Silicon in [Cl--SiH(3)--Cl](-) is hypervalent, whereas carbon in [Cl--CH(3)--Cl](-) is not. We have recently shown how this can be understood in terms of the ball-in-a-box model, according to which silicon fits perfectly into the box that is constituted by the five substituents, whereas carbon is too small and, in a sense, "drops to the bottom" of the box. But how does carbon acquire hypervalency in the isostructural and isoelectronic noble gas (Ng)/methyl cation complexes [Ng--CH(3)--Ng](+) (Ng=He and Ne), which feature a delocalized D(3h)-symmetric structure with two equivalent C--Ng bonds? From Ng=Ar onwards, the [Ng--CH(3)--Ng](+) complex again acquires a propensity to localize one of its axial C--Ng bonds and to largely break the other one, and this propensity increases in the order Ng=Ar<Kr<Xe<Rn. The behavior of the helium and neon complexes violates the ball-in-a-box principle. Why does this happen? The purpose of this study is to answer these questions and to understand why carbon can become truly hypervalent under certain conditions. To this end, we have carefully analyzed the structure and bonding in NgCH(3)Ng(+) and, for comparison, CH(3)Ng(+), NgHNg(+), and NgH(+). It appears that, at variance with [Cl--CH(3)--Cl](-), the carbon atom in [Ng--CH(3)--Ng](+) can no longer be considered as a ball in a box of the five substituents.