Nadya --

Thanks for the further information.

As I'd mentioned, the solution I proposed came out more of an intellectual curiosity to do pattern matching based on position in VFP. What I wanted to do was a more general approach to pattern matching. But, the tools we have are still rather clumsy.

Sergey's code, which you had alluded to in your original post, is simpler for the specific case.

I suspect his code also would perform faster on short strings. On longer strings, at some point, the benefits of using CHRTRAN would outweigh the repeated searching of a string caused by AT in a loop. However, I doubt you'll run into that at 40 characters!

Best on getting that up and running. And Merry Christmas!

Jay