I think it must be an overzealous/haywire word correction algorithm that doesn't believe single letter words (a, I) should stand on their own. Or maybe it was retyped by someone by hand and missing the space after those letters is the typist's idiosyncracy. I noticed typos which were similar to one another in different respondents' answers.
Is it so that you can duplicate the original source with less duplication penalty?
See: http://www.stifflog.com/2006/10/16/stiff-asks-great-programm...