Before 1850, the census recorded only the name of the head of household, and then the number of people living there, categorized by age, etc. Beginning with the 1850 census, the names and ages of all persons living in the household were also recorded. Although their relationship could be implied from the names and ages, it could also be misleading. For example, a niece or nephew, or even a younger sibling, living in the house could be mistaken for a child of the head of household.
Beginning with the 1880 census, each person’s relationship to the head of household was also recorded. However, that does not mean that their precise relationship to each other can be definitely deduced, i.e. stepchildren and half-siblings. And they would not include children who were deceased or had already moved out or were living with other relatives or were live-in employees somewhere else (or were not yet born!). As a result, when census records provide the basis of genealogy data, they often contain the wife’s first name only (not her maiden name), estimated birth years, and incomplete lists of children.
I access the census data by subscription on Ancestry.com. Over the years, they have enhanced the way they index census pages, with regard to individual census districts. When I first started compiling this database, all of North Annville Township was lumped together as North Annville Township, the official census district, even though the census taker had annotated different sections of the census with the names of the residential areas, such as Belleview, Water Works, and Annville. Ancestry.com has subsequently recognized these unofficial divisions in their index. The result is that census references for old entries in my database may have the footnote, for example, “1910, North Annville Twp”, while newer entries may have “1910, Annville”, even though the two people were actually enumerated on the same page for Sheridan Avenue in Annville.
Birth dates, in general, are subject to conjecture, since the data is often contradictory. There are many instances where a person’s age is listed sequentially on the ten-year censuses, for example, as 22, 34, 41. The 1900 census (see the fragment reproduced above) is the only one that lists the month and year of birth, rather than just an age, but even those are not 100% reliable. For example, a person whose birthdate is listed on the 1900 census as Aug 1870, appears on the 1870 census, which was taken in June, as being 1 year old, and on the 1880 census as 11 years old. In this case, I would consider all the data and conclude that the person was actually born in Aug 1869.
Nevertheless, we must keep in mind that there is no intrinsic value in the precise date of birth of an ancestor. The value lies in the context of that birth – to differentiate that person from all others, to know who that person was.