Averaging kinetic rates looks weird for identical rates due to numerical issues

[alongd]

Searching for a reaction with only two or more library sources that report identical rates may result in a weird-looking averaged rate:

At a first glance it looks as if the average is far off of the real rates, and is T-dependent, while the original values aren't. Of course, this is due to a numerical issue, as can be inferred from the y-axis values.