From wiki:
A simple variometer can be constructed by adding a large reservoir (a thermos bottle) to augment the storage capacity of a common aircraft rate-of-climb instrument. In its simplest electronic form, the instrument consists of an air bottle connected to the external atmosphere through a sensitive air flow meter. As the aircraft changes altitude, the atmospheric pressure outside the aircraft changes and air flows into or out of the air bottle to equalise the pressure inside the bottle and outside the aircraft. The rate and direction of flowing air is measured by the cooling of one of two self-heating thermistors and the difference between the thermistor resistances will cause a voltage difference; this is amplified and displayed to the pilot. The faster the aircraft is ascending (or descending), the faster the air flows. Air flowing out of the bottle indicates that the altitude of the aircraft is increasing. Air flowing into the bottle indicates that the aircraft is descending.
****
So, the "capacity" - the bottle you see - is increasing the sensitivity of the vario. It is common with older mechanical varios, which can work very well. For best performance, they should be insulated - either a thermos bottle, covered in foam, and I recall that some recommended filling the inside with copper wool, to reduce heating/cooling effects (larger thermal mass). If you take the bottle out, the sensitivity of the vario will decrease (it will become useless). Newer, electronic varios use thermistors, which replace the capacity, using ones and zeros to differentiate wildly in all directions (a techno-approximation), rather than a simple bottle.
A more detailed explanation is on the eglider site,
http://www.eglider.org/newsarticles/
totalenergycompensationinpractice.htm.
I should mention that the connections are prone to leaking, and a leak affects your compensation. Redoing the connections should be done periodically (there are good how-tos on some sites online), and in the major soaring books.
Cheers.