Ever since its inception, a popular DNA motif called the cross tile has been recognized to self-assemble into addressable 2D templates consisting of periodic square cavities. Although this may be conceptually correct, in reality certain types of cross tiles can only form planar lattices if adjacent tiles are designed to bind in a corrugated manner, in the absence of which they roll up to form 3D nanotube structures. Here we present a theoretical study on why uncorrugated cross tiles self-assemble into counterintuitive 3D nanotube structures and not planar 2D lattices. Coarse-grained normal mode analysis of single and multiple cross tiles within the elastic network model was carried out to expound the vibration modes of the systems. While both single and multiple cross tile simulations produce results conducive to tube formations, the dominant modes of a unit of four cross tiles (one square cavity), termed a quadruplet, fully reflect the symmetries of the actual nanotubes found in experiments and firmly endorse circularization of an array of cross tiles.