Despite what TMAW writes, I actually think the absence of photos of Frank in the series posted by deadesq is important evidence.
Frank is missing from LOTS of photo locations. Sure, perhaps the individual photographer misses a runner here and there, but the runner missed at each location would be at random, though with greater likelihood of missing someone as the pack became more dense.
Let's say that the probability of being missed at any location is 5%. Frank is missing from LOTS of locations. Let's work through his absence from 3 locations in the same race. The probability for this is 5% x 5% x 5% or 0.0125%. 1 in 10,000.
Based on the races I have run, the 5% probability of being missed seems an overestimate.
good work deadesq.