Pity the United States Census Bureau. It's charged with describing all Americans even though they often have trouble identifying themselves.

White or Caucasian? Black, African-American or Negro? Hispanic, Latino or Spanish? And if yes, Mexican, Mexican-American, Chicano, Puerto Rican, Cuban or other?

Describing Americans was a simple exercise in 1790; everyone was classified as free and white, other free persons or a slave.

It wasn't until 1820 that someone noticed the population wasn't so easily categorized, and "free colored" and "foreigners" were offered as choices. A long parade of tags began to be applied, decade after decade, until the latest census form in 2010 offered 15 categories as well as "some other race."

The Latino-Hispanic-Spanish question emerged last month as the latest to come before the Census Bureau as it prepares for the 2020 count. Much of the confusion has to do with the awful fact that every census form since at least 1820 would get a failing grade in an anthropology class because they mixed the terms race, color, ethnicity, descent and origin -- usually treating them as one and the same.

Which is what may happen in the 2020 form, making "Latino" a race when it is not (but then, neither is "white"). Latino is an ethnicity, which means a group of people who share a language, culture and heritage.

One way to end this confusion has been to eliminate the question on race. But that would be to pretend that everyone in this country is treated the same, and that has not been true since the beginning when it was important to enumerate "other persons" (also known as slaves) separately because they only counted as three-fifths of a person under the original U.S. Constitution.

The Census Bureau does not take lightly these questions of identity. It uses each 10-year head count as an occasion to experiment with how to improve the next decade's form.

In 2010, 500,000 households received forms with the race question worded differently to see if it would improve the response. It did.

Whether to remove the term "Negro" in the 2020 questionnaire will be based on another 2010 test (it has been included because research in the late 1990s showed that many people still preferred to use it to identify themselves).

These experiments and research are part of an admirable effort to get a full count of the population, especially those segments that have in the past been undercounted, and as a result, underrepresented.