Conflicting Birthdates in Original Sources

Hi Elizabeth,

I've never been quite sure how to address conflicting birthdates in original sources that contain primary information, specifically, the great number of WW1 and WW2 draft cards that do not contain matching birthdates. I am aware of the reasons why someone may purposely change the date, but the numbers of men I have seems very large. I have many more that do not match than those that do. These dates are sometimes different then the various census records for that person.

I know that the further back in time we go, birthdates were less important and less known. But this presents a problem when I'd like to source a birthdate. I do always check the census to see which one is more likely to be accurate, and in some cases that helps, but when those dates vary widely I am simply at a loss. Do I source each WW1 and WW2 doc, with an "or" caveat? Do I cite a range of dates between those two documents and the census? 

And happy 4th of July!


Submitted byEEon Sun, 07/05/2015 - 12:56


The issue that puzzles you might be clearer if we divide this problem into two separate concerns. You speak of "sourcing a birthdate" and then wonder how to "cite a range of dates between ... two documents." 

  • Issue 1: Citing the source.
  • Issue 2: Reporting the birth date in our research notes or our narrative.

When we cite a census or a military draft enrollment card, the issue of what birth date to use would be irrelevant to the citation.  That issue of "what birth date to use" definitely causes a quandary when we prepare a narrative, because we have to make a judgment based on the evidence we have accumulated.  In order to make that judgment, we must, in the reporting stage (whether we use a database to store bits of information, a word-processor for actual research reports, or old-fashioned note cards) record each and every date, with a citation for each date to the document that alleges that date.

To build on your examples:

  • Doc. 1—WW I draft card, 1st draft: We transcribe the document, including the date, into our research report (or we extract the date from a photocopy and enter that one bit of data into the "birth date" field of our database. Then we cite the source.
  • Doc. 2—WW I draft card, 2nd draft: We do exactly the same. As a result we have two separate documents transcribed into our report (or two separate birth dates entered into our database) and two separate citations.
  • Doc. 3—1920 census: We do the same. Now we have three separate documents transcribed into our report (or three separate birth dates entered into our database) and three separate citations.

We continue this process for as many different records as we find.

Now, at this point, if you’re doing genealogical research, you are likely to say:

"Oh, yeah. I'm using a database that will take all my data and compile a narrative bio for this-or-that person. I don't want my narrative bio to say, 'John was born 1 July 1890. John was born 17 July 1891. John was born 17 January 1891. John was a sawmill operator at the time he registered for the draft. ...'  That's illogical writing. My narrative should report just one birth date or a range of dates.”

In this case, you’d be exactly right. But we would have a different problem here: reliance upon software to do our writing. So I’ll venture into that mine field for readers who have this thought.

The solution here might be this:

  1. Enter each date under a “birth” tag and document the date.
  2. Then create another birth tag in which you write, free-form, your own sentence that can serve as a topic sentence for a whole paragraph. A simple one might be

            John was born in Amory, Mississippi, in either 1890 or 1891.1

Ideally, you would  be using software that has a narrative field attached to this sentence, one in which you can proceed to write the rest of the paragraph and the software will then insert it into the narrative at the end of the above sentence (not into source notes). The result then would be

John was born in Amory, Mississippi, in either 1890 or 1891.1 Three different sources—all of them official records—give three separate dates. John’s registration for the first World War I draft, reports his age as …. {Yada, yada, yada, for the rest of the discussion about his birth date.}

The free-form topic sentence written under the birth tag would then carry the date “1890–1891” and would be marked as the preferred birth data for the software to print when it {shudder} “writes” the narrative bio.

Of course, given your academic background, you undoubtedly have spotted another problem here. The resulting paragraph about the disputed date for John’s birth puts its one and only reference note at the end of the topic sentence instead of at the end of the paragraph or (better) linking each piece of data individually to the source that supports it.  If we’re using this kind of database then we’re stuck with letting the tail wag the dog here until we convince software engineers to create a protocol that meets research standards, rather than expecting researchers to change standards to fit whatever tool they're willing to provide.

{And now, I'm off that soapbox.}



Submitted byyhoitinkon Sun, 07/05/2015 - 16:54

I document conflicting dates in a manner similar to what the editor suggests, but with one tweak: I make all the birth dates that document the various options mentioned in documents private, and only make the 'composite' birth date (the 'conclusion') public. That way, I have all the information at my disposal, but I can easily choose to hide it from the reports by selecting that I don't want to export private information. 

Submitted bymsualumnion Sun, 07/05/2015 - 22:34


I really appreciate your very detailed and thorough response and would expect nothing else. The way you divided the problem into "citation" vs "reporting" was insightful. I actually do not my genealogy software write my narratives--I prefer to craft them (painfully)myself, so the good thing is that I would not have to worry about the issue you spoke of. I do use the software to keep track of the sources, so I rather like your suggestions there as well. 

Another thing that was helpful was the idea that I could cite (I mean "report in the narrative"-smile) the various years and their citations even if I cannot support a particular day and month of birth.  Two brief examples follow.

I found the following relevant birth data for Tobias Prather-- the day and month are consistent, but the years vary:

(1) WWI draft, states "17 March 1882"
(2) WWII draft, states "17 March 1880"
(3) 1910-1930 censuses record ages consistent with a birth in "1881"
(4) 1900 census states "March 1881."

In the example above, "March" and "17 March"' are consistent but the years are not.

I found the following relevant birth data for one Tobias Prather:

(1) WWI draft, states "15 September 1882"
(2) WWII draft, states "June 1881"
(3) SSI index reports a birthdate of "9 June 1881" --not primary
(4) 1900 census reports a birthdate of "June 1882"
(5) 1910 and 1920 census report ages consistent with a birthdate of "1883"
(6) 1930 census reports an age consistent with a birthdate of "1882"
(7) 1940 census reports an age consistent with a birthdate of "1883"

In this example, June dominates as a birth month, with varying years.

One more add-on question: How much would you weight the secondary information in the census, a known suspect in crimes of inaccuracy, against the draft cards, in cases like these? I know that primary information/original sources do not always trump secondary information. Can consistency in ages in a set of census records ever "trump" primary information? (I'm using a cruide term "trump," but of course I am referring to the process of evidence evaluation and analysis.) 

I am still really perplexed about how widespread these birthdate mismatches on draft cards are, particularly the ones where the month and day are different. 

Thanks as always for sharing your knowledge.

Robyn of "Reclaiming Kin"





Submitted byEEon Mon, 07/06/2015 - 11:03

In reply to by msualumni

Robyn,  you wrote:

Another thing that was helpful was the idea that I could ... report in the narrative ... the various years and their citations even if I cannot support a particular day and month of birth.

Robyn, it's your narrative. You can write it any way you want. (So long as you document each assertion, of course. :) )

Submitted byrworthingtonon Mon, 07/06/2015 - 08:42


Might I suggest you add one more item into your evaluation of the evidence. That is, the Informant.

Based on your 7 examples, you have several that you might know who that is, several you do not know, and at least one, where you know who it is, but that are working from second hand information.

Just a thought.


Submitted byEEon Mon, 07/06/2015 - 11:18


You also asked

  • how common these errors are on the draft cards; and
  • whether the (presumably) primary data in draft cards trump (lower-case usage of the word <g>) census information given by an unknown informant.

Errors on those draft cards are far more common than most users expect. My own father, for example, is shown as "single" on one draft registration when he was very definitely married with children. There may have been some wishful thinking there (it was his first wife) but the card is also "signed" by a signature that is not his.

Among other issues here is that draft boards often supplied stacks of cards to larger employers in their jurisdiction, to have the cards filled out at the work site, rather than the men having to take off work during the day to personally appear before the draft board.

As for whether "this" kind of source should trump "that" kind of source, the answer is always It depends.  We can't make a blanket generalization. In every case we have to analyze all possible circumstances under which the record was created, study (for anomalies and patterns) the record set in which our  record-of-interest appears, try (as you're doing) to identify the informant, and then balance all the factors that we have identified.

Then we can use our narrative to explain our reasoning and conclusion.




Submitted bymsualumnion Sun, 07/12/2015 - 08:59

Thank you again to everyone for their responses. They have all been helpful. Wading into the murky waters of the past is surely not for the thin-skinned;)


Submitted byJan Murphyon Tue, 07/14/2015 - 11:33

Robyn -- one more thing to watch out for with these draft cards --

Check to see if the card is consistent with itself -- whether the birthdate on the card is consistent with the age reported on the card.

Very often, the answer is "NO".  I suspect that men may have told the registrar that their birthday was "July 4th" and that the year of birth was back-calculated from the age, which is likely to result in the incorrect birth year for a good portion of the year, depending on the date of the official document.

I have a similar situation where the WW2 draft card and my subject's obituary state that his birth year was 1892, but the baptism record I have from the parish register which seems most likely to be his is in 1890.  The census records, his Masonic membership card, the obituary, the draft cards, all agree he was born in July, and the sources which give a birth day all agree on the day, but the years are all over the place.





Submitted byJadeon Tue, 07/14/2015 - 13:17


Just supplementing Russ's #6 and Elizabeth's #7:

You said in your original post your concern about "original sources that contain primary information."

When we look at records that appear obviously to be made by our person of principal concern, it is also important to question how they knew what they asserted.  To be sure, every person was present at their own birth, but not a single one was able to look at a calendar and write a notation as to a date.

Therefore, while many of us are convinced that we were born on a particular date, we were told this or read this or otherwise collected this information from another source.  Thus it is not ~primary information~ but secondary or unknown, and in cases of draft registrations the target person's source is usually not known.  In some sorts of documentation the person states they have a parental Bible record or someone's certified copy from such a Bible, and sometimes even a baptismal record is torn out of a church record.

None of these are primary records either, unless there is very good reason to believe that an original was written by the mother of the target person or by a midwife at the time of the occurrence.

As in Elizabeth's example, I have seen quite a few instances where a person seems really not to have a fairly reliable date as to his or her birth, and may state different ones in different records.  I also have a marriage record where specific full year-month-day ages were given for both bride and groom, but both were wrong: neither one was as old as stated, and I have concluded that each one just made up the specifics.  They stood out because none of the other entries for this calendar year had more than an age in years . . . .

This is one of the pesky problems in developing accuracy in genealogy; we have always to question our own thought process in evaluating a piece of information.

Good hunting,

Submitted byJan Murphyon Wed, 07/15/2015 - 12:41

For anyone working on the WW2 draft cards, Judy G. Russell has posted an answer to a "how old did he have to be" query on her blog.

She includes a timeline of the seven different registrations with the age ranges required for each, plus links to places you can look up the Federal Statutes online.


Submitted bymsualumnion Fri, 07/24/2015 - 14:16

Jan and Jade, thank you for your additional suggestions--both have added additional ways of helping me analyze this. That's one reason why I love this forum. I'm going to check out Judy's post. All this has made me want to check and see if my own birthdate is accurate, :)