HIP9010

[Matthew Lee Franklin]

Jason wrote:

>If a straight line fit is required, take only two data points
>
>(jcborkow at ece.rutgers.edu)
>(jcborkow at eden.rutgers.edu)

Actually to get a better line fit (one that has some statistical 
significance), you're better off taking several points and then doing a 
linear regression.  Even though it does indeed take only two points to 
make a straight line, scatter in the taking of those data may nail you.  

If there is some uncertainty in the data, using several points for the fit 
will give you a better fit of the true realtionship.  

Later,
Matt

Matthew L. Franklin     
University of Minnesota, Department of Mechanical Engineering
111 Church Street Southeast
Minneapolis, MN 55455-0150
Phone: (612)625-4074, FAX: (612)624-1398, Home Phone/FAX: (612)331-3511