I have been entering place names into genealogical records from smallest to largest location (i.e. town/locality, county/parish/district, state/province, country). Where one part is not known, I have used the double comma (i.e. town/locality, , state/province, country). I have used the notes section for additional information, such as farm name. Some programs and online sites do not include the additional comma. Other programs and sites record more information, such as farm name, street name within the place. How should place names be recorded for best accuracy and reporting?
As others have pointed out, there is a no specific standard in this area. While most products and people acknowledge the use of a hierarchical description, there is no agreement, or any standards, on what the hierarchical levels should represent. The administrative divisions within one country will differ to another. Even the US has its complications such as being composed of 50 states PLUS one federal district.
In fact, the visible representation of the hierarchy itself will be locale-dependent since not every country will expect it to be arranged small-to-large and using a comma separator. In principle, there should be a difference between how the hierarchy is represented in the data (i.e. a linked list of entities) and how they're depicted on the screen or in a report, but this is going to be dependent on the software product you're using.
As Adrian points out, the terms place, location, and address are not clearly defined. Although there are people who will insist they are distinct (myself included), there is no "controlled vocabulary" in our field.
The loss of a pair of commas is unfortunate. I'm sure you've thought about inserting a place-marker such as "UNKNOWN" into the hierarchy. This would stop your software losing an unknown entry but it may not transport well because someone else might use a different marker. I would personally avoid a specific word since it would only have meaning in a particular language, and might even be ambiguous in a different language. Although I haven't tried it personally, have you tried using a simple question mark ("?") as a place-marker?
GEDCOM has the concept of both place and address - and so, one hopes, would software based on GEDCOM. While there is lack of total agreement over the difference btw the two concepts, the GEDCOM Standard 5.5 refers to place as a jurisdiction, i.e. something "higher" than a single geographic feature such as a church or farm, making a place something such as a town / locality, etc. It follows that the farm / street / church / hospital, etc. should go into the address if your software allows, otherwise into the notes for the event in question. The GEDCOM Standard 5.5 (p.50) agrees with your idea of the double comma where the intermediate level is unknown or irrelevant. (Examples of irrelevance include the City and County of San Francisco, where no-one would write SF twice).
Beyond this not unreasonable description of what the majority probably think (no sources for this!) you're on your own. I would, however, suggest that if you look at Gazetteers, etc., it would seem logical that your places match those because then there is the future possibility of mapping between your places and those sites / maps, etc.
Think also what reports you might want to extract. Selection will be easier if you have consistent place names of a consistent level - e.g. don't enter some places as towns and other places as churches within the towns. E.g. having 2 places "St Mary's, Nantwich, Cheshire, England" and "Nantwich, Cheshire, England" might cause you issues when you want to extract reports for Nantwich. Better to use "Nantwich, Cheshire, England" as the place throughout and push the church name into the address.
But - be consistent in your own usage for starters.
As there is no agreed standard in this area, I have adopted the following principles, which are (I hope) based on common-sense and so most likely to be understood by somebody reading my work.
As the Address, I record the information (as much as is available) that uniquely and accurately identifies a single building or farm or other property -- so, typically farm or house or building name, and/or number and street name (if these exist).
[I strongly advise not using building names like "Parish Church" or "The Hospital" as the buildings identified thus can change with time].
For the Place, I order from smallest to largest division (because this is the order used by people when describing a place in English). What divisions make sense depends very much on geography, but I mostly work with UK data, so typically use: parish/ward, town/municipality,county, country, omitting any elements that are irrelevant. As AdrianB38 advises, this allows me to keep consistent place names at a consistent level in the naming hierarchy.
If there's an element of a place that's unknown (rather than irrelevant), I use ? to indicate the uncertainty. If an address is relevant but unknown, I also use ?.
Common sense isn't always enough, our ancestors not having been considerate enough to avoid tinkering with administrative units -- it sometimes has to be supplemented with copious notes, and/or some pragmatism.
As mentioned elsewhere I always use the contemporaneous place name, and make notes about present-day equivalents (and Latitude/Longitude). This does mean that I may have the same place with two names if county names or boundaries change.
There can also be judgement calls to make when dealing with places on county boundaries. Cilrhedyn is on the boundary of Pembrokeshire and Carmarthenshire. The parish church is in Pembrokeshire, but Pont Wedwst (where my ancestors lived) is in Carmarthenshire. So I should properly record their marriage as taking place at
and their residence immediately after as
but I'll confess I always use 'Cilrhedyn, Carmarthenshire' as most of the parish is in Carmarthenshire.
As you have found, there is nothing even approaching a standard way to record place names.
It all depends on what you are inputing in to and outputing to. One must be familiar with both ends of the operation in order to know what is the best way to report it.
GEDCOM 5.5.1 defines PLACE_NAME as:
PLACE_HEIRARCHY is defined as:
So GEDCOM lets you (or should I say the program you are using) define your own place structure, but does not recommend that you do so.
You need to enter your place names in the manner that your program has implemented them. That way, it will (or at least should be) exported to GEDCOM so that other programs should be able to read them again.
Your program will either have form-based entry for your jurisdictions, or is free form.
Either way, I would recommend you create a few test places in different configurations with your program, and then export the GEDCOM file. By inspecting the GEDCOM file and looking for the FORM tag in the header (which may or may not be there) and your test PLAC tags, you should be able to determine how another program would interpret your places.
As an added check, you can then try importing your GEDCOM into a few other programs and see which of your test places are being interpreting the way you want them to. But remember that these may sometimes show up badly, not because they are not exported correctly, but because the program importing is not doing it properly.
In summary, if you really want to ensure that you are entering your places in a logical manner, then spend one night and test it out. You'll find that some programs and online sites will not read certain configurations, and you can set up your places names to minimize those problems for you.
As others have pointed out there is no definitive set of rules. I have always used the comma separators in my data. I teach courses on family history research and on the use of several programs and have beta testing Family Search and several genealogy programs. I have authored the on-line book "Genealogists' Guide to North America" at www.genealogy.thepenry.net. I am hoping that the GEDCOM rewrite will include the ability to enter two place names for each fact, historical and current location. I have a particular pet peeve; I hate to enter United States for events occurring before 1776 and dislike entering United Kingdom. I prefer historical locations such as Republic of Vermont, and entering Mexico for U.S. Southwest areas before states' creations and using British Colonial America for states before 1776. I accept 1776 as the cutoff, not ratification. Of course West Virginia/Virginia, Tennesee/North Carolina, Kentucky/Virginia are other examples. How about New Amsterdam, or when Sweden had possessions in what is now the U.S? Florida has been both Spanish and British, etc. The problem with entering places that are historically correct is that Geo Mapping doesn't recognize these locations. Common sense says to enter the current location, and enter historical information as notes, or other fields.