The dissertation argues for the following thesis: Velars to have a phonological representation where place specification may be accomodated, but this hosting site is left empty. In other words, I assume velars to be placeless in their phonological representation. This hosting site can have a number of formulations, depending on the particular phonological model. It can be conceived as a Place node, like in Feature Geometry, or an element tier as in Government Phonology, or a particular dependency relation as in Dependency Phonology. According to the thesis, velars share the presence of this hosting site in their representation with labial and coronal consonants (and with vowels, of course), while differing from labials and coronals in not having anything to occupy this hosting site. The thesis is supported by phenomena from a range of languages. The placelessness of velars goes against received assumptions where coronals are considered unmarked due to their absence of place specification. I will demonstrate that quite a number of the most frequently cited cases in support of the unmarked status of coronals do not seem to constitute firm evidence for coronal unmarkedness (and in fact for markedness in general). This suggests that if coronals are still to be considered unmarked, it will have to have a different reason.