really sorry, but i think you explantion of 'rods that are gathering the information that is not focussed centrally, therefore all of that similar input can be summated and amplified to give an even more powerful signal to the brain' can not work,
as edge recognition/contrast (and so reading) is based on field receptors who are different on the fovea and the extreme part of your eyes. Furthermore, the way photons are integrated summed implies more sensibility but far less discrimination (lot of information is lost, especially about spatial location)
In fact, on the fovea you have specific field receptor (for ganglionnal cells) who are able to distinguish contrast wereas side field receptors are able to distinguish frequencies/colors (that's why moovement detection is far better on extreme part)
But anyway, the way the book is held when PR implies that it is *still* in the fovea (for the greater part), not much use of any 'peripheral vision'.
Check it out, just ask a friend to put his hand at 45° of your eyes, you cannot count fingers but clearly sees if they are moving.
I May have an hypothesis about why PR can work, but it is quite long... anyway here are 2 facts that may helps understanding it:
1) there is a spot (a negative scotome) on the retina that is called 'the blind spot of Mariotte' where the optic nerves ends on the retina, without any photoreceptor (nor rods nor cones). It is nearly impossible to notice it consciously. First we may think that binocular vision helps us to reconsctruct missing information. Wrong, as closing an eye doesn't change anything. In fact there is psychic adaptation and reconstruct based on 'what is probably here'.Furthermore, experiment made showed that stimuli presented on this spot were non-consciuously treated.
2)A section made on the superior colicullus makes you half blind (a entire field is inacessible). Sometimes, a degerenescence can make the same section, so we have tried on humans to present a stimulus on this field. It was biologically inaccessible. But when asked to 'guess' where the spot was, they gave with a very high rate the exact location of the spot (much higher than just random answer).
What to conclude from these 2 facts ?
perception and vision are tightly related but perception does not rely entirely on what is sent by the eye (trought the optic nerve) but also by oher ways (the bidirectionnal data sent by amacrines and horizontal cells in the retina). Ways that are not consciously available, but still present. And that it is possible to reconstruct missing information based on probabilytisc model (and a lot of fourier analysis that is seem to be made in the V1 (stried cortex))
Actually there are a lot of arguments in same scheme...
Oh and btw i really think (based on this perceptual model) that previewing/ trigger words is much more important than what it seems...experienced preaders, plz confirm ?
btw sorry for my poor english, not much used to any biological explain in this language...sorry if it was a pain to read 
