Abstract: The constant humidity property of some saturated saline solution can be used to design humidity generators. Current humidity generators using saturated saline solution usually require complex equipments and a long time to reach equilibrium, and are prone to environmental influences. On the other hand, air bubbles are characterized by good air-tightness and short inside mean diffusion distance, which is defined as the volume above the solution divided by the surface area of the solution in a closed container. Therefore, it is easier to achieve vapor equilibrium within an air bubble, which can then be collected pneumatically and used to calibrate humidity sensors.The conditions under which water vapor reaches equilibrium within air bubbles are quantitatively analyzed in this study. An air bubble can be modeled as a closed cylinder container with solution at the bottom. Assuming that the gradient of vapor density is distributed linearly along the axis of the cylinder, the time τ required to reach vapor equilibrium within the air bubble can be deduced from the relationship that the vapor mass diffused from the solution surface into the container within a unit time equals the increased vapor mass within the container. From the forces exerted on the air bubble during its ascent, the time t for the air bubble to remain in the solution can be calculated. In order to reach vapor equilibrium within the air bubble, τ must be smaller than t (τ < t). It follows that the diameter of the air bubble r and the depth of the solution h should satisfy the inequality below: in which ρ is the solution density, g is the gravity acceleration, D is the vapor diffusion constant, η is the viscosity of the solution, U₀ is the normal humidity at the surface of the saturated solution, U₁ is the initial humidity in the air bubble, and U₂ is the humidity of the air released from the air bubble when it bursts.To test the new humidity generator described above, a test set is calibrated against a M4 hygrometer (General Eastern Instrument) with the dew point temperature uncertainty of 0.2 ℃. The absolute values of the differences between readings from the test humidity generator and those from the reference hygrometer are no bigger than 1%. The response time is typically 10 minutes under common application conditions.The sources of error for the humidity generator using the air bubble method are also discussed. The analysis and the test show that a humidity generator based on air bubbles from saturated saline solution is simple, fast, accurate, and less affected by environmental factors, and is a real utility humidity reference, in particular can be used in onsite calibration of humidity sensors at automatic weather stations.