Posted by Paul Rawlins on April 10, 2010 in Collections

The U.S. censuses are by far the most popular documents we have online, so recently, has worked hard to improve the images, indexes, and search functionality for the U.S. censuses. The latest installment, the improved 1920 census, is now spiffed up and ready to go.

We now have new digital images, which means better image quality and readability.

Re-keying for a new index revealed an additional 250,000 names that have not appeared in our index before, and arbitrating differences between the original index and the new effort will mean better transcriptions, the addition of some 20 million alternative names, and a more accurate index.

Since I don’t have any research going on in 1920 right now, I decided to check out some of the new images. I confess, this wasn’t a rigorous, double-blind study complete with a control group and sophisticated regression analysis. In fact, my experiment was about as random and unscientific as you can get.

I had nothing to go on, no frustratingly fuzzy pages I had encountered in the past, so I just spent some time tooling around the site one afternoon looking for pages that could use a little help. It was a bit of a crapshoot, but I found a half dozen, sent my list to our scanning folks, and asked if I could get a peek at the new pages. And some of them do look promising.

This might be my best one. It’s from the Dallas 100th precinct, District 80, and the lower left-hand corner is just about illegible. Here’s a screen grab from the website.

Dallas District 80 1920 Census (old)

And here’s the upgraded scan:

Dallas District 80 1920 Census (new)

Hello, Joseph Johnstone!

Of course, not every page is going to yield a revelation. Some are just a little easier on the eyes, like these columns from Haskell, OK:

The old:

Haskell, OK, names (old)

And the new:

Haskell, OK, names (new)

And some will still keep your nose pressed pretty close to the monitor.


Haskell, OK, faded (old)

A little better?

Haskell, OK, faded (new)

OK, I think some of these enumerators must have written with a #4 pencil and a very light hand. The new images aren’t going to fill in every blank, but here’s hoping they fill in a few of yours. Let me know if you come across some good examples from your own searches.  And enjoy the improved 1920 census.


  1. Paul,

    Thank you for this – I do have one question…

    You said:
    Re-keying for a new index revealed an additional 250,000 names that have not appeared in our index before, and arbitrating differences between the original index and the new effort will mean better transcriptions and a more accurate index.

    And—this is where you come in—the new index incorporates about 20 million suggestions from users for alternate names and corrections. So, thank you.

    Am I to understand from this that if my grandfather, James Doe, was incorrectly indexed as John Doe on the original 1920 index and I had submitted an “alternative” name as James Doe and given the reason as transcription error that the new index will only show his entry as James Doe and there will be no mention of the incorrect John Doe on his 1920 entry?

    If this is indeed the case then Ancestry is really headed in the right direction.

    btw- which year census is the next one scheduled for this treatment?

  2. Monika

    #2 Oh, Andy! Now I see why you responded to me the way you did, yesterday! I know that you are NOT a man to assume! So, surely, just because the information you would give as an “alternative” name–coming from you–would WITHOUT QUESTION be accurate….surely you do not assume the same from any one else who would submit alternative data? Not everybody is a purist like you! The thought that others, less conscientious than you, could submit alternative data…only to see the original data erased on the index makes me shudder (in view of the fact that somebody gave as an alternate “fact” e.g., the birth date of one of my relatives such as to have him bcome a father at 16 instead of 40!! and that accepted that as an approriate “alternative”). And, yes, yes, Andy, I know one needs to look at the ORIGINAL record closely oneself (as you and I do). But, you KNOW how many people do not do that and would just integrate the new “alternative” into their tree…correct or not correct..without looking at the original record! This could turn into many incorrectly copied ancestry trees to be blindly copied by others.

  3. Monika, Slow down. 🙂

    Here is what I was told about the way the new 1920 census index was supposed to happen. When Ancestry incorporated the “alternatives” they were only going to incorporate those that were marked as transcription errors and that was only after a review of the document to verify that it was indeed a true transcription error of the original document. In those cases the original incorrect index entry was going to be replaced with a correct entry and the old, incorrect index deleted.

    Alternatives for maiden names, nicknames, name changes, etc. were going to be left as additions to, but not replacements for, the original index entry.

    IMHO, this is exactly the way it should be handled. The only other alternative is to not even offer the option of submitting corrections and alternatives and that is something I really don’t want to see.

    To me alternate data is the same as “See xxx” in a dictionary of book index. Something to be looked at but not always used.

    Until Ancestry disables the ability of users to Click and Copy from one tree to another we are always going to have incorrect trees blindly copied by others.

  4. Monika

    #4 – Andy, Oh, Goodie! 🙂 Thanks for straightening me out! You do this so well! 🙂
    🙂 No, I am fine with that! Do not mean to imply that I do not want to see appropriate corrections. As long as the “alternatives” are truly verified by staff, that’s GREAT. It will make it much easier to find people when you look for them!

    Love you Andy! 🙂

  5. Elle Litist

    Any improvement is welcome, so thanks Ancestry.

    A real improvement would be to flip to a negative of the image, and adjust the size and exposure by sliding increments.

    I hope the message boards are never improved to utilize emoticons.

  6. Cathy Lingenfelser

    You MUST be blind. Do you actually believe that these new images are BETTER than those they replace????

    There are far more unreadable images in your NEW batch. Did anyone take the time to compare the new view to the old?

    This is a fiasco and I am willing to bet that NO ONE is willing to take the responsibility for this mess.

    PLEASE bring back the originals….

  7. Amanda

    I’m not seeing the fiasco that someone else does? I think it’s more readable and not so blinding like the images with high contrast. I hope 1930 census is next in line for this. Thanks for the improvement.

  8. Anne

    The enhanced censuses were the single most important reason I converted my trial membership to a full membership.

    I don’t know how many of you tried to look up census information from microfilm in the “old days” (before Internet) – but even when you got it loaded, which required travel and lots of time, it was often illegible. censuses are a huge improvement and well worth the cost of membership.

  9. Amanda,

    I’m with you on the 1930 census being next.

    There is soooo much work to be done on that one (states of births not originally keyed for individual or parents, the enumeration districts thing, etc).

    They should actually have started that one first but…

  10. BobNY

    Okay, Andy, I’ll bite.

    What is the “enumeration districts thing” in the 1930 census that needs to be worked on?

  11. Bob,

    Here is a post describing it…

    Ancestry has Incomplete 1930 Census Citation
    jdweintraub (View posts) Posted: 4 Nov 2007 6:04PM GMT
    Classification: Query
    In 1930 it was decided that Enumeration Districts would have a two part number. The first number is the County number, assigned in order of the alphabet within each state. The second number is the district number. So the Hollywood Knickerbocker Hotel in the City of Los Angeles, CA, for 1930, is within ED number 19-64. That double number appears on the upper right of each of the 1930 Census sheets. Take a look at any schedules you have copied from the 1930 enumeration and you will see clearly the double number. In 1930, all EDs within Los Angeles County show the 19- prefix.

    Ancestry makes no mention of the prefix (e.g. 19-) anywhere when they reference 1930 EDs. They show a “Source Citation” for each 1930 index result, but again do not include the prefix in their “citation” or even indicate there that the material comes from NARA Series T626 (except in their text description). ***Both*** the prefix and district number make up the ED number for 1930, and any citation that omits the prefix number and the source of the material is incomplete.

    Ancestry may not know it, but this is going to create even more problems for Ancestry users (and perhaps Ancestry programmers) when the 1940 census comes out in 2012. Our research shows that in 1940, 100 of the largest cities were given their own prefixes, different from the County prefixes that they are in. Thus for 1940 Los Angeles census results, District “#1” could actually be ED #19-1 which is in L.A. County outside of the two large cities that follow, ED #59-1 which is within the city of Long Beach in L.A. County, or ED #60-1 which is within the city of Los Angeles in L.A. County. The Hollywood Knickerbocker Hotel will be within ED 60-132 in 1940.

    This lapse in proper reference citation should be corrected now for the 1930 material, across the board, and should not be repeated for the 1940 information when it becomes available.

  12. Linda Rood

    I would like to see Ancestry transcribe the State Census records starting with New York. They are invaluable, especially the 1865 census. Unbelievable amount of information at the end of each enumeration district.

  13. Linda,

    All you have to do is convince the New York State authorities that they should make those records available to Ancestry.

    Until the state authorities agree to that there is nothing Ancestry can do.

  14. Mike Hobart

    The new 1920 censuses scans do not currently use the NARA microfilm roll numbers, but rather some internal reference number. This is a major problem in documenting research and forces users to do extra work to determine the roll number. I am copying an example below:

    Year: 1920;Census Place: Attleboro Ward 1, Bristol, Massachusetts; Roll 31109_4301088; Page: 16A; Enumeration District: 3; Image: 688.

    No one not using would now what Roll 31109_4301088 would mean.

    This presumably can be readily fixed as Ancestry must have a NARA roll number new roll number table 🙂

    Mike Hobart

  15. Mike Hobart

    Following up on my comment on looking up the roll number, I took the E.D. for the above record and went to the NARA website link for the 1920 index of enumeration districts

    and found

    681. Berkshire Co. (EDs 24-50) and Bristol Co. (EDs 1-15, 226, 16-18, 22, and 216).

    So I would document the record as coming from NARA microfilm roll 681.

    But that should be what the source page shows in the images (as it does for the other censuses).

    Mike Hobart

  16. Paul Rawlins

    Sorry to take so long to get back.

    Ah, good point. Maybe what I should have said is that a truly scientific study would have been less dependent upon my whim.

    I’ll have to investigate your question a bit. I made a mistake in my initial post about the alternate names (a misunderstanding on my part and a missed email correcting me—my apologies), so let me do some asking so I get it right. That is, if you haven’t already discovered your answer.

    As for what’s next…great question. But not one I can talk about yet.

    The 1870 census was reimaged and reindexed last year as part of the work we’ve been doing on the census–though I’m not sure if that answers your question. Let me know if it doesn’t, and I will see what I can find out.

  17. Paul,

    Thanks, I really would appreciate knowing if my understanding of the re-indexing of reported transcription errors is correct.

    If so it might spur others to take time to make the effort to report them if they know such transcriptions will, at some point in the future, actually be corrected on the index instead of just merely shown as a alternative.

    In fact, if a schedule were put out of which one would be next then members could concentrate on getting those transcription errors handled first – which would be a win for both Ancestry and members.

  18. Mary Beth Marchant

    My question is about the index wherein the birth place is listed for example as “Georgia(the country”. I don’t remember if that applies to the 1920 census but is sure prevalent for the 1850/1860 census at least. I have made many corrections so I am wondering if those will be incorporated when other census decades are reindexed. I realize that many times the handwriting is extremely poor and illegible and for the surnames that is understandable. But having someone born in the US in 1850/1860 and stating that they were born in “Georgia(the country” just indicates a lack of knowledge by the indexer. I am hoping that will be fixed in the near future.

  19. Ida

    Paul, I am missing so many family members I was just hoping if they went back thru the records that they might find these people. They were in the 1860 and 1880. Most were not even related at that time. One other thing.

    While reading the other post about correcting transcription errors. For years I have been trying to get ancestry to look at the Kentucky Death Records for Jefferson County, Kentucky. Whoever transcribed them put the wrong first name. I found this over and over.

    With so many being wrong I really think this needs to be looked into. If you want an example let me know. I have many. You can also look at my tree to see what I mean. Thanks

  20. P J Evans

    I’m going to second Mike Hobart on the microfilm numbers – I want the NARA film numbers, so _other people_ can look at the original record and see what I did.

  21. Jim Livermore

    To Andy:

    “All you have to do is convince the New York State authorities that they should make those records available to Ancestry.

    Until the state authorities agree to that there is nothing Ancestry can do.”

    Bull. LDS has the microfilm. New York state has nothing to do with it.

    End of off topic post.

  22. Jim-

    As has been stated before…LDS has nothing to do with Ancestry.

    If LDS has the exclusive use of the microfilm then they have to agree to give Ancestry access it it.

    The point is that Ancestry must be given the right to use the records by whoever has control of the records; there is no unilateral action Ancestry can take on its own.

  23. Margaret

    # 25 Andy

    There is a relationship between the LDS and Ancestry. Read the source information for some of the databases including — on topic –the 1920 Fed Census. “Images reproduced by FamilySearch.”

    BTW, Ancestry does have a couple – though not very many – databases covering some NY State Census records.

  24. Margaret,

    There may be a relationship between Ancestry and the LDS concerning certain projects but each of those is a one-off deal, there is no overall relationship concerning the collections of each organization.

    The point of my post was…

    Just because the LDS has something in their collection it should not be assumed that Ancestry has access to it.

  25. Jim Livermore


    Where in my post did I suggest any relationship between LDS and Ancestry?

    The point of my post was…

    Quit pointing fingers at organizations you might think are withholding records from Ancestry. They are not. The ball is in Ancestry’s court entirely. It is all about the money.

  26. Jim,

    I know for a fact that certain organizations are indeed withholding records- and not just from Ancestry.

    The four largest counties in Tennessee have declined to make their early marriage index books available to anyone seeking to put them online.

    I wasn’t really pointing fingers, just saying that people seem to think Ancestry can just go get whatever records they want , when they want and that simply isn’t the case.

  27. Sherry

    Linda #15

    The familysearch pilot site has the 1865 and 1905 New York Census images online; unfornately not yet indexed but worth checking back to see what progress has been made.

    The 1892 New York includes an index as well as images.

    Best of all…… is free.

  28. Sarah L.

    Mike #17/18 and Margaret #26, it looks like Ancestry’s older 1920 census images were from NARA microfilm, but the new images are from FamilySearch microfilm instead. I think the strange new numbers that Mike found in the source citation point to the new source that Ancestry used.

    I looked at a transcribed FamilySearch 1920 census record for Attleboro Ward 1, Massachusetts. It included a “Digital GS number” of 4301088, which matches the last part of the roll number in Mike’s posting 17: Roll 31109_4301088.

    I’ve never heard of a digital GS number, but according to a posting on WikiAnswers, it’s an identifier for the digital version of a roll of microfilm created by the Genealogical Society of Utah (GSU):

    I checked several 1920 census entries, and the digital GS number from FamilySearch matched the last part of the Roll number on Ancestry’s pages every time. I don’t know what the first part (31109) is. It’s the same in every record I found.

    Technically, I guess the source from Ancestry’s pages would be what we cite since that’s what we used. But it’s complicated. It sounds like Roll 31109_4301088 shown on Ancestry’s page is the digital version of the FHL microfilm (1820681). If you look up that film, it refers to the NARA film that Mike listed in his post.

    Paul, if you or someone from Ancestry could help us sort this out, that would be great.

  29. P J Evans

    Many of the images from other years refer to _both_ the NARA film numbers and the FHL film numbers. If I’m going to refer to a source for a census, I want at least the NARA film number, and I’ll add the FHL film number if I have it. Using some other number, without any kind of refernce to what it is and where its from, is just Not Acceptable.

  30. BobNY

    Nor was it acceptable to do what ACOM did when they “enhanced” the 1900 census. They deleted all of the ED descriptions in major cities, e.g. New York and Chicago.

    I have been trying for an entire year to get them to:
    1. Admit that they were there before the “enhancement”
    2. Recognize they are not there now.
    3. Have an “expert” determine if they are important
    4. Look into returning them.

    That is the current situation as I have not heard anything back from ACOM for almost 8 months.

    These were the last 2 communications:

    Chris Lydiksen,
    Posted on:
    August 25, 2009 at 11:46 am
    #62 and #63 – I’ve forwarded your feedback to the expert on the case. We’ll get back to you again on this.

    Chris’ final response indicated to me that he still did not underdtand the question and basically has let it die a natural death.

    Chris Lydiksen,
    Posted on:
    August 25, 2009 at 3:50 pm
    #76 – If the problem is with the transcribed index, then the U.S. content product manager is the right person to address this. If the problem is with the search tool, then the search product manager is the right person to address this. As stated already today, the issue is currently under review.

Comments are closed.