Electromagnetic field has a density of energy U = ε/2*E2+ μ/2* H2 And a density of momentum, given by the Poynting vector S = E x H For an element of volume dV you have a four vector of energy and momentum which is [E,P] = dV * [U, S] Being E the energy in the element of volume and P the momentum of inertia. If you measure this four vector from another inertial frame Fr' which moves with velocity v respect the initial frame, Fr, you can get [E',P'] by making a boost over [E,P]. Because of the contraction of lengths, the dimension in which the boost is made shortens in a factor 1/γ, γ = 1/(1-(v/c)2)1/2, so the density of energy and momentum must be multiplied by γ. So: [U', S'] = γ * Boost( [U, S], v ) Where Boost is the Lorentz Boost function. You can transform the fields E and H themselves to the new frame Fr' by using the transformations described here: https://en.wikipedia.org/wiki/Classical_electromagnetism_and_special_relativity I would expect that after transforming the fields I would get the same values for [U', S'] that the ones get by the previous way, however, nothing farther to reality than this: If I have a E = Eo x and I apply a boost in x direction, according to the transformations of fields I get the same fields: E' = E H' = H = 0 So I will have U' = U and S' = S whereas if I transform the 4-momentum vector, I will get: U' = γ2 * U S' = - γ2 * U * v Which of these results is the good one? Can anybody help me in finding what is wrong with this reasoning?