Arch bridge springs can be connected to concrete abutments either by prestressing bars or by connectors. In both options, the torsional stiffness is substantially reduced, compared to the full arch cross sectional area. The influence of this lack of torsional stiffness on arch buckling is being researched, both numerically and experimentally. To reduce any residual stress during tests, wooden rods that simulate the arch were submerged in water and subsequently bent to the desired shape. Imperfections of the arch samples are measured. Two unequal concentrated loads are applied to the samples, thus simulating the effect of movable loads across half of the arch span. During loading, lateral deflections were measured until elastic buckling occurred. The simulation of more flexible rotation of the springs required replacing the cross section by thin equivalent side plates. Since all parameters have not been isolated, the results are limited yet. However, comparing the failure load of similar conditions, the reduction of torsion stiffness by 81.48% reduces the failure load by 26.3%. This indicates that total prevention of axial rotation may not be imperative for arch bridges.