Now that we are mostly at the mercy of computer indexing, we all need to learn new strategies for searching for information. Gone are the days when we could skim through an every name index and easily spot potential spelling variants. With a computer database, the only way to search for variants is to pick something like a soundex or phonetic option. That often means wading through several hundred potential variants and still coming up empty handed. That is bad enough. However, the problem today is compounded by shoddy indexing that is guaranteed to stymie our efforts.
I know, shoddy is strong language.
FamilySearch, my favorite place to search, is regrettably one of those databases. FamilySearch indexes are created by volunteers. Each entry has potentially been seen by three pairs of eyes. Two perform the indexing. When there is a discrepancy between the two indexers, an arbitrator settles the matter. Ideally, the arbitrator has research experience with older documents and handwriting. I have indexed thousands of records myself, so I know how difficult it is to do a "cold" reading of the record. It is one thing to be looking for a specific name and to be able to power through the handwriting and spelling; it is another thing to have no clue what the name should be and to attempt to decipher it.
However, all too often it seems like the indexers and arbitrators are not even trying. A case in point is the marriage record of Isaac Duncan and Susannah Kavanaugh of Madison County, Kentucky, which FamilySearch has indexed as Isaac Duncan and Susannah Havana. No search engine is going to link Kavanaugh and Havana in a phonetic way. I found this by searching solely for Isaac Duncan of Kentucky.
Fortunately, the image is available in FamilySearch, so it is possible to see what the indexer and arbitrator viewed when they made the decision to index Susannah's surname as Havana. Isaac and Susannah's marriage record is the last line on the page.
An examination of the handwriting around Susannah's name, reveals several features of the clerk's handwriting.
An examination of the H in Harris, shows a downward stroke on the right leg of the H that then moves upward, making a loop that moves up and over to the left, followed by a downward stroke that loops under and to the right. The K in Chas Kavanaugh's name has a downward stroke on the right that moves in the opposite direction of the H. The loop in the arm of the K moves under and then over, ending in a squatty leg. The initial letter in Susannah's surname more closely resembles the K in Kavanaugh than the H in Harris. At least one indexer should have seen a K instead of an H. Then the issue would have gone to arbitration, and then the arbitrator should have examined the handwriting on the page and arbitrated for the K. Indexers and arbitrator's are supposed to examine the rest of the handwriting on the page to help with deciphering handwriting. Clearly, that did not happen here. It's just frustrating as an indexer to have an arbitrator neglect to examine the rest of the page when arbitrating a record. When this happens, I always file feedback, but does any one ever act on feedback?
So is Susannah's surname Kavana? Not quite. A closer look at the page shows that her name was divided between two lines with "Kavana-" on the first line and "=ugh" on the second line: Kavanaugh. This was a fairly common way to divide a word in the late 18th century.
Should an indexer catch that? Maybe not.
Should an arbitrator catch that? Absolutely.
For the sake of argument, one could say that this happened because both indexers interpreted the name as Havana; therefore, it never went to arbitration. I find that hard to believe. But even worse, if that is the way it happened, it means that we have indexers so inexperienced with handwriting and old records that they are creating errors that in many cases make it virtually impossible to locate records in a database without a lot of creative maneuvering.
Is this really the best we can do?
