The technical specs of the eye are really contentious with precious little consensus. From what I've been able to research, it's been speculated that the human brain can only process ~10-24 frames per second but the eye can see much more than that, upwards of 300 FPS. How that information is determined is beyond me.
As for reading comprehension, I don't know; however, there are so many factors involved (language acquisition, diction, context, structure, etc.) that I suspect there aren't hard and fast rules. Somebody with experience in cognitive psychology might be able to answer this part of the question better.
All good points. But, we need to start to take the human machine into account and bring a little more of the science into the UX work being done. Thanks for the input.
Tobii has a few answers to this that we hope will get at what you're asking. Typical reading in English characters in normal font size will make approximately every 8-9 characters into one fixation. The duration of the fixation is around 180 – 450 msec of time (the more complex stuff you are reading the longer the fixation length). Once in a while the brain will move back if the saccade went for too long. The long saccade into the beginning of the next row is relatively difficult, hence we prefer to read narrow columns rather than wide wikipedia text lines. Additionally, accuracy is critical for medium-sized and especially for smaller-sized fonts, otherwise the participant will analyze the wrong line. We always suggest choosing slightly larger font sizes or greater spacing between the lines.
Reading happens mostly in the brain...the eye movement is only a small part. The major take-away is that the brain makes sense out of those chunks of 8 character signs.
Please email me at email@example.com if you'd like to know more.