Image from the Library of Congress. Taken from Statistical Atlas of the Ninth Census (1870). For larger version and other images, click here.

“Did you ever have the measles, and if so how many?” - The Census, by Artemus Ward, favorite author of Abraham Lincoln

The United States Census is a humongous undertaking—a once-a-decade attempt at a comprehensive national headcount. Already constantly collecting information, the US Census Bureau is primarily tasked (by Article 1, Section 2 of the Constitution) with conducting the decennial Census. Only 21 decennial censuses have been held, starting in 1790: That event was spearheaded by Thomas Jefferson, then Secretary of State.

Over its 220-year history, the Census has spurred endless debate. After all, it is America's attempt to rigorously define who “America” is, and a quick look at some Census milestones does reveal a changing country.

1790. The first US Census cost 1.1 cents per capita to conduct, and found the US population to be 3,929,214 people. In its entirety, it read as follows:

1. Names of heads of families?
2. Free white males of 16 years and upwards, including heads of families?
3. Free white males under 16 years?
4. Free white females, including heads of families?
5. All other free persons?
6. Slaves?

The final question becomes loaded when you consider that, to determine its number of Congressmen, a state’s total number of slaves was multiplied by 3/5—a slave being, at the time, legally considered 60% of a person.

1844. Secretary of State John C. Calhoun argued that, according to the 1840 Census, the North had a very high incidence of “insane negroes,” while the South had virtually none—“proof” that America's black population belonged in slavery. The Census figures in question were false, having been heavily manipulated. Some of the figures were so preposterous that Southern statistician and anti-abolitionist James DeBow remarked that “the number of insane negroes was marked, we believe, higher in one State than the whole amount of negroes registered in it.” DeBow was later appointed head of the US Census Bureau (1853-1857).

1880. For the first time, women were allowed to be Census-takers. To see images of the Census from this era, click here.

That same year, C. W. Seaton, Chief Clerk at the Census Bureau, discovered the Alabama paradox, in which apportionment can decrease the number of a state’s congresspersons even if the state has no fewer people. In 1880, Alabama held eight seats in the 299-seat House of Representatives. When the House increased to 300 seats, Alabama kept only seven seats.

1890. The Census data became so vast (taking the better part of a decade to complete) that, for the first time, it was tabulated by machine. Invented by Herman Hollerith, the punch-card-reading device lowered tabulating time to two-and-a-half years. Hollerith started his own business, the Tabulating Machine Company, in 1896. It later became IBM.

1940. Just prior to the US’s entry into World War II, the Census Bureau began “sampling”: asking one out of every 20 surveyed households further questions on in-depth topics (e.g. fertility). By law, all such private information gathered during any Census must be held from public release for 72 years. Emphasis on must: even the CIA isn’t privy to it. Of all Censuses awaiting full public release, this one is the oldest. It will become available in 2012.

1970. California topped Texas as the nation’s most populous state.

2000. On the threshold of the third millennium, controversy arose over matters of definition: i.e., who counts and who doesn’t. For example, gay and lesbian couples were counted in an oblique way: those filling out the Census's long form were asked about their genders and about their relationships to household members (the Bureau then tried to put two and two together). Problems ensued.

For statistical purposes, this data could never be used to calculate or even to ballpark the number of homosexuals in the U.S., because only couples were counted and, at that, only those living together. And, if a same-sex couple checked the “Married” box (rather than the “Unmarried partner” box), the Census Bureau’s software read it as an error, automatically registering them as either a heterosexual couple or—if the age difference between them was 15+ years—as parent and child.

Other controversies that year:

• Many Utah Mormons, living in foreign countries as religious missionaries, went uncounted.
• “Hot box imputation” is a method used by the Bureau to get at the unknown number of residents in a building. It involves using the number of residents in the nearest similar building to estimate, and usually adds a fraction of a percent to a state’s population. In 2000, Utah’s fraction was a shade under what the state needed to earn another member of Congress. The state took it to the Supreme Court, in Utah v. Evans (2002), but ultimately lost.

The 2000 decennial census cost about $23.09 per capita. Since it found 281,421,906 people in the U.S., its price tag ended up being $6.5 billion.