Why does water rise in the burning candle experiment?

Objective: 
To study the rise of water in the inverted glass which covers a burning candle placed in water. 

Equipment:
three plates, four similar candles, three drinking glasses of same size.

Introduction:
The experiment described in the first part is very famous and is used by many teachers and students to show that there is 21% oxygen in air. In this demo experiment I will show that the real physics of rising water is very different.

Procedure:
Put a candle vertically in a plate. Light the candle. Put some water in the plate so that a small lower portion of the candle is in water. The candle keeps on burning. Cover  the burning candle by an inverted glass. The candle goes off and water rises in the glass. How much water will rise in glass depends on the thickness of the candle and how much time you allowed the candle to burn before you covered it.
Use a candle and cover it quickly after burning. As the candle goes off, very small amount of water rises in the glass. It could be hardly 5% of the volume of the glass. Leave this set up as it is and take another plate, put a similar candle, pour water, light the candle and wait for some time. If a fan is running nearby put it off. Now cover it with a glass of the same size. This time water rise will be much more. 
Now take the third plate and put two candles in it. Pour water in the plate and light all the candles. Wait for some time and then cover both by a glass of the same size as used in the previous trials. This time the water rise will be very high, may be 40-50%.
What is the Physics of this rising water? When candle burns the air surrounding the flame becomes hot. The flame itself is very hot gases. The pressure of this surrounding air is the same as the atmospheric pressure as all air is connected. As pressure remains the same and the temperature rises the density goes down from the gas law
PV = nRT. For a given volume n will decrease if T increases.  When you cover the candle(s) you trap this less dense air. As the oxygen is consumed and the candle goes off, the air (gases in fact) inside the glass cools down. As the number of moles n is now fixed, decreasing the temperature will decrease the pressure and this will suck water in the glass. In equilibrium the temperature in the glass will be the same as the room temperature, the pressure will be P= P0-hrg, where P0 is the atmospheric pressure and h is the height water rises.
If you cover the candle just after the burning, the air trapped is not that hot. The density is thus not much lowered and hence on candle going off the water rise is not much. On the other hand if you burn two candles together the surrounding air becomes much more hotter and hence the water rise is high.

Discussion:
The experiment described clearly shows that the rise of water has no relation with the oxygen content in air. In fact for each oxygen molecule consumed, you produce a molecule of CO2 among other products. Also the solubility of CO2 is lower than that of O2. So there is no question of decrease in pressure inside due to consuming oxygen.
There is another factor that contributes in rising water in the glass. At higher temperature the saturation vapour pressure of water is also high. As the air in the inverted glass is in contact with water, it will contain saturated vapour. When the candle goes off and the temperature falls, saturation vapour pressure also decreases and some of the vapour condenses. This also decreases pressure inside and helps in rise of water.
Please note that water starts rising only after the candle goes off.