On Mon, 16 Jul 2001 17:46:18
Bill Potts wrote:
>Gene:
>
>That's all very interesting, except that we were not talking about
>calculating the precision. Rather, we were talking about the degree of
>precision implied by the use of the extra decimal digit.
>...
Well, Bill... I'm not sure that my reply was dealing with that aspect actually
('talking about calculating the precision').
Please remember that my comment was aimed at basically two things.
First, that for all intents and purposes, when we talk about "accuracy" *in practical
terms* we're "tackling" that by adding decimal places to a measurement (just like Gene
mentioned, the more the measurement's mean reflect the true value, evidently the more
it would be associated with a higher number of decimal places).
To illustrate this it suffices for one to "go to the extreme". Imagine we were able
to match *exactly* the measurement with the true value (i.e. perfectly accurate).
Evidently in this case the result would "carry" all decimal places one could think of
(i.e. infinite... ;-) ).
Now, with precision, the story is quite different (and I wasn't really talking about
that aspect of metrology at all). The illustration I was alluding to is a cartoon
picture showing a big dart board with a *lot* of darts around a tiny spot, but a spot
which was way off the center target! This is *precisely* ( :-) )what Gene was
mentioning, "the scatter" is very "tight".
My second point was, if I'm not mistaken, that when measurements (especially height)
are lower than a meter, it's standard practice in "old" metric countries to show the
results in cm, as in 85 cm.
So... It seems you "took on an unnecessary... 'fight' " here, Bill. I guess that in
essence we've both made our points adequately in our own way...
But in essence the common practice that I know of (and that I was alluding to) is to
simply refer to the specific 'decimal place accuracy' for a certain measurement (or
that a value would be accurate to a certain decimal place), already assuming that
precision matters are already being adequately addressed.
(Even though these are "randomly distributed", it would be assumed that they would
be... "under control" or known to be at an acceptable level, so to speak, by the
individual conducting the experiment. Just as an illustration in one of our labs
students are required to "demonstrate" that by taking several measurements under the
"exact" same circumstances to verify that aspect, also repeatability. But once
they're "happy" with them and have "confidence" in the instrument, they wouldn't
"bother" continuing with this kind of exercise for every single experiment).
Anyway, I just wanted to share how we use this "jargon" among engineering and academic
circles.
Marcus
Get 250 color business cards for FREE!
http://businesscards.lycos.com/vp/fastpath/
