You can read more about it here:
http://www.typophile.com/node/39060

Pieter

Great! Thanks Pieter. R.

While I generally agree with Adam's analysis in the other thread that serifs are not essential parts of letters, I think in the case of a design like this, when the I has strong bars, attachment to the end of the bar is legitimate. I don't really think of these bars as serifs; rather, this is a variant form of the letter I that has bars at the top and bottom, and the fact that this variant quite often occurs in the context of sans serif types strongly suggests to me that these bars be interpreted as something other than serifs. This variant also occurs frequently in handwriting, and despite the absence of serifs on any letters.