Wednesday, November 7, 2007

Be Careful with Indexes at

The key feature that makes and other commercial web sites so valuable are the Indexes. Without them, we would be reading handwriting on page images just like we did with microfilm images in years past. The Indexes on genealogy web sites have many excellent features - wild cards, many search fields, exact or Soundex searches, etc. They significantly raise the odds of finding the information that we are searching for.

However, sometimes there are flaws in the indexes, or in the data indexed. I pointed out flaws in the California Death Indexes here some time ago. Now I have found another.

Ancestry has a database called "Indiana Marriage Collection, 1800-1941." The Ancestry source description of this database says:

" Indiana Marriage Collection, 1800-1941 [database on-line]. Provo, UT, USA: The Generations Network, Inc., 2005. Original data:

* Works Progress Administration, comp. Index to Marriage Records Indiana: Indiana Works Progress Administration, 1938-1940.

* Jordan Dodd, Liahona Research, comp. Electronic transcription of marriage records held by the individual counties in Indiana. Many of these records are on microfilm at the Family History Library in Salt Lake City, Utah."

One of our CVGS members was looking for a marriage of Benjamin Reynolds to Cerilda Flinn in Indiana in the 1860's. If he goes directly to this database (which is a logical thing to do), he gets the following results:

* If he searches for the name as "Benjamin" and "Reynolds" - he gets 7 matches, but none to Cerilda Flinn.
* If he searches for the name as "Benj" and "Reyn" - he gets 9 matches, but none to Cerilda Flinn.
* If he searches for the name as "Ben*" and "Rey*" - he gets the 10 matches, but none to Cerilda Flinn.
* If he searches for "B" and "Rey*" - he gets 82 matches, but none to Cerilda Flinn.
* If he searches for "Cerilda" and "Flinn" - he gets 1 match for Cerilda J. Flinn married to Benjamin J. Reynolds on 22 Sep 1867 in Crawford County IN. The same match comes up if he uses wild cards for Cerilda.
* If he leaves the names blank and searches for the Spouse name of "Benj*" and "Reyn*" - he gets 10 matches including the one who married Cerilda Flinn.

But if he didn't know Cerilda's given name, or surname, he would have missed out on this information. Before this search, he thought the name was Sirelda from another record. The Soundex search does not find the record with a given name of "Sirelda" because it works only on the surname. [As a side note, there are 31 "Cerilda" entries in this index! I've never heard the name before.]

I found this record just by happenstance. I put "Ben*" and "Reyn*" in the search box on the Ancestry main Search page. That gave me a long list of matches in different databases. When I clicked on the "Indiana Marriage Collection, 1800-1941" link, I found the record quickly. The search found the spouse's name in this search. It is evident that the search box results for this specific database only finds the name in the "Name" column, not the "Spouse" column. The main Ancestry Search box finds both. I don't know if this holds for all databases or just this one. I tried more spouse's names in the database search box, and while many resulted in matches, some did not - I'm guessing 5% to 10% did not show up as a match.

The lessons here are:

* The Ancestry main search box may provide more matches than the specific database search box.
* Not all names of persons in a specific database are found by a search using "Given Name" and "Last Name" searches.
* In this specific database, some names of spouses are not included in the "Name" database - they apparently were not indexed.
* Searches in specific databases should include not only the search in the given name and last name boxes but should consider putting the target name in the "Spouse's Name" fields if that is available.

Is this's fault? Maybe not, if they took the database from other sources (see above). Perhaps the WPA list did not include the name(s) missing from the "Name" column.

We are so spoiled by the availability of these databases with excellent indexes and search capabilities. If I don't find someone after I use all of my "tricks" to find them in an index, I often will assume that they aren't there. On census records, I have resorted to using spouse names when known, but I haven't used them on vital records indexes and other databases previously. I will now!

No comments: