Dear Karen

The real concern of your japanese colleague seems to be the relationship
between the results of the measurements and the reality. There are various
parameters  (Temperature, Humidity, potential difference )influencing the
SIR. However it is difficult to separate these influences. So it might be
best to test at various Temperature-Humidity- Voltage levels and set up a
Statistical Model. I don't think there is enough space here to explain
that. But Montgomery explained this type of modelling very good in his book
- Design and Analysis of Experiments-. Looking at the humidity the subject
is no easy so I would make a linear attempt since it is as wrong or right
as any other one but easy to handle and the same for the voltage applied.
Looking at the temperature the relationship can be derived by the Arrhenius
law.This law states that every process that is thermaly activated is
regulated by its activation energy Q. So what you need to do is to measure
the value of interest V at two temperatures and solve the equation:

V(t)= V0x eexp(-Q/RT)
T= temperature in Kelvin
R= Stefan Bolzmanns gas constant

Like that you will find V0 as well as Q

But whenever you go for a statistical model never forget that your model is
only valid within the range you measured. Extrapolations are, now how
should I say, not very sensible.

Cheers

Guenter