I was curious about this myself so I ran a test.
Loaded a couple hundreds rounds of 40 S&W using WW231. (My testing ammo all was from the same lot of primers and powder, cases were the same headstamp and the setting on my loading press were the same - powder was checked every 50 rounds.)
On a hot summer day I fired some of the test load over the chronograph. Over the next few months as the ambient temperature dropped I fired more of my test loads over the chronograph. Generally speaking this testing removed the variables except for temperature from the velocity testing.
For WW231 the results are that in my gun, shooting my load, WW231 is temperature sensitive, BUT inversely sensitive. Velocities went down as the temperature was higher. My velocity loss was about 1 - 2 % for every 10C of temperature. Given the limited sample size and typical Sd, the velocity loss calculation is not any more precise than that.
This difference does not explain what you saw.
In your situation I would argue that the velocities provided in reloading manuals are "guides" not an expected value. While you may have had some examples where your velocity matched the reloading manual, my experience is that reloading manuals only give you a rough starting point for velocity. Compare different manuals and you will find a wide variation in "velocities" for similar loads.
When loading your 158 loads, the seating depth will impact the velocity as will the shape of the bullet (how much of the bullet is nose to be left outside of the case and how much is base to be stuffed inside the case which changes the pressure and burning characteristics of your load). Additionally the amount of crimp you used, the brand of brass, cylinder gap, even changing from one brand of 158 SWC to another brand of 158 SWC can all cause a change of velocity. The temperature sensitivity of your powder is not the first thing that needs to be investigated.
Hope this helps.
