A series of atmospheric pressure measurement made with a precision barometer (resolution 0.5 Pa) was made automatically.
(i) A short series of measurements in the morning showed what seemed to be stable results and a mean, standard deviation and standard uncertainty were calculated according to the rules.
(ii) After lunch, the mean pressure seemed to have shifted significantly more than expected. When the whole day’s data was examined, it was revealed that the barometer reading was oscillating with a large-amplitude sinusoidal curve of long period of several hours. It turned out that the thermostat control of the barometer had failed, leading to a corresponding oscillation in gauge temperature and subsequent barometer reading.
Unknown errors => Measurement uncertainty.
An arguably greater challenge to the metrologist than statistics!