Many watercolor versions of violet or mauve manufactured decades ago included fugitive pigments. Some of these — the basic blue violets (PV1 and PV10) still used in specialty printers inks and layout colors — are not suitable for artistic use. They produce a luscious dark, rich color that sadly starts to fade as soon as the pigment has dried. The fading is so complete that in some brands a tint will leave nothing but white paper.

AVOID. These paints are only useful if you want to do a concept art environmental erasure of your own work. Substitutions. The most saturated and lightfast blue violets are obtained with a mixture of ultramarine blue (PB29) and quinacridone magenta (PR122).

lightfastness test sample

unexposed (top); exposed 800+ hours (bottom)

A recent paint from Holbein, combining lightfast ultramarine violet (PV15) with fugitive PV7. The CIECAM J,a,b values for bright violet (PV7+PV15) are: 27, 65, -32, with chroma of 73 (estimated hue purity of 68) and a hue angle of 334.

AVOID. A very intense, dark valued reddish violet, wonderful for decorative work or paintings that are explicitly intended to be photographically reproduced but not preserved for more than a few months. Substitutions. The most saturated and lightfast blue violets are obtained with a mixture of ultramarine blue (PB29) and quinacridone magenta (PR122).

TOP 40 PIGMENT  Cobalt violet PV14 (often labeled "cobalt violet deep") is a very lightfast, semitransparent, nonstaining, moderately dark valued, moderately dull violet to red violet pigment, available from 4 registered pigment manufacturers worldwide. The ASTM (1999) rates its lightfastness in watercolors as "excellent" (I) and my own tests consistently concur. PV14 undergoes a moderate drying shift, lightening by about 17%; the hue typically shifts toward red in tints. The average CIECAM J,a,b values for cobalt violet deep (PV14) are: 46, 33, -32, with chroma of 47 (estimated hue purity of 47) and a hue angle of 316.

PV14 is one of the most expensive pigments used in artist grade watercolors, and is frequently mimicked with less expensive pigments, especially in student grade paints. The Maimeri cobalt violet and Daniel Smith cobalt blue violet (listed below) are less lightfast mixtures of cobalt blue (PB28) and quinacridone rose. Many art supply manufacturers do not offer PV14 at all, suggesting it is unpopular in the retail market or not profitable to sell. Among artists listed in the section on palette paintings, only Charles LeClair recommends it.

There is noticeable variation in the color and texture (granulation) of this pigment, with one cluster of paints at around hue angle 330 (reddish), and a second at hue angle below 320 (bluish). Rowney Artists cobalt magenta is the reddest and most intense of the paints listed here: a lovely, semitransparent middle violet with a good granular quality; in masstone it has an interesting duotone (magenta/violet) appearance. The recently revamped Winsor & Newton cobalt violet is similarly reddish. Blockx cobalt violet has a similar hue and chroma, but is slightly lighter valued, coarsely granular, and as inert as sand in wet applications. At the other extreme, Holbein cobalt violet light is the bluest in hue and one of my favorite cobalt pigments: lighter valued, more intense and more opaque than other brands, with a more homogenous, glowing purple color. Daniel Smith is in the middle of the hue range, but darker valued and less saturated than the others; the Utrecht paint is the same hue but is thinly mixed and even more unsaturated.

Although some artists disparage this pigment (Michael Wilcox calls it "gummy and weak"), genuine, high quality cobalt violet is a spectacular paint in broad wash applications — morning skies and magnified florals — and evocative in flesh tone shadows. The "red" shades offered by Rowney, Blockx and Winsor & Newton are effective as the pink component in caucasian flesh tones. Like other granulating pigments (such as viridian), the typical pigment contains a very broad range of particle sizes; heavier wash applications may show a whitish overcoat, especially in paper depressions, produced because the smaller (less saturated) pigment particles are the last to sink out of solution. The hue is readily mixed from ultramarine blue (PB29) and quinacridone rose (PV19), but the poetic granular quality and crystalline color permanence are unique and well worth exploring. See also the section on cobalt pigments.

TOP 40 PIGMENT   Ultramarine violet PV15 is a very lightfast, semitransparent, moderately staining, dark valued, moderately dull violet to very dark valued, moderately intense blue violet pigment, available from 6 registered pigment manufacturers worldwide. The ASTM (1999) rates its lightfastness in watercolors as "excellent" (I), but I found some brands began to opacify and whiten slightly, with a very small resulting color change, after a month of full sun exposure. In watercolors PV15 presents a very small drying shift, holding its value and losing only 10% saturation. The average CIECAM J,a,b values for ultramarine violet [red shade] (PV15) are: 36, 21, -37, with chroma of 42 (estimated hue purity of 45) and a hue angle of 299; for ultramarine violet [blue shade] (PV15) they are: 22, -4, -57, chroma 57 (estimated hue purity 61) and hue angle 266.

There are significant color differences across various brands of this paint. A red violet version is available under the same color index name and chemical description; most manufacturers only provide the blue hue. (Hilary Page's quirk of adding "R" or "B" to the color index name has no sanction from either the SDC or the manufacturers.) Both the red violet and blue violet hues are manufactured as a chemical modification of ordinary ultramarine blue, which is mixed with sal ammoniac (for the blue hue) or dry hydrochloric acid (for the red) and heated to 150° C for several hours. Although technically PV15 is any ultramarine that has been chemically treated as described, all ultramarine violets contain significant amounts of unaltered ultramarine blue (PB29).

Winsor & Newton ultramarine violet is the lightest valued, most intense and most characteristic "blue" violet of the brands tested here; it is less active than other brands in wet applications. The Daniel Smith and Rembrandt ultramarine violets are similar in value range and chroma; the MaimeriBlu is the reddest of the violet shades with good saturation. The Rowney Artists and Old Holland were the least saturated brands tested here. In my paint tests, I discovered Lukas cobalt violet was actually an ultramarine violet that whitened under sunlight exposure; Lukas replied to me that they use genuine cobalt violet only in their dry pan formulation (!). The M. Graham ultramarine violet and Blockx ultramarine violet contain substantially more unadulterated ultramarine, producing a color that appears significantly bluer, darker and more saturated: in tints these paints shift toward red to nearly match the traditional violet color. (Actually, the hue angles of all the ultramarine violet paints are very similar; the "bluer" shade arises in part because the color is both darker and more saturated.)

Daniel Smith ultramarine red is the only commercial source I know of for the PV15 red shade, a granular and soft pinkish violet, close in hue to a diluted, dull manganese violet.

Ultramarine violet does not mix well with yellows, producing to my eye a lifeless gray. (Stephen Quiller recommends it as the mixing complement to cadmium lemon, but see the chart at mixing complementary colors.) The color can be easily reproduced by other mixtures — ultramarine blue with a touch of quinacridone rose, for example. Good as a blue violet for a muted palette, and sometimes useful for subtle gray violet shading in portrait work or to capture the colors and textures of twilight skies. See also the section on sulfur pigments.

TOP 40 PIGMENT   Manganese violet PV16 is a very lightfast, semitransparent, lightly staining, dark valued, moderately dull red violet pigment, available from 4 pigment manufacturers worldwide. The ASTM (1999) rates its lightfastness in watercolors as "excellent" (I) and my 2004 tests agree. In watercolors PV16 undergoes a moderate drying shift, lightening and losing saturation. The average CIECAM J,a,b values for manganese violet (PV16) are: 31, 39, -21, with chroma of 44 (estimated hue purity of 40) and a hue angle of 331.

The PV16 pigment is very consistent across paint manufacturers. Daniel Smith manganese violet was a principal source for this pigment: slightly bluer and lighter valued than other brands, and blossoming more when rewetted. The Winsor & Newton permanent mauve, previously available only in dry pans, is now (2005) available in a tube formulation; it has a slightly less saturated, and darker color that lightens into effective tints and lifts almost completely to produce cutouts or sculptured edge effects. The MaimeriBlu mineral violet is more staining and very opaque. Hilary Page noted discoloration in her sample of Old Holland, which may not have been a single pigment paint; my swatch began life as a dull, scabby purple, not at all pleasant to look at, and the vehicle (or pigment?) discolored to a browish cast after about a month of sunlight exposure. Two thumbs down!

PV16 is the most lightfast balanced purple pigment available in watercolors, and its recent addition to the DaVinci and Winsor & Newton is a welcome development. While it has an assertive and distinctive pigment personality, this becomes less conspicuous when the paint is part of a shadow mixture. It is attractive both in full strength and wash applications, but for the most characteristic color appearance it must be applied with confident, juicy brushstrokes and left to dry without fussing or retouching. It is especially good in floral painting, both as a muted floral color and to add texture and body to browns mixed with deep yellow or orange paints. The same hue can be mixed from ultramarine blue (PB29) with quinacridone rose (PV19) or quinacridone violet (PV19), depending on whether you want more saturation or darker values. In many respects, PV16 handles like thioindigo violet (PR88) and can play a similar role in landscape or botanical palettes. See also the section on manganese pigments.

TOP 40 PIGMENT   Dioxazine violet PV23 (and its sister form, PV37) is a lightfast to impermanent, semitransparent, heavily staining, very dark valued, dull violet pigment, available from about 30 pigment manufacturers worldwide for use in plastics, inks, paints and foods. The hue is similar to (but much darker than) ultramarine violet or cobalt violet deep. Its tinting strength is very high, on a par with phthalo green (PG7) and phthalo blue (PB15). In watercolors PV23 and PV37 show a very large drying shift, lightening by 38% and losing more than 20% saturation. The average CIECAM J,a,b values for dioxazine purple (PV23) are: 20, 15, -27, with chroma of 31 (estimated hue purity of 34) and a hue angle of 299.

Because it is so dark, the color appearance of this pigment is very similar across paint brands, though clear differences do appear in the paint undertones (tints), and there are very large brand differences in the paint lightfastness (see below). The most robust paints in my tests were manufactured by Daniel Smith (which gives the pigment a lightfastness rating of "1, excellent"), Winsor & Newton and M. Graham. The Daniel Smith carbazole violet is very dark and concentrated, producing a slightly grayed violet in tints, perfect for rendering shadows, and it appeared to be among the most lightfast paints. Winsor & Newton winsor violet (dioxazine) is more intense and slightly less staining than other brands, and dilutes down to lovely glowing tints. The M. Graham paint, recently discontinued (see under PV37) is dense and dark, but seemed slightly more susceptible to fade in tints.

There were some glaring lightfastness failures in other brands of watercolor paint, such as Utrecht, Schmincke, Rowney Artists and MaimeriBlu. The very expensive dioxazine pigment is so diluted in the Utrecht paint that it cannot produce a dark value even at full strength, fades noticeably, and bronzes heavily in masstone. Other brands should not be used until they change the manufacturer suppliers of their pigments.

There is considerable confusion about the lightfastness of this pigment in watercolors. Let me set the record straight. The ASTM (in D5067-99, Standard Specification for Artists' Watercolor Paints) gives separate lightfastness ratings for a "red shade" ("fair" [III]) and a "blue shade" ("poor" [IV]) of dioxazine violet. Michael Wilcox's paint guide merely repeats these ratings without doing any lightfastness tests of his own. However, Hilary Page's paint guide, which is based on actual lightfastness tests, is incorrect to say that "the blue shade does not seem to exist". In fact, a range of color variations are produced from different crystal forms of the pigment and from different manufacturing methods to refine and grind it. The problem is that pigment manufacturers seem to assign hue designations at their whim: among the SDC Colour Index pigment descriptions supplied by the pigment manufacturers, one finds the designations "bluest shade," "blue shade," "reddish shade," "red shade" and even "yellowish shade" (!). So without knowing which pigment manufacturer made the pigments that were tested by the ASTM, and on what grounds that manufacturer described the pigment color (colorimetric values? manufacturing methods?), the ASTM lightfastness ratings are uninterpretable.

dioxazine violet lightfastness samples (2004)

after 800+ hours of sunlight exposure, brands show large differences in fading or discoloration: (left to right) Daniel Smith, M. Graham (PV23), M. Graham (PV37), Winsor & Newton, Schmincke, Rowney Artists, MaimeriBlu, Utrecht

Manufacturer and my own 2004 lightfastness tests indicate dioxazine violet actually has better lightfastness than reported by the ASTM, easily reaching "very good" (II) lightfastness in the best brands reported here. The lightfastness reported as 6,7 or 7,7 is equal to or better than the lightfastness I observed in naphthol red, PR170, perylene scarlet, PR149, quinacridone pyrrolidone, PR N/A, perinone orange, PO43, and even some brands of quinacridone rose, PV19, all pigments that are considered acceptable for artistic use. Overall, art materials manufacturers are clearly using pigments from different pigment suppliers, and the pessimistic ASTM ratings are either unrepresentative or flawed.

Examined overall, however, PV23 is a pigment that (1) is not transparently labeled "blue" or "red" by manufacturers; (2) produces highly variable lightfastness test results; (3) may produce unreliable lightfestness test results within a single pigment, given the wide range in lightfastness test results across different grades; and (4) therefore presents a difficult sourcing problem for paint manufacturers, who must themselves do rigorous testing in order to be confident in the quality of the pigment and pigment manufacturer they are dealing with.

PV23 is a good choice for color point 6 on the color wheel, is useful for reducing the saturation of paints on both the warm and cool sides of the color wheel, and produces potent dark shades when mixed with the likes of phthalo green (PG7) or quinacridone violet (PV19). It is probably too strident or strongly tinting to make an effective shadow color, and I feel it is untrustworthy in tints.

CAUTION. This is a pigment that puts responsibility squarely on the artist. Many artists may conclude that there is reason to reject it out of hand, given the problems described above. Others may choose to use it because it is acceptably lightfast in the best brands and in applications near full strength, or may use it in limited applications such as field sketching. However, it is clearly reckless to use it in significant works without doing your own lightfastness tests on the brand you use, and it is risky to rely on it in tints or in light valued mixtures with green, yellow or red paints. Substitutions. PV23 is very close in hue, saturation and value to indanthrone blue (PB60), which is consistently more lightfast and is a better hue match to the blue chromaticity of skylight that illuminates outdoor shadows. The same hue at similar lightness and saturation can be mixed from ultramarine blue (PB29) with quinacridone violet (PV19), which I recommend you use if you are concerned about PV23's lightfastness or find its aggressive staining hard to work with. The many purple convenience mixtures made with ultramarine blue and quinacridone rose seem in my tests to have about the same lightfastness as dioxazine violet, and therefore are not really practical substitutions. See also the section on dioxazine pigments.

Dioxazine violet PV37 is a lightfast, semitransparent, heavily staining, very dark valued, dull violet pigment. — In 2000, M. Graham switched from PV23 to to the chemically more complex but reportedly more lightfast form of dioxazine, PV37, in its formulation of dioxazine purple. However I didn't find a significant difference in the lightfastness of the newer pigment in comparison to PV23: both began to fade in tints at about 550 hours of sunlight exposure (BWS high 6), and remained solid in masstone well into BWS 7. The paint is dark and concentrated with very good tinting strength, producing a slightly grayer violet in tints. Many artists recommend dioxazine violet as a foundation shadow color, glazing over the purple with paints that describe the surface colors of objects; I suggest you try indanthrone blue (PB60) for that purpose. For more information on dioxazine pigments, see under PV23.

Crystal violet PV39 sounds like a drug, and it is. An intoxicating blue violet color, it loses most of its brilliance within a few weeks of daily sunlight exposure. The average CIECAM J,a,b values for crystal violet (PV39) are: 20, 8, -49, with chroma of 49 (estimated hue purity of 54) and a hue angle of 279.

AVOID. The most saturated and lightfast blue violets are obtained with a mixture of ultramarine blue (PB29) and quinacridone magenta (PR122).

lightfastness test sample

unexposed (top); exposed 800+ hours (bottom)

 Cobalt violet PV49 is a very lightfast, semitransparent, nonstaining, granulating, moderately dark valued, moderately intense red violet pigment, available from just 2 pigment manufacturers worldwide. The color is a dull shade of the optimal "red" subtractive primary; the handling characteristics are indistinguishable from its sibling, PV14). The average CIECAM J,a,b values for cobalt violet light (PV49) are: 50, 52, -26, with chroma of 58 (estimated hue purity of 50) and a hue angle of 334.

Daniel Smith cobalt violet is a lovely granular pigment, a pinkish violet with darker violet accents; it blossoms when rewetted, and lifts completely. The Utrecht paint granulates equally well, but with a slightly lighter and more homogenous value and a slightly redder hue; it also seems somewhat more concentrated but is slightly less saturated. — PV49 has poor tinting strength but an attractive, unique hue and granulation. I find it works best when used for skies or sandy, earthy landscapes; it can be used either in its natural hue or mixed with near transparent pigments such as the iron oxides, quinacridones or phthalocyanines to give a different background tint. See also the section on cobalt pigments.

 Thioindigo violet PR88 is a marginally lightfast, semitransparent, heavily staining, very dark valued, moderately dull magenta pigment, offered by only 2 pigment manufacturers worldwide. It is one of the many red lake pigments. The ASTM (1999) rates its lightfastness in watercolors as "very good" (II), with the caution that "pigments described as thioindigoids have varying degrees of lightfastness," depending on the manufacturer of the pigment used in the paint; my 2004 tests of the paints listed here were also "very good" (II).

thioindigo violet lightfastness samples (2004)

after 800+ hours of sunlight exposure: (left to right) Daniel Smith, Winsor & Newton,
Grumbacher, MaimeriBlu

In watercolors, PR88 undergoes a moderately large drying shift, lightening and losing saturation by 20%. The average CIECAM J,a,b values for thioindigo violet (PR88) are: 29, 50, -2, with chroma of 50 (estimated hue purity of 47) and a hue angle of 358.

This is a fairly consistent pigment across manufacturers. The Winsor & Newton paint has been discontinued and the Grumbacher line has lapsed. Of the two remaining brands, the Maimeri paint is more transparent and is much more active wet in wet. The Daniel Smith permanent violet is darker valued at full strength and is so unsaturated it appears maroon, but it dilutes out to a pleasing hue and microscopic pigment texture and is somewhat more lightfast. PR88 mixes to some very attractive dusky violets with ultramarine, cobalt or prussian blue, as well as lovely browns with orange and yellow paints. All brands backrun readily.

AVOID. I don't feel the pigment is reliable enough to provide lightfast mixtures, and there are more lightfast pigments with very similar color appearance. Substitutions: I suggest quinacridone violet (PV19) or perylene violet (PV29) instead. Both paints are as dark as thioindigo violet, and both have substantially better lightfastness. The perylene violet is warmer and produces less intense brown mixtures.

Many artists mix their violets from ultramarine blue (PB29), cobalt blue (PB28) or ultramarine violet (PV15) and a rose or violet quinacridone (PV19). Well, here are the same purples, premixed for you as convenience paints you can use straight out of the tube. The problem with these convenience mixtures is that they appear in some cases to be less lightfast than the dioxazine violet they are designed to replace!

lightfastness samples of convenience purple paints (2004)

after 800+ hours of sunlight exposure, the samples show significant fading or discoloration: (left to right) Daniel Smith cobalt blue violet, Daniel Smith rose of ultramarine, Maimeri, Utrecht

These representative convenience purples, all mixed from ultramarine blue and quinacridone rose, show substantial discoloration or fading. Daniel Smith rose of ultramarine offers a less staining and lighter valued alternative to dioxazine violet (PV23). The Utrecht permanent violet is mixed with dioxazine, which shifts slightly in relation to the more permanent ultramarine, making the color combination unstable.

AVOID. As shown above, I found across several brands that these premixed purples are not as transparent or lightfast as dioxazine violet (PV23) or ultramarine violet (PV15). If you require a basic purple paint or purple mixture, then you will do much better to use manganese violet (PV16) as the basic pigment. If you mix purple colors yourself, for example with ultramarine blue (PB29) or indanthrone blue (PB60) and quinacridone violet (PV19), I'd suggest you can expect no better lightfastness than the commercial paints provide.

KEY TO THE PAINT RATINGS. Summarized as numbers: Tr = Transparency: 0 (very opaque) to 4 (transparent) - St = Staining: 0 (nonstaining) to 4 (heavily staining) - VR = Value Range: the value of the masstone color subtracted from the value of white paper, in steps of a 100 step value scale - Gr = Granulation: 0 (liquid texture) to 4 (granular) - Bl = Blossom: 0 (no blossom) to 4 (strong blossom) - Df = Diffusion: 0 (inert) to 4 (very active diffusion) - HA = Hue Angle in degrees of the CIELAB a*b* plane - HS = Hue Shift as the undertone hue angle minus the masstone hue angle, in degrees of the CIELAB a*b* plane - Lf = Lightfastness: 1 (very fugitive) to 8 (very lightfast) for paint in tint,full strength - Mentioned in pigment notes: Chroma: For the masstone paint on white watercolor paper. - Drying Shift: Change in masstone color appearance from a glistening wet to completely dry paint swatch, in units of lightness, chroma and hue angle in CIELAB. For more information see What the Ratings Mean.

Last revised 07.I.2015 • © 2015 Bruce MacEvoy