>>Your integers have only 32 bits, numbered from 0 to 31. You overstepped.
>>
>>(and this is not the bittest question in the world - I've seen bitter...)
>
>I suppose that this function cannot be applied to BigInt type, which if I understand correctly, has 64 bits. No?

VFP doesn't do 64 bit integers. If such a number is encountered, it is automatically converted as IEEE float, with standard 15,5 significant digits.