When my maternal grandfather was approaching 65 he assembled his paperwork in order to claim his pension. As he did so he was surprised to discover that after a lifetime of writing out Dec. 6, 1910, as his birth date that in fact his birth certificate said that he was born on Dec. 6, 1911, a full twelve months later. This was problematic as his health was in serious decline and he really needed to get that pension. For the historian it presents an interesting question: how do we explain the discrepancy between what a man believes to have been his date of birth and the date listed on his birth certificate.
Now, the first thing to do is to ask how my grandfather came to know that his date of birth was Dec. 6, 1910. The answer is: probably the same as most of us know our birth date. Our family members who can remember when we were born tell us that was the date. In short, it is based upon what we might call oral tradition. In the case of my grandfather there was the interesting situation that his eldest sister was still alive when the discrepancy was discovered, and she was old enough not only to remember but to have actually assisted in his birth. This was the early 20th-century in rural Canada after all: he was born at home, not in a hospital. Aunt Dorcas was quite certain that, yes, Grandpa had been born in 1910, not 1911.
So what we are really dealing with at this point is a discrepancy between oral tradition and written records. Now, there is a tendency I think for many historians to favour records over oral tradition, but in this case we run in to trouble if we do, and this because of another datum, namely that my grandfather's younger brother was born early in 1912. If my grandfather had been born in December, 1911, my great-grandmother would have given birth to him whilst several months pregnant with another child, a biological improbable scenario. Quite simply, if his brother was born early in 1912 then my grandfather could not have been born in December, 1911.
The balance of the data suggests that in this case the written record is probably mistaken and the oral tradition probably correct. Yet the historian, at least any worth the name, will not stop there. In fact, this is where historiography really gets going, because the question now to be asked is "Why did this error arise?" Recall that I said this was rural Canada in the early 20th-century, a land before cars and telecommunications. Recall that I also said that he was born at home. Recall that he was born in December, which a century ago would have been the dead of winter in the region where he was born. So his family would have had to physically go into the county clerk's office to get him registered, and this without car and in the dead of winter. Under these conditions it is quite probable that the birth was not officially registered until several months after he was born, quite likely in the New Year. This would have been quite standard practice a century ago. Thus I infer that what simply happened was that when they went to get the birth registered whomever filled out the paperwork at the clerk's office had gotten to used to writing "1911" for the year and simply made a typographical mistake. This seems by far the easiest way to account for the totality of the data.
The interesting thing to note is that whilst this is the easiest way to account for the data I actually do not have a source that tells me that this was the result of a typographical error made by someone who was used to writing 1911 rather than 1910. And that's okay, because history is not the enterprise of adjudicating which sources are true and which are in error, even though (as in this case) one might recognize demonstrable error in the sources. Rather, history is the enterprise of explaining why the extant data relevant to a certain event (in this case, the birth of my grandfather and the registration of that birth) takes the form that it does and only that form. Sometimes the best explanation might actually appear in our sources, and if so that is a happy evidentiary boost, but very often we have to infer it using our imagination. The trick is to develop an imagination disciplined by reason and informed by data.