Rules
Deaths by year by age: how to read the age field
Deaths by year by age: how to read the age field in a memorial record, the errors that cluster around it, and how to turn an age into a birth-year window.
An age is the least interesting number in a memorial record and the most useful one. Be clear about which of those you are relying on.
It is the least interesting because it says almost nothing about a person. It is the most useful because it is a search key: an age plus a year of death gives you a birth year to look in, and a birth year opens the records that actually contain a life.
This page is about one field in a memorial record rather than about any single death: the age a death notice or an obituary carries, and how far that number can be trusted.
What to take away
- An age in a record was reported, calculated or estimated, and only the calculated kind can be checked. The three look identical on the page.
- Treat an age as a birth-year window rather than a birth year, and widen the window when the number is round or was reported rather than worked out.
- An age-sorted view of an archive describes who gets written about. It describes nothing about how long people live.
Where the number comes from
An age in a record is rarely measured. It is one of three things:
How an age reaches a record
Reported
- Source
- Informant memory
- Checkable
- No
- Error risk
- High
- Common in
- Notices
Calculated
- Source
- Two dates
- Checkable
- Yes
- Error risk
- Inherited
- Common in
- Modern records
Estimated
- Source
- Nobody knew
- Checkable
- No
- Error risk
- Looks factual
- Common in
- Older records
- Reported.Someone who knew the person supplied it. This is how most ages reach a death notice, and how many reached historical registers, where an informant (often a relative in shock, sometimes a neighbor), answered from memory.
- Calculated.Someone subtracted a birth date from a death date. Reliable if both dates are right, and inheriting every error in either.
- Estimated.Nobody knew, so a plausible number was entered. Older records do this more than modern ones, and the estimate looks identical to a fact on the page.
Only the second is checkable, and only if the record shows its working. When a source gives an age and a birth date that do not agree, you have learned something valuable: at least one of the two is wrong. The disagreement tells you the entry was assembled rather than copied from a single document.
The errors that cluster around ages
Rounding. Reported ages drift toward round numbers, particularly for the very old. An age of 90 in a record is weaker evidence than an age of 87.
Error patterns around ages
- 90weaker evidence than 87
- 1 yearmost common discrepancy
- 2 weekscalendar shift possible
- 1 sourceno matter how many copies
Boundary arithmetic. Someone who died shortly before a birthday is a year younger than the simple year-minus-year subtraction suggests. This produces the single most common one-year discrepancy in any index. It is why a difference of one year between two sources usually means the sources used different arithmetic, not that either was wrong about the person.
Shading. People adjusted their stated ages during life, for work, for marriage, for immigration, for sport. The age recorded at death may descend from a number that was never accurate.
Calendar changes and date order. Older records may use a calendar or a date convention that is not the one you are reading them with. That can move a birth date by days or by a fortnight, and occasionally by a year.
Inherited errors. An age that came from a published notice, which came from a family's recollection, is one source, no matter how many places now carry it.
Using an age as a search key
This is the practical value of the field, and it is worth doing methodically.
- Take the age and the death date, and derive a birth-year window rather than a birth year: the full range consistent with that age, which is normally two calendar years.
- Widen it by a year in each direction if the age was reported rather than calculated. Widen further if the number ends in a zero or five.
- Search birth or baptismal registers across the whole window, not the single best year.
- When you find a candidate, confirm it with something that is not the age: a place, a parent's name, a sibling, an address.
- Record the window you searched and what you found, including the years you eliminated. That negative result is worth as much as the positive one to whoever looks next.
What an age-sorted index of a year can and cannot tell you
Sorting a year of memorial records by age produces a shape, and it is tempting to read the shape as a fact about mortality. It is not. It is a fact about coverage.
A memorial archive contains the deaths that were written about. Whose deaths get written about depends on public prominence, on which professions leave records, on which countries and languages the archive draws from, and on how long a person had been out of public life before they died.
Every one of those filters correlates with age. An archive that skews toward people who became known in their working lives will show almost nothing of the two ages where most human deaths actually cluster.
For statistics on how long people live, the sources are the National Center for Health Statistics, which publishes US life tables and mortality data built from complete death registration, and the state civil registration offices that register each death. Neither of them draws on selected notices.
Those national returns are collected internationally in the WHO Mortality Database. An index like this one is not a sample of anything and should not be read as one.
An age-sorted view is genuinely good for narrower things, and they are still worthwhile. It finds people who were contemporaries, it shows a generation thinning, and it gives a researcher a birth-year range to work with when nothing else about a person is known.
Before publishing an age
- Say whether the figure was reported or calculated, if you know.
- Keep the birth date beside it rather than replacing it; the date carries more information.
- Where sources disagree by a year, give both and note the likely arithmetic cause instead of picking one.
- Leave it blank rather than estimating. A blank is honest; an estimate becomes a fact the moment someone copies it.
The registers that a birth-year window is searched against are the ordinary family history sources — census schedules, state vital records, church registers — and the National Archives resources for genealogists map out what exists and who holds it in the United States.
Where this sits
The way the same field behaves in the first hours after a death is different, and covered in the age line in a fresh notice. The parent index is deaths by year, and the monthly unit beneath it is deaths by month. For why an age is one of the least verified fields on day one, see deaths today.
Common questions
Two sources give ages one year apart. Which is right?
Usually neither is wrong. One has subtracted years and the other has counted birthdays, and somebody who died shortly before a birthday sits between the two answers.
Should I record an age when I only have an estimate?
No. Leave it blank and record what you do have. An estimate becomes a fact the moment somebody copies it.
Why do so many old records show ages ending in zero?
Because reported ages drift toward round numbers, particularly for the very old. A round age is weaker evidence than an odd one.
Can I use an age to tell two people of the same name apart?
Only as a first filter. Confirm with something that is not the age: a place, a parent, a sibling, an address.
Does an age-sorted index show anything about mortality?
No. It shows the age profile of people who were written about, which is a fact about coverage.
