The following article was written by my good friend, William Dollarhide. Enjoy…
Dollarhide’s Genealogy Rule No. 29: The page on the census where your ancestor’s town was enumerated has no page number.
Name indexes to the U.S. Federal Censuses have become a godsend to genealogists. Beginning in the late 1960s, several private publishers began indexing the censuses. The pioneer publisher was Accelerated Indexing Systems (AIS). Over a period of some twenty-five years, this Utah company indexed all censuses for all states, 1790-1860. AIS first published the census indexes in book form, and later microfiche. Most of them have since been converted to online databases. All but a few of the AIS databases have been redone, first by companies like Heritage Quest, and later by FamilySearch.org, Ancestry.com, and others. The early indexes were often full of mistakes, and genealogists had to devise techniques to work around them.
Indexing Mistakes
Genealogists have learned the hard way that census indexes are often incomplete, or have names misspelled, names missing, or other problems. For example, if you are researching for people named Henry James or James Henry, you will learn to always check for both the surname and given name when looking at census indexes. Names such as these are prone to be reversed. As an example, and in response to our earlier article, “Census Mistakes,” the following was received from a reader:
“After reading your column online, I thought I would write and relate a mistake I found researching my family. I was looking for my great-grandfather and his family in the 1920 census for Washington, DC. I was able to locate his brother, but not my great-grandfather. Having learned where my great-grandfather lived about 1910, I was able to locate them in the 1910 census. But I could not find them in the 1920 census index, although according to the City Directory they did in fact live in DC. With the address from the City Directory and a map of enumeration districts, I finally located my great-grandparents in the 1920 census. It seems that the enumerator had taken my great-grandfather’s name “Meyer Perry” and reversed it to “Perry Meyer.” Thus, his wife “Rebecca Perry” became “Rebecca Meyer,” and so on with their four children (In addition, my great aunt’s name was re-named in the census). Looking up the soundex for “Meyer,” I went back to the index and immediately found the family listed under this name.”
This reader’s example can be repeated by many genealogists who have discovered that an ancestor is not found in a census index, yet the family is clearly in the census schedules. Note that his research involved the use of city directories for the same area where his ancestor lived — which is a good way to confirm whether or not someone resided in an area at about the same time a census was taken. In fact, city directories have proven to be a more complete and more accurate listings of the residents of an area than most census lists. (The best collection of city directories for a location are found at the local public library of that city. See our article, “Old City Directories” for more details). Our reader used another resource to help him find his lost ancestor — a map of the census enumeration districts for 1920. These maps were microfilmed by the National Archives and are available at the Family History Library and other libraries around the country.
The recently opened 1940 federal census schedules, for example, have some valuable finding aids, including maps, ED converters, and lists of city directories, all found at a special website setup by the National Archives just for the 1940 census. See
ww.archives.gov/research/census/1940/general-info.html#questions.
Spelling Errors in Census Indexes
There are two questions here: what percentage of entries are actually in the index somewhere, and what percentage are entered in spellings reasonable enough for the user to find? Missing persons have been reported to be as high as 20% of all entries in some of the early AIS census indexes, but if one looks at the names of missing persons by using alternate spellings, the error rate may drop to 5% or less. Therefore, a genealogist needs to think of alternate ways a surname can be spelled.
William Thorndale, a professional genealogist who has studied the accuracy of the AIS census indexes in detail, suggests that the accuracy rate is higher than most people think. Thorndale’s excellent essay, “Census Indexes and Spelling Variants,” in The Source: A Guidebook of American Genealogy uses examples of spelling variants that account for most of the so-called missing names. Thorndale said, “A Salt Lake professional once complained to me that the 1800 North Carolina census index was so bad that it missed all of the known Fespermans known to be in the state and listed in the census. It did, but this professional might have thought of looking under Fisperman.” He goes on to say, “If you are looking for Fesperman, simple prudence says you will look under F/Ph/Pf + vowels + s/ss/z/st + p… and if the index puts it under Fisperman, so what?” Let’s see, that means that possible alternates for Fesperman might be Phesperman, Pfesperman, Fistperman, Fasperman, Fosperman, and more.
Thorndale summarized five categories of spelling errors in census indexes as the following:
1. Calligraphic look-alikes: Daniel/David, Nathan/Mathew, Ball/Bell/Boll/Bull, Sanderdale/Lauderdale
2. Phonetic equivalents: Lydecker/Litaker, Myatt/Maillotte, de la Hunte/Dillahunty, Hansel/Ansel, St. Cyr/Sincere, Ratton/Wroughton, Vanlandingham/Flannagen.
3. Translation equivalents: Calbfeisch/Veal, Rubsamen/Turnip seed, Silver/Silber.
4. Truncates: Fitzgerald/Gerald/Jurrell, O’Sullivan/Sully, Haythornthwaite/Haythorn, Strohmaier/Maier, de Villeponteaux/Pontoux.
5. Spelling irregularities: Cowper pronounced Cooper, Coke pronounced Cook, Featherstonhaughs pronounced (so it is said) Fanshaw.
If your ancestor’s name is not in the census index, check the spelling. Think of any way a name can be misspelled. With a name like Dollarhide, I have found 36 different spellings of the name, because of the interchangeability of vowels, double letters, dropped letters, and changing the “d” to a “t”. (Dollarhide/Dollahide, Dalerhide/Dolerhite). I once found the name indexed as “Dotterhide” and decided that the indexer was reading a scratch on the microfilm copy that crossed the two l’s making them look like two crossed t’s. Opening a printed index, I start with Dal, and go through the rest of the vowels (Del, Dil, Dol, Dul). Generally, it will be the interchangeability of the first few letters of a name that will cause the name to be indexed in different places in the alphabetized list.
Using alternate spellings with the early AIS indexes, one can usually get the error rate down to 5% or less. With the Heritage Quest and later indexes, the error rate drops to as low as 1%, due to this company’s extraordinary efforts in checking and rechecking the names for accuracy during the compilation of a master name index. The indexing by FamilySearch.org is state-of-the-art, where there are typically three people involved in every entry, two working independently, and one who rechecks any entries that disagree. And, we have the actual images to look at now, not just the index, and the ability to browse through census locations has made the census more accessible and more accurate that ever before.
Dollarhide’s genealogy rule No. 51: An Irish blessing: May all yer fatters be in the index.
I found in the 1940 census that my individuals were not listed in the index though they had lived at a particular address all their lives from the time they married. I then went and looked at the actual census form, found their district, found their street, checked the house numbers, and lo and behold, the census taker skipped their house. The address went from 476 to 478, 479 etc. They lived at 477, so this was not the fault of the census index person, but of the census taker who never recorded their information.
I found where the census taker (and the person he lived with) put my husband’s grandfather’s middle name as his last name. His name was Grayson Wray Palmore, and he is listed on the 1940 census at Grayson P. Wray. I found him only by remembering what address was on his marriage license less than a year later.
I have learned that it pays to check the films of the actual records. My great grandfather and his family does not show on any index for the 1900 census. We have an extremely unusual surname and, at first, I thought that might be the problem. But from what I have been able to determine, the 1900 census for Moody County, South Dakota, at least the majority of it, was not indexed. The area I was interested in was Flandreau, the county seat. It was also the location of the Indian School (Lakota). Missing! However, the actual census records are clearly there.
Thanks, Laurann. If enough people like you were to respond, it might be possible for someone to make a list of all of the unreported missing census pages. We don’t always know who to blame – because missing families may have been the fault of the census takers going door-to-door (badly), rather than missed by the indexers years later. It took me hours and hours of searching the 1930 Pierce County, Washington original census schedules, page by page, to finally learn that the entire town of Mossyrock, Washington was not listed anywhere. I can’t blame Ancestry or FamilySearch.org for something that happened in 1930.