The Feynman period is the contribution of a subdivergence-free graph to its anomalous distribution, or equivalently the simple pole of the amplitude in dimensional regularization. We used an algorithm recently developed by Borinsky to compute numerical approximations to the periods of 1.3 million distinct graphs in phi^4 theory, including all subdivergence-free graphs up to and including 13 loops. This large data set allows us to accurately determine distribution parameters, growth rates, and correlations between the period and numerous graph-theoretic properties. In my talk I will show some of these numerical and qualitative results. In particular, I will address the role of O(N)-symmetry factors and their interplay with the numerical value of the period. Talk based on https://arxiv.org/abs/2305.13506