Looking at something like this sensor readout table for an x-h2s from dpreview, I was trying to understand how one would interpolate to find the readout speed for 1080p240. The 240p on this camera is exceedingly soft, supposedly because of the sensor resolution and needing more lines to be skipped because there is not enough speed available to do a full pixel readout even at the 1.38x extra crop.

So when we talk about reading the sensor lines, does it read the lines pixel by pixel, or somehow read the entire line in one instance?

I had initially assumed it was pixel by pixel across each line. In this thinking I was looking at the relationship between rez and timing, which definitely doesn't appear to be linear (6.2k --> UHD 4K is 3x MP drop but only 20% speed increase). But when considering lines vs MP, for each doubling in lines there is a 4x increase in resolution. So depending on if these readouts are based on total res or just number of lines will definitely affect those calculations.