# TU-58

Fred Cisin cisin at xenosoft.com
Wed Dec 2 19:01:04 CST 2015

```>> I tried measuring a whole bunch of circles, and I can't find any
>> rational reason why dividing the circumference by the diameter never
>> came out even! :-)
On Thu, 3 Dec 2015, Johnny Billquist wrote:
> You need to measure more of them! You've just been unlucky.

OK!
I started to wonder whether I needed more bits in my floating point.
They just wouldn't resolve in the 24 bits of single precision.

Using such a bizarre number is just plain irrational!

Seriously, though,
in the computer math class, I did make the students manually
compute the some bit representations, including PI.
I found several students had encountered the same thing that I had run
into, of some grade school teachers misinterpreting their textbook and
declaring, "PI is about 3.1416 or EXACTLY 22/7"! (insertion of "EXACTLY"
was theirs)
In fifth grade, I got into a lot of trouble for argueing with that teacher.

I think that "tire" is a good term; it implies some of the special issues
that come up from compressing from only one side.  If you measure the
distance from roadway to axle, and imagine a tire that is a perfect circle
of that diameter, some of the rotations per mile issues clear up, although
obviously not explaining the amount of force required for propulsion
relative to what you would need with uncompressible round tire.
("spherical chicken in a vacuum")

Tony mentioned selecting based on OD, and then remachining the ID.  With
my extremely limited machinist skills, I'd be more inclined to look for a
match of the ID, and then let elasticity take care of at least part
of the OD discrepancy.   I've tried to machine rubber.  It obviously can
be done.  But only by somebody more skilled than I.

--
Grumpy Ol' Fred     		cisin at xenosoft.com
```