scaling properties of dielectric constant

Dear all,

I'm wondering about an easy triangular lattice of holes-calculation. I 
simulated that lattice with radius/lattice constant=0.45 once with 
epsilon=11.4 (like e.g. in the book of Joannopoulos) and once with 
epsilon=2.45. There appeared a bandgap in tm-polarization between band 2 and 
3 with the bigger epsilon, but none with the smaller value. But shouldn't it 
look the same (due to the scaling properties of the master equations)?
Or is it only a resolution/mesh-size-problem?

Thanks for your help
Philipp

Above is my ctl-file:

(set! geometry-lattice (make lattice (size 1 1 no-size)
                         (basis1 (/ (sqrt 3) 2) 0.5)
                         (basis2 (/ (sqrt 3) 2) -0.5)))

(set! default-material (make dielectric (epsilon 2.45)))

(set! geometry
       (list
        (make cylinder
          (material (make dielectric (epsilon 1)))
          (center 0 0 0) (radius 0.45) (height infinity))))


(set! k-points (list (vector3 0 0 0)
                     (vector3 0 0.5 0)
                     (vector3 (/ -3) (/ 3) 0)
                     (vector3 0 0 0)))
(set! k-points (interpolate 128 k-points))

(set! resolution 64)
(set! num-bands 12)
(set! mesh-size 10)

(run-tm)