[RASMB] Re: An early Christmas puzzle

[John Weeks]

Arthur Rowe <arthur.rowe at nottingham.ac.uk> was asking a question about
temperature control in his apparatus.

I can't pretend to know about AUC- I joined the list as a lurker to try to
understand a bit about what some of our customers do with our software.
However, I did  spend about 25 years in a lab, some of that time spent
using temperature controllers.

1. The temperature controller's spec (unless there is something to say
otherwise) tells you how closely the controller will control the measured
temperature. It does not include things like calibration error in your
thermocouple (or RTD or thermistor, or whatever).

2. In my experience, the largest source of error in temperature control is
not in the controller, but in temperature gradients. The temperature may be
known to within 0.5 degree where it is being measured, but what is the
temperature at the ends of the sample? (Hopefully the sensor is placed near
the middle of the sample chamber)

>Given the type of error distribution expected, can we really accept that
>it is better, using 2 instruments, to get 2 estimates for an s value that
>dis-agree than 2 which agree? Or is this believing in Santa Claus?

I think it's Santa Claus. As I said, you know that the controller is
keeping the measured temperature within a half degree, but you don't know
much about other sources of error.

I will now remove myself from an argument where I don't really have
expertise :)

Regards,
John Weeks

WaveMetrics, Inc.
Phone (503) 620-3001
Fax   (503) 620-6754
email   support at WaveMetrics.com