Recent theoretical efforts aimed at understanding the nature of antiferromagnetic ordering in GdNi2B2C predicted double-q ordering. Here we employ resonant elastic x-ray scattering to test this theory against the formerly proposed, single-q ordering
scenario. Our study reveals a satellite reflection associated with a mixed-order component propagation wave vector, viz., (q_a,2q_b,0) with q_b = q_a approx= 0.55 reciprocal lattice units, the presence of which is incompatible with single-q ordering but is expected from the double-q model. A (3q_a,0,0) wave vector (i.e., third-order) satellite is also observed, again in line with the double-q model. The temperature dependencies of these along with that of a first-order satellite are compared with calculations based on the double-q model and reasonable qualitative agreement is found. By examining the azimuthal dependence of first-order satellite scattering, we show the magnetic order to be, as predicted, elliptically polarized at base temperature and find the temperature dependence of the out of a-b plane moment component to be in fairly good agreement with calculation. Our results provide qualitative support for the double-q model and thus in turn corroborate the explanation for the magnetoelastic paradox offered by this model.
