We have investigated compliance with the σ-product rule at triple junctions with a common  axis using computer simulations and in real thin films, using electron microscopy. Large discrepancies exist between the simulations, which assume randomly distributed misorientations, and the experiments. Although for both cases the rule fails at most triple junctions, the real material shows a much higher than expected occurrences of compliance. This result can be explained in two distinct ways: first, there might be a preference for triple junctions that obey the rule (implying a distinct influence of the triple junction itself). Alternatively, the misorientations in the real crystal might not be randomly distributed: if there is a preference for boundaries that have only small deviations from coincidence, there will be an increase in the compliance with the rule.