2004 lightfastness tests
This page summarizes the materials and procedures used in my 2004 paint lightfastness tests and the results and recommendations based on them. The results for individual paint samples are published in the guide to watercolor pigments.
These tests supersede my 1998 lightfastness tests, which used an alizarin crimson exposure scale and averaged my results with those reported by the ASTM, paint manufacturers and other published sources. I now believe the averaging method merely perpetuated inaccurate industry or ASTM pigment ratings.
My fundamental recommendation is that artists test their own materials for themselves, using the methods described at doing your own lightfastness tests. The principal reason is that all published lightfastness ratings are unreliable, as explained in the section lightfastness with a grain of salt.
lightfastness test procedures
The test location was near Sebastopol, California (latitude 38.4°N) at a private residence located on a south facing hillside about 8 miles from the Pacific Coast.
Up to 36 paint samples, each diluted in a gradation from masstone to tint, were painted in close array on single 12"x16" sheets of Arches block watercolor paper, for a total of 21 sheets (over 750 samples). The masstone end of each sample was painted across parallel black lines made with an indelible (Sharpie) felt pen to assess changes in paint opacity (hiding).
Sheets were stapled side by side on two 48"x48" pieces of 3/4" fir plywood under cover sheets of 1/8" Acrylite OP-4 (UV transmitting) clear acrylic. Strips of aluminum metal tape were applied to the underside (sample side) of the cover sheets to mask one half of each paint sample and two blue wool exposure cards, and the sheets were bolted to the panels at regular intervals across height and width to ensure tight contact and consistent registration between test papers and the metal tape.
Displays were mounted outdoors for a total of 820 hours on a south facing residential deck and at an angle perpendicular to the sun's rays, brought out each day of clear or partly cloudy weather no earlier than 10am and returned inside no later than 6pm. Displays were kept indoors on rainy, overcast or foggy days.
A spreadsheet was used to record daily weather information and calculate accumulated sunlight exposure; across all days (including rain days), average sunlight exposure was 6.5 hours/day at an average maximum temperature of 83° and average maximum dew point of 52°.
Samples were inspected daily through the first week of exposure, every two days through the second week of exposure, weekly through six weeks of exposure, then biweekly thereafter: BWS 3 was reached after 40 hours, BWS 4 after 80 hours, BWS 6 after 320 hours, and BWS 7 after 750 hours of full sunlight exposure. Color change was evaluated visually under indoor daylight and through a narrow window cut in a 3"x5" gray card, as specified in ASTM D5398.
As I explain in the section on lightfastness with a grain of salt, variation in the results of lightfastness tests are inevitable. You should be skeptical of published lightfastness ratings, including my own, and should do your own tests on your own paints if at all possible.
Specifically, the lightfastness ratings of watercolor pigments, published by the American Society for Testing and Materials (ASTM) in their technical report D5067-99, are inaccurate for several important pigments, noted below. In some cases, the observed variability in pigment lightfastness from one watercolor paint brand to the next was so large that no single lightfastness rating can be meaningful. (See, for example, PY3, PB27, or PV23.) No paint company or professional artist should rely on these ratings until the ASTM has repeated the tests across a larger sample of pigments.
That said, I summarize here the major test results that confirm or contradict the testing results published elsewhere:
alizarin crimson (PR83). This pigment is unsuitable for professional artistic work, period. This has been confirmed in every reputable lightfastness test conducted since the late 19th century.
genuine carmine (NR4). Useful as a food coloring, completely useless as an artist's pigment.
genuine rose madder (NR9). Though slightly more lightfast than alizarin crimson in my tests, this pigment is also unsuitable for professional artistic work.
2004 lightfastness tests in progress
naphthol reds (see separate pigment links under synthetic organic pigments). Many of the lightfastness problems I discovered were in paints made with naphthol red or orange pigments. The exceptions were PR112 and PR188 (which were the most durable of all). The pigment PR170, offered by a few watercolor brands, seems a marginal pigment to me and should be avoided. I recommend you test for yourself any paints made with a naphthol pigment.
aureolin (PY40). Most brands blanched and turned gray, and the pigment reaction varied acrosss brands. Choose an alternative light yellow paint.
cadmium yellows (PY35). This is usually a very reliable pigment, but I found a few instances where the paint darkened significantly under exposure to light, generally in the light or lemon shades that contain more zinc and probably also contain some lead, which blackens cadmiums. As this change appeared after only a few weeks of sunlight exposure, it is easy to produce in the simplest lightfastness test.
synthetic iron oxides (see page of "earth" pigments). With a few exceptions, all the iron oxide pigments held up well in my 2004 tests; I found very little of the pigment darkening reported in my 1998 test results.
convenience greens (labeled hooker's green [deep or light], sap green, permanent green [deep or light], prussian green, olive green, emerald green, etc. under PG7 and PG36). I found several problems with convenience green paints, including paints by reputable manufacturers. Surprisingly, the worst problems appeared in dull and/or dark green mixtures; the bright, yellow green paints fared well. Be cautious in your use of any convenience green until you have tested it yourself.
iron blue (labeled prussian blue or antwerp blue, PB27). I was surprised by the variability and peculiar behavior of this pigment. The best brands seem dependable, but some brands showed significant fading. The pigment variant antwerp blue is especially unstable and should be avoided.
quinacridone magenta (PR122). A pigment rated as "fair" (III) in the ASTM (1999) tests, I found that its lightfastness was equal to or even exceeded that of many brands of quinacridone rose, which is generally believed to have "very good" lightfastness. And this reliable permanence was found across all brands using the pigment, without exception. There is absolutely no reason to disparage this useful pigment.
dioxazine violet (PV23). The ASTM watercolor lightfastness test booklet (1999) offers a conflicting but discouraging evaluation of this pigment. I replicated these results but with a different emphasis. My tests showed its lightfastness is variable across paint manufacturers, but the best pigments are easily more lightfast than a high grade naphthol red, such as PR188 or PR112. The problem is ... finding paints made with the high grade pigment! Here it is important that you do your own lightfastness tests, and if you cannot, you should avoid this pigment.
convenience purples (see partial list under purple paints). It is now common for paint manufacturers to provide violet or purple paints as a mixture of ultramarine blue and a quinacridone rose or magenta, on the assumption that the mixture of two relatively lightfast pigments will also be lightfast. However, my tests showed that many of these convenience mixtures are no more lightfast than the dioxazine violet they are used to replace, especially in purples with a red rather than blue color. For masstone or heavy color applications dioxazine violet is probably just as useful; for tints an ultramarine violet (PV15), cobalt violet (PV14) or manganese violet (PV16) should be used instead.
colors whitened with zinc oxide (for example most brands of naples yellow). I frequently found that convenience paints formulated with zinc oxide (PW4) whitened under sunlight exposure. This occurred because the white component of the color became more opaque. I did not observe a similar effect in the (few) convenience mixtures formulated with titanium dioxide (PW6), which also has a very good track record in architectural paints and coatings. In general I recommend you avoid any paint containing PW4, and when white mixtures are wanted in your paintings, use PW6 to make them.
Artists are free to use whatever pigments, materials or techniques they want. The intent of my lightfastness testing was to (1) make artists aware of what lightfastness means and how it is measured, (2) inform artists about the lightfastness of the paints (inks, papers, techniques) they use, and (3) provide that the buyers of artworks are informed of any permanency issues raised by the choice or use of materials.
It seems to me that artists who choose to ignore these issues are either misinformed or misguided. At worst, they poison market confidence, and depress painting prices, for all other artists.
The same comment applies to the paint manufacturers whose technical or marketing materials do not report any lightfastness ratings for their products, or who publish misleading or fabricated lightfastness information.
In my own tests, I found that Art Spectrum, Daniel Smith, DaVinci, Maimeri, M. Graham, Utrecht and Winsor & Newton choose quality pigments and publish accurate lightfastness information about them the paints performed as promised in my own tests (both in 1998 and 2004), and they get good ratings in all the evaluations I have discovered. I feel these brands can be used with relative confidence.
Paints by Blockx, Holbein, Lukas, Old Holland, Rembrandt, Schmincke, Sennelier and Yarka present a mixed bag. Some of these brands publish no lightfastness information at all, or clearly rely on generic pigment ratings that do not describe the manufactured paints. Some claim adequate lightfastness for pigments that are known to be fugitive or that faded in my tests more often than random quality lapses by the pigment manufacturer would explain. And some claim to use nothing but lightfast pigments even though the ingredients in some of their paints are indisputably known to be impermanent. Overall, I feel these are paint brands you should use with caution: always test their products for yourself.
Unfortunately I obtained samples of Daler Rowney paints at the time they were revamping their line, with the result that I tested both old and new color formulations, and while I found serious problems with some paints these seemed to be in the old formulation.