A very interesting project. 1st, I would like to say that as scientist; it is our role to put forth a hypothesis and then conduct experiments to prove or disprove that hypothesis. The observations are always exactly what they are. As long as the experiment and results are repeatable, then you can be fairly confident that they are correct observations even if they don't match what you expected them to be.
That being said, I looked at the experiment procedures mentioned. Dry air at normal room temperature and pressure, has a density of approximately 1.2 kg/m^3, while steam (boiled water) has a density of approximately 0.6 kg/m^3. When heating the water, you added energy, causing the water molecules to move around much faster until they change from a liquid to a gas. As the liquid water became a gas (steam), the density of the air immediately above the surface of the water was potentially reduced by 50%, effectively creating a localized vacuum. By sealing the opening of the flask, you prevented dry air from reentering the flask from the outside, thereby maintaining the vacuum. The bell was now suspended inside a container that had about 50% less "stuff" to help propagate the sound wave. You didn't mention it, but I suspect that as the flask (and water/steam) cooled, the sound from the bell became even more quite. This is because as the steam cooled and changed back into a liquid the air around the bell became even less dense (more of a vacuum) and since the stopper was still in place; air from the outside was still unable to get in and take its place. This vacuum sealing effect is used in canning food. With less and less of a medium to propagate the sound wave, the bell would sound quieter and quieter.
I hope this matches your observations and answers your questions. Don't hesitate to post back with further questions.
I hope this helps.
“Education never ends. It is a series of lessons, with the greatest for the last.”
~ Sir Arthur Conan Doyle (Sherlock Holmes)