages of a set of images (and "Yorkshire Marriages" has a large number of sets of images - you can't say "Image nnn of Yorkshire Marriages"), you can discover the identity of a sub-collection which is meaningful "Deanery of Craven, Bishops Transcripts, Skipton 1600-1845" or "PE165/32 Scarborough St Mary Parish Marriages 1885-1889" - but I suspect that you have to be careful that a set of images does not cover more than one say parish, in which case the first images can be misleading.

You have to go to the images and work out what physically they are part of - which to me under lines the need to associate the original physical reality with your source. It gets messy - particularly if you want to refind the record either in another providers' "collection" or possibly even in FMP's collections should they subsequently recognise that they are creating monstrosities and re-organise them!

Yes, many “Collections” or as you call them “meta-collections” are named incorrectly or contain items not associated with the title of the collection, but this does not mean that they are not the Source of your data. For example: I could be reading a book about the work on the atomic bomb, and within the book I find a reference to a cousin of my father. That book is still the source of that fact, “Worked on the Manhattan Project”, even though the title of the book, the publisher, etc contain nothing relating to my relative. My grandmother’s sister was a painter, I found in a book about Christian Krohg that she studied under him. Again the book title has nothing to do with the fact that I’m sourcing. Name the collected source because it is where you found the information not because it has the correct title for the information found within it!

For the Yorkshire Marriages I would still name it as the source of a Banns, but indicate in the Notes that the collection also contains Banns when someone references for follow up the source!

RE: Re-organizing collections. Yes this is where getting our source material from online collectors is a PITA but sometime this is all we get. I’m in the habit of taking lots of notes about my online sources, retaining images of the documents and trying to leave myself some hope if the collector changes their storage index, or even goes out of business. Over 40 years of doing this I’ve lost a number of sources both online and in book form because the collector/library/individual/book no longer exists where it was at the time that I collected the fact.

[davidf]

I have been pondering ... (always a dodgy thing to do when not adequately caffeinated)

I'm a lumper and also someone who uses FH not to directly "create output" (CDs Books etc.) but as a repository of "facts" about people that I draw on when "word-processing Family History". I'm also on V6 - where I believe Sources and citations are a little clearer and less complex?

One of the major issues in the lumper/splitter debate seems to be with Census records where we have an extra entity about which we want to hold data - the household. Both splitters and lumpers have to make different compromises to enable the household to be recognised. If we did not need to, the "text from source" would not be cut-and pasted from FMP/Ancestry etc (with a quick check against the image of course) but would be just the specific schedule line that related to the individual - with column titles and possibly with header information. But we like our households.

Splitters put the household details on the "text from source" where the "source" is the collection of lines and header information that make up the household construct. Details common to say the 1871 Census (notes, date, repository etc.) will get repeated for each household in the census - or omitted.

Lumpers either put the entire household detail in the citation "text from source" for each person in the household of interest (and yes, have to change them all if they discover that they have made an error is transcribing - rare?) or they may only put the detail in the head of household's census fact and reference it either in the fact note of others - or set them as "witnesses" (which creates a foreign key link in dbms terms). I have been experimenting with putting the household transcription in a shared note, but the interface means you have to link the note through the All tab (which really is not ideal for day to day recording).

("Non-census facts" - e.g. occupation place of birth etc. I just reference to the generic source "1871 E&W Census" - my eye can scan down to the census fact and pull up the required detail)

Has anyone got thoughts about putting the household transcription in the Media Record Note ("Picture Note" - and uncheck "use for Caption"!)? Again linking the media record to the citation (lumpers) or the split-source (splitters), is just adding a foreign key and we already do. Two households on one page or a household that spans more than one page does not "over" complicate it.

For those of us using FH as a repository of facts this might be workable; I suspect though that getting direct output to include the "Picture Note" may be tricky - but is that a matter of configure once and forget?

[ColeValleyGirl]

I'm also on V6 - where I believe Sources and citations are a little clearer and less complex?

Not really -- V7 just introduces new ways of constructing and viewing them; the underlying stuctures are the same.

[davidf]

... For example: I could be reading a book about the work on the atomic bomb, and within the book I find a reference to a cousin of my father. That book is still the source of that fact, “Worked on the Manhattan Project”, even though the title of the book, the publisher, etc contain nothing relating to my relative. My grandmother’s sister was a painter, I found in a book about Christian Krohg that she studied under him. Again the book title has nothing to do with the fact that I’m sourcing. Name the collected source because it is where you found the information not because it has the correct title for the information found within it!

I agree that books (secondary source?) can be valuable sources for "co-incidental"
information about an individual. (Alan Hodgkin's Biography, Chance & Design, has a single line that references my late father's work on radar during WW2 - which tells me more than my father ever did!).

But I don't think you should compare them to meta-collections. The "fact" is that Hodgkin mentions in Chance & Design that my father solved a problem with gun-laying radar for Lancaster rear-gunners. His book is my source, albeit secondary and well after the event.

Meta-collections are almost like a subsection of an archive (or archives in the case of Yorkshire Marriages) or library where the index cards have been pulled together "to make it easier for researchers". If the National Archives book stacks were organised by "Meta-event" (say WW2), I would not reference my father's work by saying "TNA book stacks, 3rd aisle, 2nd shelf from the top"; I would break that down into the "created source" - in this case Hodgkin's book.

[NickWalker]

I've discussed this so many times over the years but really everyone is a splitter at least at some level.

A 'lumper' (known as method 2 in Ancestral Sources) makes a decision as to what they consider a source to be. For example, if someone kept a diary for 30 years they may decide that a source is 'John Smiths Diary' or they may decide it is 'John Smith's Diary 1923' and have another source for 'John Smith's Diary 1924', etc.. Or possibly if the entire Smith family kept a diary they may have a source called 'The Smith Family Diaries'. Unless they have a source called 'All the diaries in the entire world' then a lumper is still making a fairly arbitrary decision as to how much to lump and where to split.

In the same way a lumper may decide to have a source called 'England Wales Censuses' or they may decide to have several sources 'England and Wales Census 1841', and so on for each year. So again they've made a decision as to where to split.

As a 'splitter' (method 1), I've also made a decision as to where to split and decided that it will be at the household level for a census. For birth certificates I've decided that a birth certificate is a source, etc. This makes absolute sense to me and really doesn't have anything to do with the way Family Historian and GEDCOM works.

However, I am also a database designer and the way that GEDCOM/FH is implemented means that lumping (method 2) causes data duplication in circumstances where transcriptions of a source are required and I find myself unable to use a system that requires data duplication. I like to have transcriptions of birth certificates, census entries, baptism entries, etc. and so I have a source for each one of those. However, I don't have transcriptions of birth/marriage/death index sources and as these index references are only linked to one fact, I split those.

[KFN]

Davidf: Maybe my understanding of what you call a meta-collection is incorrect, I’ve never seen the term in our library. This could be nothing more that a “finding aid” an index of information for multiple books, or collections (collected content) that is separate from an individual book or collection! A “finding aid” works like a book’s index of topics, but is outside of the book and can index multiple books and collected content.

When I talk about a collection, it can be a collection of individual books, or a bound collection of individual documents, or a film/fiche/roll/image set/etc that has been linked together logically into one viewable set. Very often a library (be it private/corporate/public) to reduce storage of paper, to use as backup, to use as user facing information or because the data was “born digital”, create “collections” of computerized or scanned images. A finding aid is created so that researchers can find specific pieces of data within that collection quickly.

[AdrianBruce]

...
Yes, many “Collections” or as you call them “meta-collections” are named incorrectly or contain items not associated with the title of the collection, but this does not mean that they are not the Source of your data. ...

Yes. But also.... (Sorry, I couldn't work the phrase "Not Only But Also" into my reply.)

This speaks to the need to analyse "stuff" sometimes to a deep level, rather than take the "Collection" level name from the provider on trust. There are minor changes that data providers do without telling us - FMP's "Cheshire" PRs were originally named without "Cheshire" in their title - I think they were "Diocese of Chester..." something or other. However, people went into FMP, searched for "Cheshire", didn't find it and complained bitterly that it was a complete nonsense that FMP had published Cheshire PRs. Well, they had, they were just named in an unexpected - but correct - fashion. So FMP renamed the "collections" to include the word "Cheshire".

I attempt to get round this sort of issue by always describing the relevant image at the detailed level as well as providing the (then-correct) collection name. Hopefully, the detail will enable someone to find the thing even if the collection name has changed, or indeed, they're on a different provider or, as I said above, if they're in the Record Office, nowhere near a PC.

This is why it takes me an evening sometimes to sort out my first source-record and / or first citation for some stuff - never mind the advertised name, just what the heck am I looking at?

[fhtess65]

Indeed they are a PITA...I have learned while gather citation information from all the major database sites to note the original repository and check ITS catalogue to make sure I have the details correct. There are errors at FMP, Ancestry, FS etc...

And FMP is particularly bad about not including anywhere close to enough detail for a full citation - in many cases, no reference number etc. If you can get to the front of the roll, you might find it there, but sometimes it's not possible to do that either (half the film-strips give you image 32 of 56, allowing you to type in a 1 and get to the beginning, the other half doesn't and scrolling there is painful, and not always possible).

Teresa

although the various providers seem a touch inconsistent in their naming of what a "collection" is. Collection? Dataset? etc.

If by this you mean the sort of "meta-collections" that FMP are creating; they are a PITA!

I think "Yorkshire Marriages" is a good example - you have to be very careful to check: (1) Is this a record of marriage or of banns, (2) Is this the parish register or some bishop's transcript? Sometimes the "transcript" (in truth more an index page) will tell you, sometimes not (or even tell you wrongly - parish registers are not usually written up in a consistent hand month after month!)
<snip>

[KFN]

Nick Walker said:

However, I am also a database designer and the way that GEDCOM/FH is implemented means that lumping (method 2) causes data duplication in circumstances where transcriptions of a source are required and I find myself unable to use a system that requires data duplication.

Yes up to this point GEDCOM has not done a very good job of Normalizing it’s data structure. I have been a database designer since the early 80’s

Typically in a Library we would list all of the volumes of a person’s diary as a single item/collection (a source), but not include other people’s diaries in that collection. So a Lumper would then identify the volume in the citation.

[ColeValleyGirl]

In spite of much discussion about somewhat related topics, this has not progressed to the point where we have an agreed specification for a new item on the wish list, so I'm putting it on hold until somebody volunteers to progress it.