I like Mark's idea.
I for one am not a fan of measuring in the dark due to safety considerations, as well as to my ability to sight tangents better in daylight. I have found myself finishing a measurement close to dusk, when the temperature has fallen below the temperature of the pre-cal, but only a handful of times over many years.
Experienced measurers may rarely find large differences between pre- and post- cal numbers, I believe. Usually, I have the same experience as Mark. My post-cal constants are smaller, due to the higher temperature at the end of measuring or by the time I drive to the cal course. In this case, I don't remember ever getting a surprise post-cal constant. There's no need to shorten a measured course in this scenario, IMO.
I have been using 100% nitrogen in my bike tires, which tends to reduce the expansion/contraction a little, though not as much as the solid tires that Duane uses. Still, for measurements that are done in constant or increasing temperatures all the way through, Duane's method seems logical to me.
I would appreciate hearing from more measurers, FS's, and certifiers about Duane's method as restricted by Mark's requirement. It seems to me it could result in more accurate measurements, as well as reducing unnecessary post-calibrations and unneeded shortening - and lengthening - of courses.