For those interested in a different approach related to the thread’s subject see:
https://geodesy.noaa.gov/PUBS_LIB/NAVD.pdf Article 11 “Survey Instrumentation and Procedures,” by Harold Beard. It includes a description of “Double simultaneous leveling” used during leveling for NAVD88. While not mentioned in the article but alluded to by Larry Scott, there are uncorrectable refraction biases due the one-way run. The FGCS leveling specifications https://geodesy.noaa.gov/FGCS/tech_pub/Fgcsvert.v41.specs.pdf allow the method but with conditions.
For those who have never seen one, I attach a photo of the rod used.
I'm a fan off double simultaneous leveling using 3 meter rods with the 301.555 centimeter offset between the left and right sides of the Invar strip.
I haven't needed them for years, and if anybody has a use for them just come to Pasadena, California and pick 'em up.
I can see how refraction can cause a bias because your uphill line of sight is closer to the ground than the downhill one, and the refraction won't be the same.
But how does a two-way run correct for that any better than a double rod one-way run?
At best, the different times would give you two different temperature gradients, but they would tend to be similar.
Seems legit to me. a little extra care in reading and recording saves A LOT of walking (time). We did traverses like that in Alaska "diamond traverse" is what the BLM guys called it. Saved a lot of effort. I don't think we ever had to re-run one because you knew it was critical to get the readings right the first time out there.
Thanks for all the answers, I've learned a couple of things out of this thread.