For the record - one femtometer is roughly half the width of a proton. Femto is a prefix meaning 10^-15. Therefore, a femtocell would have to cover one 10^-15th the width of a regular cell. GSM can reach (optimistically) 50 km from the base station, so we have (5m*10^4) * (10^-15) = 5m*10^-11 = The width of the inner electron cloud of a carbon atom. I can't imagine a cell zone of this region being useful, but I certainly give props to the engineers who can design something this small! Please call them, say, millicells or something more accurate - and stop peeving my pet.

But a cell occupies volume in three-dimensional space, so maybe a better calculation takes the volume of a femtocell as 1x10^-15 the volume of a full-size cell.

That would put the linear dimensions of the femtocell at 1x10^-5 that of a cell, so if a cell is 5x10^4 m then a femtocell is 5x10^-1 m - or half a metre. I think this is a bit smaller than what they have in mind, but it's not so wildly inaccurate...