Precision and linearity

[Nigel Smith]

In article <9603291834.AA29206 at mercury.med.pitt.edu>, bap at MED.PITT.EDU
(Bruce Phillips) wrote:

>         Recently, Dwight Lynn had the audacity to communicate:
> 
>> I'm sorry to inform you that you cannot determine a linear relationship 
>> with just two points.  You'll need to set up another centrifuge at 3X for 8 
>> years and 8 months.
> 
> Dwight Lynn
>
>         Au contraire, two points are the surest way to establish a straight
> line.  In the realm of good science, I am willing to get the points BEFORE
> drawing the line.

Two points certainly define a line, and if want a straight line then two
points is all you should take.

However, if you want to prove to your peers that the line *is* straight, I
would suggest you plot a few more...

      Nigel