Humidity measurements
For a long time, I have had a Honeywell thermometer with a humidity reading. Lately, it has been stuck at 15%, which made me wonder whether my apartment is really so dry, or whether the hygrometer is broken. As a person with a technical background, I decided to find out!
In a nearby mall I bought five different hygrometers of two different brands:
Originally I though that taking measurements for a day would be enough. This is what I saw (you can guess which line corresponds to which device on the picture above):
This answers my original questions: My Honeywell device really is broken and indeed, my place was way too dry. After turning on a humidifier in the morning, the humidity levels started going up. The dip around 5pm is when the humidifier ran out of water. Interestingly, it seems that when the humidifier is running low on water, it is not capable of keeping the humidity levels up.
However a new question arises - which of the measurements is right? Clearly the measurements cluster into two groups (red/orange and green), which happen to correspond to the two different producers (Accutemp and Taylor). This suggest different calibrations between the two companies.
To resolve the issue, I noticed that Toronto Public Library lends CO2 monitors, that so happen to also measure humidity:
Another day of measurements gave me this:
The library device readings are clearly closer to the Tylor group, though they are the lowest among all the functioning devices. This is somewhat suspicious and puts into question my original expectation of (close-to-)perfect alignment with one of the two groups.
While the calibration of the hygrometers is thus very uncertain, and the correct value is probably "somewhere in the middle", it was nice to see that all the devices show values that are very tightly correlated. The correlation coefficients between any two devices were between 0.88 and 0.98 and can be appreciated for example on the following comparison (which seems to show a hint of nonlinearity):
