* Citing the United States Census

[Gary_G]

In the FHUG post on Citing the English Census, there was a discussion that helped define the minimum set of parameters needed to adequately specify a particular UK census page. Has anyone seen a similar article for the United Sates Census?

I am aware that the Evidence Explained book tries to give some guidance, but it basically seems to have one copy absolutely every possible parameter describing the census page. This leads to a rather bloated citation. From what I've been able to determine from the tools by Steve Morse, copying everything is not actually required in providing a unique reference. The problem is that I'm not sufficiently familiar with the USA census structure to be able to determine the essential elements to record.

[ADC65]

For the Title, I use:

Code: Select all

US Census <Year> - <State>, <County>, <Town/City>, <Address>

For example:

Code: Select all

US Census 1950 - California, Contra Costa, Richmond, 4121 Huber

For the citation, I use:

Code: Select all

State: <XX> / ED: <00>-<000> / Sheet: <000> / Schd: <000>

For example:

Code: Select all

State: CA / ED: 69-067 / Sheet: 009 / Schd: 080

This is enough to ensure a unique ID, at least as far as I've been able to ascertain, and I've got a few hundred US Censuses in my tree.

I do not record Supervisor's District Number, as it is not needed to make the reference unique, but you may prefer to capture it. In some years the Enumeration District has just the last three digits, I just record these years as having two zeroes in the first position. The Sheet Number is normally of the form xxA or xxB, for example 05A, 14B, etc., but in 1950 it is just a number. I record everything with preceding zeroes just to keep things neat.

I'm sure there will be plenty of other advice, but this system gives a comprehensible Title (I don't use head of household names) and a unique citation. I have the citation 'blank' as above in the Ancestral Sources template so it is easy to overtype for each entry.

[Gary_G]

Adrian;

What do you do with administrative district, ward, tract, and block, in the big cities?
I seem to remember that these can be needed in the ED's of large cities, because the sheet/page numbers may not be unique at the ED level.

I notice that your reference doesn't show a county, yet I'm reasonably sure that the ED number is only unique within a given county. There must be a reason. Can you explain?

Wrt. "schd."; are you referring to the family number or order of enumeration number?

[ADC65]

Hi Gary,

The "Schd:" refers to the lowest level on the census, so for example on the 1940 Census, it refers to the 'Number of household in order of visitation', on the 1930 Census it refers to ''Number of family in order of visitation'. This is because this number is unique to the person or family group living at a location (disregarding errors introduced by the enumerator). So there may be multiple families per dwelling, but each will be given their own family number.

As far as I am aware, the ED is unique, certainly within State, but I will admit that it is not something I have studied in depth and I am open to correction. Looking through the few hundred censuses I have recorded, I do not see any duplicates across counties, but that would be an incredibly small sample of course. There are some online tools where you can track an address based on State / ED across different census years, and they don't seem to need county if you haven't got it.

I don't record the administrative district, ward, tract, or block in the citation. I consider these data I would capture in the Place / Address of the Census entry itself. Again, I don't set myself up as an expert, but as far as my experience goes, I have not met a situation where State / ED / Sheet / Schedule do not uniquely identify a family, and where State / County / City/Town / Address do not provide a unique Title. Your mileage may vary, as they say.

[Mark1834]

My system is very similar to Adrian’s. My custom source template has dedicated fields for country, year, place and address, and the detailed citation goes in the Publication Info field - National Archives reference for England & Wales; ED, Sheet and Dwelling Number for US.

[Gary_G]

Thanks, Adrian and Mark.

Trying to infer a rule from data from one district or region seems a bit risky, but what else can one do?
I wish more was written on exactly how the overall US census numbering system was set up.

I will likely write to Steve Morse, who has published several online tools for the US census, and ask if he has any background articles to share. If I find anything of interest, I will post it in this thread.

[ADC65]

Thanks Gary, I'd be glad to hear of any feedback. It's Steve's tools I was mainly referring to above, so if anyone knows, he does!

[Gary_G]

Adrian;

I just wrote to Steve Morse to make sure there isn't a better explanation, but I also read through the FAQ on Steve Morse' site and found the answer to some of my questions.

The FAQ is at: https://stevemorse.org/census/faq.htm. I've extracted some of the key points below, but read the full fact text, if you find my synopsis confusing.

The relevant FAQ numbers are:

107. What's the difference between this site and the 1930 NARA site?
209. I found the ED number. Now how do I find the person in the census rolls?.
209A. How can I use block numbers to simplify going frame-by-frame within an ED?
513. Why do some 1940 EDs have letters at the end?
514a. Why do 1930 and 1940 EDs consist of two numbers separated by a hyphen?
514b. Why does the number before the hyphen change between the 1930 and 1940 censuses for some locations?

In summary;

• The block numbers (when present). are not necessary, but help quickly narrow down a manual image search when browsing the census. In certain large cities, the EDs are defined by listing the blocks within the ED -- each such block is identified by the streets encountered when going around the block. Each such block is also given an identifying number or letter. If a 1940 Enumeration District involved more work than a single enumerator could handle, that district was subdivided and assigned to several enumerators. When that happened, letter suffixes were appended to the end of the ED. An example is 16-167A, 16-167B, and 16-167C.

• In 1930, the Census Bureau decided to assign two-part numbers to each ED. The first part of the number (prefix) denoted the county within the state, and the second number (suffix) denoted the census district within the county. To make things confusing, Ancestry.com does not show the full 1930 ED number, but just the suffix. Note that the NARA description drops the County prefix from the ED number although the format of a 1930 ED is the County prefix, a hyphen, and then a specific number. The One-Step descriptions for 1930 always shows the entire number.

• To find a person in a census roll; [This shows the key data to cite, but also see the point on the block number]
  You'll need to know the census microfilm roll number. Then go to the microfilm and go through it, frame-by-frame, looking for the ED number, and then within that ED looking for the street you are interested in. Then go down that street looking for the house number. And, finally, go through the listings of all people residing at that house number until you find the person or family you are looking for.

[Gary_G]

Adrian;

I just received a reply from Steve Morse.

You are correct in stating that state/county/ED are the items needed to uniquely identify a census record. But you need to add to that the page number and line number. With those five coordinates you can locate any entry in the census. The other items that you mention, such as administrative discricts, tracts, blocks, etc are simply noise when it comes to identifying an entry.

One more point. When I say page number, I do not mean the number displayed by the various genealogy sites such as ancestry. Rather I mean the number that physically appears on the census page.

[Gary_G]

While Steve Morse has given the items to cite for a US Federal Census, I found that the indexing of the New York State Census films potentially introduced a bit of a wrinkle. See, for example, https://www.nysl.nysed.gov/genealogy/nycqueensx.htm

It seems that, if one already has the film number, this is not an issue and one could cite them in the same manner as the US Federal census.

However; the online census resources typically don't provide any hint of the film number for a state census. To find it in the imaged index and then the original census films (which are not online and only available at the state archives), one often needs one or more of the Administration District (A.D.), Election District (E.D.), block etc. Perhaps this is the source of the confusion about needing to cite those elements. If one is just citing the source of what was found online, then the elements are not really relevant. However; if one is trying to fully cite the source-of-the-source and does not already know the film number, then those elements are needed somewhere in one's citation. As the source-of-the-source films are rather difficult to access, perhaps one can safely go with just following Steve's recommendation for all US census returns.

[ADC65]

Thanks for the update, Gary.

It's pleasing to know that Steve agrees with the uniqueness I have assumed in the fields used. I had not thought of using line number rather than (what I call) schedule number, and I might revisit that. The complication I envisage is you would need to use a range of line numbers to identify a group or family, e.g., "Lines 12-15" rather than the simpler "Schedule 42", but this might be more accurate than the sometimes suspiciously random schedule numbers.

I also agree to use the data on the scanned sheet, rather than page numbers that Ancestry et al supply. I think what Steve might be referring to is the Image Number in the counter at the bottom of the screen, which implies page number, but is not. However, Ancestry in particular are very sloppy about recording things like folio and page numbers in the UK census, and I would recommend always transcribing that sort of information from the scanned sheet itself rather than relying on some very iffy transcriptions.

I appreciate what you say about the microfilm and source-of-source stuff, but in general I do not worry myself with this sort of detail. If you were to find a particular image of a census on my website, and wanted to check it or see if there was a better image available, or whatever, you would be able to find it in just about any system armed with citation I supply. Adding which microfilm roll and which mountain vault it's stored in or any other data to the citation just clutters it up, in my opinion. Perhaps there's a place for that, maybe in a note, but for me it's not adding anything useful (there are sometimes exceptions, of course).

I think it's useful you asked the original question, and hopefully you have a basis to start with now. I did go through a number of iterations before I finally settled on this one, and as you know, re-working the same data a number of times becomes tedious very quickly.

[Gary_G]

Adrian;

I believe Steve was only trying to indicate a particular person by using the line number. If one wanted to indicate a family, then I would definitely use a family number (if available in that census year). That is what Evidence Explained suggests and it makes a lot of sense. Family number (or similar) is also the equivalent of a schedule number in the UK census, so using it may cause less confusion.

I also tend to work off the scanned sheet, but be careful. Your citation needs to show what you used in a way that one can find and review the record, if needed. If you used the website search-engine to find a record and the actual image shows something different than what was used in the search, a subsequent search still needs the "as-indexed" info to find the record on the site. If one is a stickler for details, the best way to show the data is to place the as-indexed data in quotes and follow it with the "corrected" data in square brackets. I have a number of records that have huge indexing/transcription errors in peoples names. One absolutely cannot find them again by searching for the spelling of the name one transcribes from the image.

The source-of-source is really only useful, if one can actually consult it and, even then, only if it could be different from what the version cited. I know that "purists" will argue up-and-down that it needs to be there. However; I generally include it if it is easily located and unambiguous. If the site only provides half the source-of-source info needed to locate the record, it really is rather useless to include it.