After many hit or miss attempts to measure the travel of the bellows using a vernier as a depth gauge to determine the average movement per degree of temperature change I finally spent two minutes to set up a camera and take pictures of a setup using a 1 inch travel indicator, an indoor outdoor temperature device and waiting for the sun to go down to make some measurements. Zero reading at 78 degrees. 52 pictures at 5 minute increments later the temperature dropped to 70 degrees. The bellows collapsed by .200, or a movement of .025 per degree of temperature change. At the 74 degree reading the collapse was .100. Earlier primitive tests showed a greater collapse per degree at the lower temperature range of 60 to 40 degrees. It appears that the .025 per degree of temperature is quite linear in the 80 to 60 degree range. Coincidentally this .025 per degree corresponds very nicely with the chain/pulley ratchet which appears to have a .025 tooth spacing. Due to the fact that this clock was designed during the 30's there obviously would have been no problem in getting the required temperature variation. I was concerned with the modern digitized thermostats that can hold a room temperature at less that one degree variation may not be sufficient change to operate the Atmos. A small identification mark on the mainspring ratchet wheel can be an aid in determining if the clock is winding properly. If everything is in order and the clock does not wind, perhaps a different location that will subject the clock to a greater temperature change may be all that is required. A location in a room that may be subject to the heat from the sun during the day may be all that is required. If a 2 degree change in temperature does not wind the clock there is probably a defective bellows. A rapid change in temperature for a short period of time may not be suficient due to the fact that the bellows is insulated to some extent by the clock case which can affect the time required to raise or lower the temperature of the bellows.