Castle Drogo: thermal expansion and asphalt

The main problem with the roof has been the very early use of asphalt for a flat roof, before its properties were fully understood. It expands on hot days and contracts on cold days. This makes it crack and let water through. I was interested to look up the coefficient of thermal expansion because I have never really studied it. Turns out it is measured in units of "per degree Celsius". Length doesn't come into it because it is calculated as expansion per unit length - in other words, it is a strain. https://en.wikipedia.org/wiki/Thermal_expansion A brief internet survey suggests that asphalt may be non-linear and complicated. Here is one source https://d3dqsm2futmewz.cloudfront.net/docs/smartWebArticles/mat_thermprop.pdf If the coefficient is say 4 x 10^-5 per degree Celsius, the max imum temperature range between hottest summer and coldest winter might be 35 degrees Celsius. Over a 10 metre roof section it could mean a change in length of 14mm. That's quite enough to make big cracks.