Two consumer hygrometers in one room disagree less on dew point than on relative humidity

Hypothesis

Relative humidity is a ratio that depends on temperature. The National Weather Service page on dew point versus humidity states that relative humidity can be misleading and that the higher the dew point, the greater the amount of moisture in the air; its example gives 100 % RH at one temperature and 50 % RH at another for air that is more humid in the second case. A consumer temperature-humidity display reports RH at the temperature of its own sensor. The premise of this hypothesis is that spots within one room, at different heights or near a window or a radiator, differ in temperature by amounts such a display resolves, while the moisture content of the air is more nearly uniform if the air mixes. The hypothesis: for two such devices in one room, the difference between their dew points, each computed from the device's own temperature and RH pair, is smaller than the difference between their RH readings, after the fixed inter-device offset established side by side has been subtracted.

Prediction

In a log of paired readings, the spread of the RH difference between the two positions is larger than the spread of the dew-point difference, and the RH difference correlates with the temperature difference between the positions. During and shortly after a local moisture source near one device (a shower, a kettle, laundry drying), the prediction fails, because the air is then not mixed.

Proposed test

1. Place the two devices side by side for 24 hours and record their temperature and RH offsets.
2. Move one device to a second spot in the same room (a different height, or a different distance from the window) and log both at fixed times for 14 days with a conditions column, as in the indoor temperature and humidity protocol.
3. Compute the dew point for each pair with a published formula; compare the distributions of the RH difference and the dew-point difference, and check the correlation with the temperature difference.
4. Swap the devices between the positions and repeat, to separate device from position.

Status

No result is claimed. If the air is well mixed the hypothesis follows from the definitions; the test asks whether that assumption holds at the scale of a room with consumer sensors that may state no accuracy figure, and how often local sources break it.