Sunday 14 September 2014

Next lightfast test results - 19 months

My watercolour lightfast tests on the outside
of a north-facing glass window.
I painted a number of watercolour swatches in January 2013, cut them in half, and placed half on the inside of a north-facing window for two months. I checked the samples and noted any changes. Then I put the exposed sections in plastic sleeves and stuck them to the outside of a north facing window, and scanned them in October after a total of seven months. I reported on them here.

I returned them to their north facing window, which is a pretty harsh environment for watercolour, and have just scanned them. Note I am in Sydney, Australia, so that is the same as a south-facing window in the Northern Hemisphere, though Australian sun is very harsh. The biggest changes happened during the first seven months. Not much further change has occurred.

Sheet 1
The first row is all fine except Aureolin, PY40, which should not be used as a watercolour. Notice the light wash is now creamy and not yellow at all. The mass-tone is dull and greyed.
I couldn't see any change in the bottom row of the first sheet. All Daniel Smith unless marked otherwise.

Sheet 2
The top row of sheet 2 all look fine :-)
The bottom row of sheet 2 also looks fine. I expected a possible change in the last one, Permanent Red Deep, but can't see any. As before, Quinacridone Rose looks different on the screen but not in real life. This is true of Quinacridone Coral and Quinacridone Red as well, so I guess there is some change but not one we see, or a slight difference in the angle of the sample in the scanner.

Sheet 3
The top row of sheet 3 had Opera Rose as my control as I expected this to fade, just as I expected Aureolin to fade above and the crimsons to fade below. There is only the slightest loss of brightness in the original though so far. The scanned colour looks much more dull as the scanner doesn't pick up the florescent dye. Rhodonite has continued to dull, but so has the sample that was in a draw unexposed to light as apparently it is exposure to oxygen that changes the colour of rhodonite.
On the bottom row, all samples seem unchanged.

Sheet 4
All the samples on the top row seem unchanged.
On the bottom row, Prussian blue PB27 has faded, as noted after the 2 month and 7 month scans. Vivianite has lost some more of its 'blue' hue. Kingman Island Turquoise Genuine has faded slightly and Sleeping Beauty Turquoise Genuine has dulled down further to a grey-green from its original Turquoise. The other samples appear unchanged.

Sheet 5
I can see no changes in any of the samples in the first row. I am delighted that these wonderful Primatek green pigments have held up. 
In the bottom row, all samples are unchanged. :-)

Sheet 6
All of the top row samples seem unchanged including the Primateks Hemetite Burnt Scarlet, Garnet and Piemontite Genuine. I love all these granulating earthy colours :-)
The samples in the bottom row also appear unchanged. 

Sheet 7
In the top row, there is a possible slight yellowing of Manganese Blue PB33 Old Holland but it is very slight. Flesh Tint, Old Holland, with is a mix of a number of pigments, has really faded out.
On the bottom row, a number of faded colours are obvious. These are largely unknown Japanese pigments bought in Tokyo as a fun set for sketchbook work couple of years ago. At a guess, the yellow is genuine Gamboge, which has faded. The yellow has also faded out of what was a green on the left. The vermilion is fine but the alizarin crimson has also faded. The burnt sienna must have had some gamboge in it as well, which has faded leaving an indian red hue. On the right, the prussian hue has also faded. Thank goodness these were not finished paintings! These colours may be OK in a journal where they are protected from light, but should not be used in work to be framed.

Sheet 8
The final sheet has more traditional Chinese pigments that were bought as pellets many years ago in Hong Kong. On the bottom row you can see that once again the genuine gamboge yellow has faded badly, as has the indigo. The vermilion red has faded a bit, the alizarin crimson a lot and the bright red has faded in the lighter wash. Even the burnt sienna has faded in mass-tone. I don't know what these pigment numbers were but it is interesting to see what faded watercolour looks like! It is no fun in a finished painting.

 On the bottom row, the watercolours by Liquitex, Art Spectrum, Daler Rowney and Lefranc et B all appear unchanged.

It is a very valuable exercise to do a light-fast test of the colours you use as the manufacturer's labels may not always be accurate. Keep away from PY40, PR83 and, it seems, PB27.

September 2014 update: My other lightfast tests, using the colours I have chosen for my palette and other extras, were reported here. They have continued to hang inside my window and I see no changed in any of these colours. Whew!


  1. PB27 not lightfast? That's really disappointing. My tubes all say it should be very lightfast. I think I had better do my own testing. Prussian blue is one of my favorite blues...

    1. I was really surprised by Prussian Blue. I had not heard or read anything to expect a problem. To be fair, it was exposed in very extreme conditions - no one puts watercolours outdoors - but it had changed after only two months while inside, so it is definitely one to watch. I would suggest you do further tests of the brand you use as the pigment may have come from a different source and may be fine.

    2. Hi, am re-reading this again. Have you looked at Bruce MacEvoy's website Handprint? There is a lot of technical information there. He was satisfied with his tests of PB15 (Phthalocynanine Blue) so I am assuming the un lightfast part of the PB27 must be what they add to PB15 to make it darker. I am about to do some of my own tests on the colours I use.

    3. Sorry meant to give you the link to that other website

    4. I did a bit of research and found this regarding Prussian Blue's Lightfastness:

      "Prussian Blue pigment has a chemical reaction to UV light exposure. It is different than the standard fading that happens in other pigments. This is caused by the oxidation of ferrous ferrocyanide salts. Sun bleaches the FeIII (Iron) which requires time to re-oxidize away from light. Because it is possible to regain color intensity over time in shade, manufacturers have lazily marked it as LFI."

      You can read more here:

      My thoughts are, who is hanging watercolour paintings in the sun! :P

  2. Thanks Jane. This independent testing is invaluable, are you planning to test any of the newer colours on your palette such as the pyrrol crimson?

    1. I have done other tests of my palette colours and they are all fine. No visible change in well over a year in a north facing window. I'll scan those too though they are difficult to read.

    2. Here's the link to the previous post so you can see all the colours

  3. Eeek! Thanks for the warning on the Prussian blue and the sleeping beauty turquoise. I love both of them and had just switched them into my palette.

    1. You can mix a Prussian blue hue with Phthalo Blue GS and a little of a warm red. I was also very surprised about the Prussian Blue - I've only ever read that PB27 is a reliable pigment. Not in the harsh Aussie sun perhaps?

  4. Hi Jane , I like your colour samples. I was talking to Mark at Seniorart in Malvern in Melbourne as he is a stickler for quality materials and proper methods. I saw the sample of green apatite that had been up on display at the shop and had faded really much for ASTM 1 rating. You mentioned handprint by Bruce MacEvoy which is a great site.
    I have also come across one called where they list all pigments with their ratings and even which manufacturer makes them as oils, w/c, acrylic etc. It also lists any quirks the pigment may have.I think the Primateks may have been rushed out onto the market a bit too quickly. For more good stuff on colour there is Michael Briggs who has a very common sense site for colour. He has a colour lab in Sydney and teaches at the Julian Ashton school and the university as well. Another colour resource is from Don Jusko at with a big site of interesting findings. He has a 36 colour wheel where the paints stated from various manufacturers do line up as "best fit"
    complementaries.He even has colour pencils as well on his site.

    1. Great resources pw6 - thank you. I'll check them out :-)

      I haven't done my own lightfast tests with the Primateks. I tend to use them in sketchbooks rather than 'exhibition' works. Aussie sunshine is vicious - much more so than some other countries - and that is a factor we need to be aware of. I remember hearing that when some stained glass windows were shipped out to Sydney for the Cathedral, they were installed and looked...white. Colourless as our dazzling sunshine bleached them out. They had to be redone.

  5. Wow! thank you for doing these, I have about 4 different colors myself so I need to try doing one of these too! Did you ever try testing them after putting a gloss over them?

    1. I haven't tried putting a gloss over them and doing further tests. Glass can act as a bit of a protection, especially UV glass. I know of watercolourists who work on canvas and paint a varnish over their watercolours rather than framing them under glass. If it is UV blocking it's possible they would test better.

      I have sprayed UV varnish over my own watercolours on canvas, and one hangs on a wall that gets hit by light from a north-facing window (not recommended!) but if it is fading I haven't noticed....yet!

  6. These are great, thank you so much!
    I was going to order Quin Rose, but accidentally ordered Quin Pink instead. I was really worried as it said it wasn't as lightfast. But this test reassures me. Is this a color to watch, or do you think it is safe to use? I hope I didn't waste my money :D

    1. I think you'll be fine with Quin Pink. It is very similar to Quin Rose but perhaps more on the blue side. Lovely colour, not to be confused with Opera pink or Opera rose, which is fugitive.

    2. Thank you so much! I am so relieved!

  7. Jane thanks. For the colors that are not light fast and only stay in a journal, would they fade as well? It’s a shame to not use such beautiful colors like sleeping beauty turquoise. I’m not one to start selling my paintings. I just sketch and paint with them.

  8. Hi Jane....what do you think about Daniel Smith Opera, SB turq and Prussian behind UV glass? Would the paintings be worthy of sale?

  9. Jane, bless you for doing these tests and sharing your results. It's so helpful. Interesting I just saw a video saying Anthraquinoid Red is not ligtfast, especially in Moonglow, and Handprint also warns. But it seems fine here, so I feel better about using it now. You can do much better testing where you are than I can, so I really appreciate this!