If I remember correctly G mentioned previously that they only tested one pair of forks per competing brand. If this is the case again then it is a bit weird, statistically, to round up all forks for which n=1 to within 0.1 K, and for the one brand for which n≥1 (I assume more than one Odyssey fork was tested) to round up to 650 K+. It's not wrong perse it's just a bit of a weird way to represent your data in my opinion. What is the number of replicates, average number of cycles and standard deviation for Odyssey forks?