[XeTeX] Devanagari Nukta problem

[Alastair M. Paterson]

Hi!

Some Devanagari is being typeset using XeTeX (MikTex 2.7).

When the combination of: consonant + short i + nukta (unicode: consonant 
+ 0947 + 0937) is used, the nukta is being displayed as the unicode 
dotted circle with the nukta beneath it rather than having the nukta 
appearing correctly beneath the consonant.

If this character combination is typed into Wordpad, for example, and 
using the same font, it is rendered correctly.

Any suggestions?

Thanks

Alastair