We have studied in this paper the effect of the screen area illuminate
d or number of pixels used (with pixel-ratio-unity) on two possible mo
dels which in practice are used for producing and predicting colours s
pecified by their coordinates (x, y, L). The first model provides the
mixtures through the response curves (luminance versus DAC values) of
the individual phosphorus ; this is the most usual model used for CRT
calibrations. The second model is calculated from the response curve o
f the achromatic (white) which is obtained by making the three DAC val
ues equal. It was found that the luminance of the white is attenuated
by the number of pixels, the lower the DAC value, the slighter the att
enuation. For instance, in the CRT used, this attenuation disappears f
or DAC values under n = 35. Likewise, the attenuation is not appreciat
ed when, for each DAC value provided, the area is equal or smaller tha
n 100 x 100 pixels. If we were to assume any of these constraints, bot
h models would lead us to highly acceptable results. For greater scree
n sizes than 100 x 100 pixels, the errors in the luminance using the m
odel 1 are unacceptables and they can be reduced using the model 2, up
to a maximum DAC value beyond which it doesn't apply, however the pre
cision in the coordinates x, y, remains similar. Although these attenu
ation problems don't appear in high quality CRT monitors, in many expe
riments (specially in psychophysics) an ordinary monitor can be used t
hanks to the acceptable accuracy in predicting luminances and chromati
city coordinates achieved with model 1 and the solutions which, with s
ome restrictions, model 2 gives to these problems.