Do I need two voltage measurments for a split phase system?

I’m in Canada, measuring the current on two legs of a split 240v phase, 120v on each leg, using two CT sensors.

I’d like to add voltage measurement to this, but it occurred to me that I will only be measuring voltage on a single leg. I take it this means I should plug the AC-AC transformer into a circuit on the leg that is most likely to have reactive loads with a power factor less than 1, and the other leg will basically just be wrong?

Ideally, you should measure both voltages, and if you have the 3-phase emonVs, and upgrade to the emonTx4 or the emonPi2, then you can do exactly this - I recently extended the ‘Docs’ article to include the North American system.

However, I think you are inferring that you have maybe an emonTx V3 with the plug-in a.c. adapter? In this case, you need to choose the leg that’s “typical” of the voltage that the two legs deliver. I would only consider the load power factor(s) if it makes the voltages on the two legs significantly different. As you write, the one you use will give correct real power and the other will be wrong - but most likely less wrong than using a guessed and constant voltage estimating apparent power all the time. With experience, you might find that one leg is consistently different to the other, and in this case a small adjustment to the calibration might spread the errors out and improve the overall accuracy.