# Uncertainty in 1/x

Discussion in 'Science' started by Newstein, Jan 17, 2019.

Tags:
1. ### NewsteinNew commenter

If x = (10.0 +- 0.1) cm, 1/x = (0.10 +- ?)

2. ### sparkleghirlStar commenter

As a scientist, I'd look at it like this:

1/10.1 = 0.099 and 1/9.9 = 0.101
Both of these are 0.001 away from your value for 1/10 so that's what I'd give for the uncertainty.

Another way of looking at it in science wouold be to say the original uncertainty as a % is 0.1/10 which is 1%, so we'd expect an uncertainty of 1% on the answer as well, which is what 0.001 out of 0.1 would give.

This might not be mathematically rigourous, i don't know, but it's what we'd do in a lab. I think.

rehaank, CheeseMongler, BG54 and 3 others like this.
3. ### gainlyStar commenter

Sparkleghirl's analysis seems perfect to me.

Newstein likes this.
4. ### BG54Occasional commenter

Yes - also if we use fractions as exact values, then the mid-point for the range of 1/x would strictly be 1000/9999 or 0.100010001.... rather than 0.1, but that's probably nit-picking.

5. ### rehaankOccasional commenter

I just did exactly this on a post it note - inverse propagation of uncertainty i think

6. ### NewsteinNew commenter

Take a micrometer reading like x = (0.10 +- 0.01) cm. The uncertainty in 1/x then is
(1/2) (1/0.09 minus 1/0.11 ) = 1.02 cm^-1
Well, I have to accept this as well.

7. ### NewsteinNew commenter

It transpires that x and 1/x have the same % uncertainty.