Many death records list an exact age at death in years, months, and days. But including the number of months rather than just a total number of days into the year introduces a couple of ambiguities.
Example: John Smith died on 10 March 1850 and his death certificate reports his age as 25 years, 1 month, and 12 days.
If I subtract 25 years, and then 1 month, and then 12 days, I come up with the birth date
January 29, 1825.
The first ambiguity is that this assumes you start counting days with January 30 being number 1, January 31 being 2, etc. But the reporter might view the birth date, January 29 as day 1, so our inferred birth date is off by 1 day and should really have been
January 30, 1825
(It's essentially the same issue as how you count years when the age at death is reported as "in the year of his age".)
The other ambiguity is more substantial. If I take the death date 10 March 1850, subtract 25 years, then subtract 12 days, and then subtract 1 month (assuming birth date is numbered as the "zeroth" day) I come up with
January 26, 1825.
All I did is switch the order of subtraction of months and days and I produce a different birth date. Compounding these two ambiguities, we could be off by as much four days from what the reporter intended.
Is there a convention for reporting inferred birth dates in a range because of these ambiguities, like the 1750/51 convention for dealing with the ambiguity of old style and new style? Has anyone ever seen a death certificate that provides consistent instructions for computing age at death? Is there any evidence that people have been consistent?