Jim,
Sorry to question your experiment, but I truly believe your finding is questionable. Being a specialist in structural mechanics and stability of structures in particular, I can tell you this:
* Forces in a set-up like yours are distributed according to effective stiffness E*I . Determination of E*I (Young's modulus times area moment of inertia) doesn't require an experment. Thus, I think one rod was subjected to a higher load than the other.
* Straight line loading is paramount. I am not convinced your setup ensures completely axial loading due to the eccentric placement of specimens and the apparatus itself. Testing one by one in a professional testing machine would give more confidence in the result.
* Boundary conditions play a significant role in achieved stability load. In this case, deformation of the aluminum ends and shear load due to friction may have increased the collapse load of one vs. the other. You really need to mimic the simply supported end conditions of the genuine pushrods by providing a ball and cup type of condition at both ends, thus eliminating adhesion, end deformation and misalignment.
I would be most interested in your findings, but I also think simple calculations will provide a reliable answer. I may provide a calculation shortly.
Kind regards,
Knut Sonsteby