America was the first country to start a census on a consistent basis. The goal was to ensure accurate representation, since the House has proportionate representatives based on state population size, and the Senate just two senators from each state. Thus, correct population counts by state were (and still are) of paramount importance in the democratic process, as well as to determine the country’s military and industrial capabilities, and for taxation.
And early censuses (1790-1840) were just that - a count of the number of individuals. For verification purposes, the census included the name of the head of the household, but no other persons were identified; there was just a count of the number of men and women, boys and girls. For genealogists, these censuses are a disappointment, since tracking of individuals is impossible. Even the tracking of families can be next-to-impossible if the surname is common! Then add to this the fact that the early census data did not include every state (we only started with those 13 colonies!), and it is even more limited. But those first colonies are where most everyone lived, so it is not as bad as it might look! Furthermore, censuses were taken in territories even before they became official states (see below). Surprisingly (to me), all the census data has been retained (and microfilmed) in its original format, little has been lost or damaged (one glaring exception is all of the data for 1890), and it is still quite legible in most cases.
There are problems inherent with this data, as with any data. It is important to remember that a census taker (enumerator) fills in the form, and he/she can mishear or misspell the information given. The enumerator wants to fill in every line, so he/she may also press for an answer, and not mark these as “unsure” answers. Some information is even filled in by the enumerator without asking (like race). Most of the time the census is “taken” 9-5 Mon-Fri, so usually the wife relates the information for the entire family; therefore, particularly in the case of borders, etc, the information may be quite questionable. For many years, the enumerator would take the original interview on one form and transcribe the data onto a final census form - an error-prone process. Finally, people without permanent homes, or moving from one to another, may be missed in both places, or counted in both places, a problem compounded by the fact that the enumeration used to be taken over a several month period. Genealogists say that census data is not a “primary” source, like birth and death records or passenger lists, since it takes place “after the fact,” based on recollections. But even primary sources have errors, and the census is a “primary source” in terms of detailing household composition on this one day—it is a relatively accurate “snapshot.” Finally, it is of interest to look at the whole neighborhood, since especially in the past or in small towns, most people knew their neighbors quite well; thus, this neighborhood information will yield insight into status, lifestyle and family history (e.g., is the house average-priced or cheap? Do relatives live close by?).
It is also important to keep in mind that in years past, the census was conducted in quite a different way. In 1880 and subsequently, the census was conducted over a one-month period, which minimized double-counting and missing families who moved. But in prior years, the census had taken up to 18 months -- resulting in some families being more easily missed on the census altogether. Also, the date of the census was at various times, which must be kept in mind when examining date (such as age); again, it is only starting in 1920 that the date was standardized as January 1st. The information on past years is as follows:
The census bureau hired local literate persons to personally travel home-to-home to interview the families; in the late 20th century, this was changed to a mail=form. Great pains were taken to interview the residents of every home; the census is used for political and economic reasons, so it was important to make sure that all constituents were represented. Even still, there are cases of people being missed, or perhaps pages being lost.
Some of the earlier census data was lost—such as Virginia in 1790 and 1800, apparently in a fire in the early 20th century (the same fire that destroyed the 1890 census). Surely, Virginia, home of many of our earliest presidents, must have been enumerated. And I assume Colorado too was lost in 1840-1860.
The census data is available for free for viewing at the National Archives (15 sites nationwide) and at the Morman Church Family History Centers. All is also available online for a nominal fee, with some years for free (e.g. 1880). The online versions have computer indexing which allow searching by first name only or name and age, etc, to make it easier to find individuals (such as the Widmann’s in Oregon in 1880), while the Archives and FHC generally are soundex indices - just for last name within a state. There have been special interim state censuses; the appendix details information on these.
State censuses were created by state governments and were usually taken in between the federal censuses, with similar information. However, many of the ones listed here actually predate the first national census in 1790, and were most likely used for determining taxation. Many of the original documents no longer exist. Some may not be microfilmed, and are available only at the national archives.
I looked at two states to find information on gaps that I had. The first was Missouri, where I was interested in finding out more about the Dickenhorst family in the 1850’s, specifically when the two girls died. However, even though Missouri records were supposedly available for 1852 and 1856, most of these special censuses no longer exist, including Lafayette and Saline counties. The second state I examined was Iowa, when I was searching for information on the two Widmann children that died in the 1870’s/1880’s. This was successful in that I did find Sioux City and Waterloo for two years I wanted, 1885 and 1895, but unsuccessful in that neither census included the two children I wanted (1880 Oregon did have Carrie, but that took me a LONG time to find!). I finally found church records for the two children in an earlier church in Waterloo. Finally, I was also interested in Waterloo for information on George Dankenbring’s first marriage and son.
I had expected that the special interim census data years would have limited information, and in fact they often have more than a normal census -- this is the only place where I have seen the question of religion. And the 1925 Iowa interim census is incredible for genealogists -- it is 6 pages per person, and includes the father's and mother's (maiden) names, place of birth, and age at last birthday! However, I would expect that many of these censuses were not so detailed, and may not even enumerate every individual in the home.
This is a summary of the availability of special census data for just the states and time periods in which we have interest, and the date (in parentheses) the state was admitted into the Union:
CA -- (1850) -- … 1935
CT -- (1788) -- 1636, 1756, 1762, 1774, 1798
DC -- (n/a) -- … 1867, 1878, 1885, 1888, 1897, 1905, 1906, 1907, 1908, 1909, 1912, 1913, 1915, 1919, plus police censuses for 1885-1919 (heads only)
DE -- (1787) -- 1776, 1798
IL -- (1818) -- … 1855, 1865
IN -- (1816) -- 1801, 1815, 1825, 1835, 1845, 1865, 1871, ...
IA -- (1846) -- … 1865, 1867, 1869, 1873, 1875, 1885, 1895, 1905, 1915, 1925
MD -- (1788) -- 1701, 1704, 1708, 1710, 1712, 1755, 1762, 1776, 1778
MO -- (1821) -- … 1848, 1852, 1856, 1860, 1864, 1868, 1876
NY -- (1788) -- 1698, 1703, 1712, 1714, 1723, 1731, 1737, 1746, 1749, 1756, 1771, 1795, 1798, 1801, 1804, … 1875, 1885, ...
NC -- (1789) -- 1784, 1785, 1786, 1787, 1798
OH -- (1803) -- 1790, 1801, 1802, 1807, 1811, 1815, 1819, 1823, 1827, 1831, 1835, 1839, 1843, 1847, 1851, ...
PA -- (1787) -- 1776, 1798 (and the PA state archives has "septennial censuses" which are tax lists with names of taxable inhabitants for 1779-1800.)
OR -- (1859) -- (n/a)
WA -- (1889) -- (n/a)