Let's say we wanted to subtract these measurements.
We can do the calculation exactly:
45.367 - 43.43 = 1.937
But let's take the idea that measurements were rounded to that last decimal place.
So 45.367 might be as small as 45.3665 or as large as 45.3675.
Similarly 43.43 might be as small as 43.425 or as large as 43.435.
So our difference may be as large as
45.3675 - 43.425 = 1.9425
or as small as
45.3665 - 43.435 = 1.9315
If we express our answer as 1.937 that means we're saying the true measurement is between 1.9365 and 1.9375. Since we determined our true measurement was between 1.9313 and 1.9425, the measurement with more digits overestimates the accuracy.
The usual rule is to when we add or subtract to express the result to the accuracy our least accurate measurement, here two decimal places.
We get 1.94 so an imputed range between 1.935 and 1.945. Our actual range doesn't exactly line up with this, so we're only approximating the error, but the approximate inaccuracy is maintained.
