About fifty years ago, Chevreul showed what these accessory causes are, by exposing to light a number of dyed colors under varied conditions, e.g., in a vacuum, in dry and moist hydrogen, dry and moist air, water vapor, and the ordinary atmosphere.  He found that such fugitive colors as orchil, safflower, and indigo-carmine fade very rapidly in moist air, less rapidly in dry air, and that they experience little or no change in hydrogen or in a vacuum.  The general conclusion arrived at was, that light, when acting alone, i.e., without the aid of air and moisture, exercises a very feeble influence.  Further, it was determined that the air and moisture, without aid of light, have also comparatively little effect on dyed colors.  Abney and Russell, in their experiments with water colors, obtained similar results.

These conclusions are exactly in accordance with our common knowledge of the old fashioned method of bleaching cotton and linen, in which the wetted fabric is exposed to light on the grass, and frequently sprinkled with water.  If the material becomes dry through the absence of dew or rain, or the want of sprinkling, little or no bleaching takes place.

The one color which Chevreul found to behave abnormally was Prussian blue.  This faded even in a vacuum; but, strange to say, on keeping the faded color in the dark, and exposed to air, the color was restored.  It was shown that, during the exposure to light, the color lost cyanogen, or hydrocyanic acid, while in the dark and exposed to the air, oxygen was absorbed.  Chevreul concluded, therefore, that the fading of Prussian blue was due to a process of reduction.

The prevailing opinion, however, is that the fading of colors is a process of oxidation, caused by the ozone, or hydrogen peroxide, which is probably formed in small quantity during the evaporation of the moisture present, and both these substances are powerful bleaching agents.

It would be extremely convenient to have some rapid method of testing colors for fastness to light, and I believe it is the custom with some to apply certain chemical tests with this object in view.  The results of my own experiments in this direction lead me to the conclusion that at present we have no sufficient substitute for sunlight for this purpose, since I have not found any oxidizing or reducing substance which affects dyed colors in all respects like the natural color-fading agencies; further, I am inclined to the opinion that the action of light varies somewhat with the different coloring matters, according to their chemical constitution and the fiber upon which they are applied.

With respect to this last point, Chevreul actually found that colors are faster to light on some fibers than on others, and this fact, which is generally known to practical men, is abundantly shown in the diagrams on the wall.  As a rule we may say that colors are most fugitive on cotton and most permanent on wool, those on silk holding an intermediate position.  Still there are many exceptions to this order, especially as between silk and wool.