I've just been provided with a dataset of 83,000 records and it contains a weird and wonderful selection of observer name formats. I'm astonished at how well the import wizard is coping with making sense of them, but the following are tripping it up:

De Winter, M
O'Dwyer, J
Oakley-Martin, D
Orr-Ewing, D C

It splits each of these on the comma into two separate observers, resulting in M, J, D and D C appearing in the grid of names to be matched.

Now that I've found the problem with these I can hopefully pre-process the data to reformat these into something that it can handle, but just in case someone fancies diving into what must be some very clever code at some point I thought I'd mention them. (Then again it may be better to leave the code alone in case the magic is broken).

Keith