Note its a logarithmic graph. If the noise floor is 1E-5 ohms, it would look smooth above 1E-4 ohms, and then would show as randomness once the signal declines below 1E-5 ohms.Right after 110 K, the R(ohm) data points jump randomly between 1E-5 and 1E-7.
I very much doubt that this is due to the accuracy of the instrumentation (rather than the actual change in data points), for which the change should be very gradual as the accuracy reduces down the scale.
That is to say, the randomness should increase slowly from 110 K toward the lower temperature if this is anything to do with the limitation of the instrumentation in measuring the smaller values, rather than the sudden jump to the datapoint randomness right after 110 K shown in the graphs, which is something interesting deserving to be looked into further possibly with better equipment.
So in that case, we cannot say anything about the 1E-7 vs 1E-5 ohms being meaningful.