I recently carried out this experiment

Abstract

A barometer was attached to the inside of a 200l (44gal) drum. On a hot, humid day, when the temperature reached 32°C and the pressure was 1018hPa, the drum was sealed and then cooled to 10°C. The pressure in the drum fell to 998hPa. I repeated the experiment on another day when the temperature reached 32°C, but this time the air was very dry. The barometric pressure was 1020hPa. I again sealed the drum and cooled it to 10°C, and to my surprise the pressure only lowered to 1012hPa. It seems that the cooling of the air only accounted for a difference of 8hPa – the condensing water vapour having had a much bigger effect than I thought it would – so I researched the idea. It turns out water vapour has 1670 times the volume of liquid water. At 32°C, 85% relative humidity and 1018hPa pressure, the air consists of 2.7% water vapour, whereas at 10°C, the same volume of air can contain no more than about 0.8% water vapour, at which point it becomes saturated. The difference in water vapour quantity in the humid air as it was cooled from 32°C to 10°C was 2.7 − 0.8 = 1.9% 1.9% of 1018hPa is 19.6hPa 1018 − 19.6 = 998.4hPa At 32°C and 50% relative humidity, with a similar atmospheric pressure, only about 1.5% of the air is water vapour. This dry air will remain unsaturated until it is cooled to 12°C. So the difference in water vapour content in this case was 0.7%. 0.7% of 1020hPa is 7.1hPa 1020 − 7.1 = 1012.9hPa As the air inside of the drum represents a small part of the total atmospheric pressure, the effects of the cooling of its non-water part (<0.1hPa) may be ignored. Although the calculated figures are not exactly the same as those I measured, the small differences may be accounted for by small errors in my readings of relative humidity and pressure. Conclusion: When moist air is rapidly condensing into cloud it lowers the pressure of the immediate area. Therefore clouds do not only indicate a low pressure cell, but they also help to create it.