I was recently struck by the rendering of the end of the Mangalasutta at www.tipitaka.org.  This site applies Western punctuation, and the part in verse appears in quotation marks.  It is followed by the quotative particle 'ti', with the result that in Roman script the text ends in maṅgalamuttama’’nti, with similar renderings in other script.  Now, ignoring punctuation, this resolves into the three Pali words "maṅgalaṃ uttamaṃ ti".  Such a final 'ti' should not count as part of the metre of the verse.

This raises the question of how sandhi and Western punctuation interact in Indic languages.  What do other members know of such interactions?  One might think that there should be no sandhi where punctuation intervenes, but that's not true even in English (mine at least), for even clauses can take enclitics, eg. "That it's broken's not the point",  Prose Sanskrit is particularly prone to sandhi occurring across minor pauses, so interaction is even more likely to occur there *if* Western (= Hindi) punctuation is used.

The odd transfer of the 'ṃ' to after the punctuation probably occurs because 'ṃ' is ths closest Pali comes to having a word-final consonant.  Writing maṅgalamuttaman’’ti or similar woulf imply a full final consonant, but Pali does not need a virama in writing systems with conjuncts or consonant stacking.

Richard.