Deller is not the fella

LegacyUser

Malcolm Brian Wain said: It would be very useful when entering a surname in "Search Historical Records" if it were possible to filter out names known to be unconnected, rather than to tick the box for "exact spelling". I search for Tillyard, of which there are 10 know spelling variations, and I get an overwhelming number of entries for Deller or Dellar. In 50 years of FH research I have never once found any Tillyard spelt Deller. It is just so different and clearly wrong.

LegacyUser

Tom Huber said: Phonetically, the difference is not that great, but you might want to try playing around with the asterisk (*) as a wildcard, along with the exact checkbox.

Just a suggestion.

LegacyUser

Paul said: I agree with Tom. To be clear, you should input "Dell*r" with the Exact match box checked and you will get both your Dellar and Deller results and none for Tillyard.

Without using an exact name search, I can get hundreds or "Wright" results in my search for "Wrightson" relatives. Whilst you, and most users, would probably think that perfectly reasonable - compared to your example - in my 35 years of research I have only come across Wrightson being incorrectly recorded as Wright on about six occasions.

Now, if like me, you have received literally hundreds of "record hints" in Family Tree for "John Wright" individuals, on pages of persons named John Wrightson, you really would have something to complain about! The exercise of dismissing these as "not a match" (up to 25 instances against each individual) really can be exasperating.

As wildcard and exact match options are available, you must use them to eliminate these undesired results. Try cutting down on unwanted results in Find My Past and Ancestry - far more difficult than in using FamilySearch.

LegacyUser

Adrian Bruce said: While I sympathise with anyone who's got utterly weird (so-called) matches like Deller for Tillyard and Wright for Wrightson, I also sympathise with anyone who's tried to program for name variations across the whole global set of languages.

My note says that FS use a database of name variants - I don't know if language comes into it but even that's quite tricky. Are names in 20th century USA to be matched according to English language match rules? Or to the rules based on the language that gave rise to them? (Bearing in mind that a name may be composed of an "English" given name and, for instance, a Polish family name...)

As I recollect, D and T are very similar phonetically, so if any phonetic considerations went into the FS database, then that's probably where it all started going pear-shaped.

Regrettably, I think lots of us - myself included - have come to rely too much on phonetic etc matching and don't use wild-cards enough for when those phonetic etc matches just can't work. Then again - have you tried producing wild-card versions of Pickstock? Virtually every letter there can, and does, change, apart from the first - though why I've not seen "Bigstock" in the wild, I don't know.

LegacyUser

Adrian Bruce said: As an aside, my pet hate in FMP is that when you ask it to include name variants, it allows for blanks in the given name - thus searching for "William Jones" and variants, gives a match on "Wm Jones" (good) and on "Jones" (useless!)

LegacyUser

Tom Huber said: In your situation, search on Til* or a variant

LegacyUser

Tom Huber said: Variations would include Tilly* and so on

LegacyUser

Tom Huber said: And don’t forget to check the exact match box next to the name.

LegacyUser

Malcolm Brian Wain said: but there are 10+ spelling variations of the Tillyard surname sought. All begin with T and end with D but in between...... will try the wildcard trick which I was unaware of Thanks

LegacyUser

Malcolm Brian Wain said: Thanks for your advice. I will try that.

LegacyUser

Adrian Bruce said: You may find that you need to make 2 passes using wild cards - I know that on one name I was chasing in Devon, while I could describe all the variants using one wildcard pattern, that brought in too many false positives. I had to do two wildcard patterns to pick up all the variants but it greatly reduced my false positives.