He is using the 1951 scotopic vision chart -- scotopic vision is produced exclusively through rod cells which are most sensitive to wavelengths of light around 505 nm (green-blue) and are insensitive to wavelengths longer than about 640 nm (red). The sensitivity to light in scotopic vision varies with wavelength, though the perception is essentially black-and-white. The Purkinje shift is the relation between the absorption maximum of rhodopsin, reaching a maximum at about 500 nm, and that of the opsins in the long-wavelength and medium-wavelength cones that dominate in photopic vision, about 555 nm---so the relative brightness calculations don't reflect actual subjective human vision/perception in the real world.
As one member put it here in post #29 here:
http://laserpointerforums.com/f44/new-tool-calculate-relative-brightness-wavelengths-nm-61238-2.html
"There is one glitch with the scotopic curve where lasers are concerned, however - it is rather unlikely for a person to view a typical laser beam using scotopic vision! In fact, I can't remember EVER viewing a laser beam with scotopic vision!
Here's a big hint...if you can even vaguely tell what color the laser beam is, then you are not seeing it with scotopic vision!
What you are using when you view a laser beam under subdued or nighttime conditions (but can still tell what color it is), is mixed-mode vision (both rods & cones active).
In this case, the response curve is the combination of both the photopic and scotopic curves! The relative ratio of each will depend on how active each visual system is - the darker it is, the more it will shift away from photopic and towards scotopic."