Many years ago, when I was taking a class on genealogy at the National Archives in Washington, D.C., the instructor assigned us the following exercise:

We were handed two pieces of paper, each a copy of a page from the U.S. Census for the Arizona Territory for 1880. The pages were not related, other than coming from the same census. Among the entries on each page was a man I will call William Smith (since I don't remember his actual name) and the members of his family. Our assignment was to determine whether two men named William Smith were enumerated in the Arizona Territory in 1880 or whether the same man was counted twice. To help us, we could draw on other written sources (this was the pre-Internet era), but we had to use the Census as the basis for whatever conclusion we reached.

Needless to say, the instructor deliberately picked a tricky case to throw at us. The listings for the two families did not agree in several respects. Worse yet, the two families were enumerated on different days in different parts of the territory. On the face of it, there was not much to suggest that the same man (or his wife) could have talked with a census taker twice within a matter of weeks.

That, however, turned out to be the explanation. And the only way to figure it out was to carefully compare the two listings, noting where they were consistent and where they were not, and then to ask whether it was physically possible for a family to move from the first location to the second in the amount of time in question. In other words, you could only build a credible argument working from internal to external evidence.

Unless you are exceedingly lucky (or exceedingly sloppy) in your research, it is almost impossible to trace your family tree in the United States today without using the U.S. Census as an investigative tool. But as my experience in the classroom illustrates, federal census records can be full of traps if you fail to appreciate the many quirks and complexities those records contain. If you happened to go looking for a William Smith in the Arizona Territory in 1880, you would encounter the same dilemma I did, except that I only had to deal with the problem hypothetically, whereas you would have to resolve it for real.

There are a number of general publications you can consult for assistance with census research. Two that come quickly to mind are The Researcher's Guide to American Genealogy by Val Greenwood and The Source: A Guidebook of American Genealogy by Arlene Eakle and Johni Cerny. But as helpful as such books can be, other books have appeared in recent years devoted entirely to census research--a welcome development for anyone wanting to go beyond the basics to dig deeper into the potentially rich layers of census data. Here are four such books you may want to consider:

1. The Census Book--A Genealogist's Guide to Federal Census Facts, Schedules, and Indexes by William Dollarhide. It really doesn't make much sense to jump into census records without first understanding what they are and how they were compiled. We are not talking here about abstract issues. Historically, census records were created the hard way: page by page, in longhand, by people traveling house to house as best they could, never knowing what would await them at the next gate or doorway. Consequently, your ability to extract useful information from censuses depends to a considerable degree on how good you are at remembering what the census taker was doing when he wrote down what you are reading. It was not exactly a scientific process.

In this book, William Dollarhide does a nice job of getting you oriented on the background of the Census and on how this history can impact your search. Wisely in my opinion, he moves quickly to make you aware of one of the most important--and most overlooked--facts about the Census: the census taker was not hired to record a household as it existed on the day of his visit but rather as it existed on "Census Day." If you don't understand this concept (and, therefore, don't know which day "Census Day" was in any census year), you are virtually certain to draw false conclusions from the data. This is a good example of where a little knowledge can go a long way, or where a lack of same can be disasterous.

Dollarhide's book includes a sizable section organized by census year (1790-1930) so that you can look up a particular year and determine what census records are available for that year, including what states and territories existed that year and which census indexes have been published for that year. He also provides charts showing whether the Family History Library in Salt Lake City has indexed (on microfilm) a particular county for a particular census year. Most such countywide indexes go back to 1850 or 1860 but not (with some exceptions) much earlier.

Dollarhide also provides a lot of information about non-population censuses, such as agricultural and slave schedules.

2. Finding Answers in U.S. Census Records by Loretto D. Szucs and Matthew Wright. In terms of general layout, this book is similar to Dollarhide's in that it begins with an historical overview of the Census and proceeds to examine in detail the records for each census year, followed by a discussion of non-population schedules. It differs, however, in that it includes numerous illustrations of actual census pages to give you some feel for the documents you will be working with. It also includes various tables of information, such as phonetic substitutes for letters in people's names, frequently misread letters in census entries, and a checklist of records you may want to examine when census records are illegible or missing.

Another feature of this book sure to be of interest to readers is the section on Internet sites where you will find census information. Although the descriptions of each site sound like the companies in question may have had a hand in writing them, the descriptions are still quite useful.

Both Dollarhide's book and this one provide reproducible worksheets for collecting census data. The sheets vary in format according to the questions asked in each census year.

3. Your Guide to the Federal Census for Genealogists, Researchers, and Family Historians by Kathleen W. Hinckley. If you want to pursue census research to even greater lengths than the above two books contemplate, you should consider Kathleen Hinckley's in-depth survey of census records and how to work with them. By necessity, Hinckley covers essentially the same ground as Dollarhide, Szucs, and Wright, but she does so in more analytical detail, with numerous asides to help reveal the amount of personal history buried in the enumerators' script.

For example, in her chapter on "Research Strategies," she cites the case of James and Margaret Brown, who appear in the 1900 Census for Denver, Colorado. James is not someone you have ever heard of, but Margaret most assuredly is. You know her as the steamship Titanic's "Unsinkable Molly Brown." Hinckley recreates the day the census taker visited Molly's neighborhood, showing the exact route the enumerator (a woman) took as she made her rounds of a section of downtown Denver. As we follow the enumerator door to door, we learn a great deal about the families in question, at first by studying the census page and then by drawing on other sources that Hinckley consulted to flesh out the story. As it happened, neither Molly Brown nor her husband was home when the census taker arrived, leaving Hinckley to speculate that Molly's mother, who lived with her, was the informant.

Hinckley's book offers many other valuable insights. She reminds readers not to repeat the common error of depending heavily--much less exclusively--on census indexes. To prove her point, she compares four different indexes of the first 50 names from the same record (the 1870 Census for the Colorado Territory). One of the indexes was prepared by a local genealogical society, one was prepared by a professional indexing company, one was prepared by a genealogical software developer, and one was prepared by a genealogical publishing company. By arranging the four indexes side by side in her book, Hinckley shows where they agree and where they disagree. Amazingly, of the 50 names in each list, only 12 names appear on all four lists. That is an agreement rate of only 24 percent. Obviously, indexing of handwritten documents is a very human undertaking. The lesson for the rest of us? Don't stop at indexes; go to the original source.

I was also gratified to see that Hinckley's book provides a short explanation of "enumeration districts" for census-taking. If you are not familiar with this term and how it applies to census research, you should check this book. William Dollarhide's book (above) also explains the term.

4. The American Census Handbook by Thomas J. Kemp. This book is completely different from the first three. Of the four, this is the biggest (over 500 pages), and its size is a tipoff to its content. Although the title might suggest otherwise, the book is devoted entirely to the subject of census indexing. If you want to know if someone somewhere has produced an index for a particular state, county, or town for a particular census year, this is the reference to consult.

The author has attempted to run down every obscure and not-so-obscure source for published census indexes. If an index is available on either of two Internet sites (ancestry.com or rootsweb.com), he tells you so. Most of the book is arranged by state, then by county, so you can quickly narrow down on the areas in which you are interested. There is also a much smaller section arranged by topic, such as African-American, Native-American, Military, etc.

Probably for reasons of space, Kemp does not give you any guidance on using the indexes in his book, so you might want to refer to Kathleen Hinckley's book before diving into this one. You will, however, find here a full citation for every listed index to facilitate locating the index in a library or buying it from a publisher.

Let's say, for example, that you are curious whether any index is available for Harlan County, Kentucky, for any census year. Kemp's book identifies ten such indexes: one for 1820, three for 1850, one for 1860, one for 1870, one for 1880, two for 1900, and one for 1910. None of them, according to Kemp, was accessible on the Internet at the time he compiled his book.

William Dollarhide's book (above) overlaps with Kemp's to the extent that both men list countywide census indexes. But Kemp's listing in this regard is much more detailed. Dollarhide is only working with indexes available on microfilm at the LDS Church's Family History Center.

To obtain any of the four books reviewed here, contact the publishers as follows:

1. The Census Book
Heritage Quest, $24.95
www.heritagequest.com
1-800-760-2455

2. Finding Answers in U.S. Census Records
Ancestry, $16.95
www.ancestry.com
1-800-262-3787

3. Your Guide to the Federal Census
Betterway Books, $21.99
www.familytreemagazine.com
1-800-448-0915