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!

Lightfast tests - seven months

I painted a number of swatches of as many colours as I could find back in January, cut half away to store in a dark drawer, and placed them in a north facing window for the last 7 months. For the first two months they were inside the window, for the next 5 they were stuck on the outside protected only by the plastic sleeve they were in. There was almost no change at all after the first two months, but a bit more now.

First are photographs of each full sheet. They are too large to fit the whole sheet in the scanner, which was not very thoughtful. Next time - A4 sheets!

Then I will go through each scanned section so you can see what colours faded and which have remained - so far.

Lightfast tests page two seven months

Lightfast tests page one seven months

All Daniel Smith colours. The only change is Aureolin,  6th on the top line. PY40 is known to go grey and you can see this very clearly in the diluted wash on the left. Please don't buy or use this colour.

All Daniel Smith colours. No fading evident after 7 months. Note that Quinacridone coral looks different in the scan, but not in real life. One to watch. 

All Daniel Smith colours. It is difficult to see on this scan, but Rhodonite, second on the top line, has lost some of the rose and greyed slightly. Not recommended. All others are fine so far.

All Daniel Smith colours. Prussian Blue 2nd on the bottom line has changed slightly, and was the first to change when I checked last time. Vivianite Genuine has lost some of the blue hue, so is not recommended. Turquoise Genuine and Sleeping Beauty Turquoise have also changed - the Sleeping Beauty has noticeably greyed. All others seem stable to far.

All Daniel Smith colours. I couldn't see any changes in any of these colours.

All Daniel Smith except the final Old Holland Emerald Green. I couldn't see any changes in any of these.

Most of the top row are Daniel Smith, but you may notice Old Holland Manganese Blue with it's genuine granulating pigment, Art Spectrum Sap Green and Old Holland Flesh tint, which has lost it's red tones so is not stable.
On the bottom row are a set of traditional Japanese pigments. Not that the yellow pigment used in traditional Chinese and Japanese colours is not lightfast so has faded out of the greenm the yellow and even the colour that looks like burnt sienna. The Blue on the right has also changed. 

This final image shows traditional Chinese pigments and the faded yellow is even more obvious. The crimson is probably Alizarin PR83 but I don't know for sure - it has obviously faded.
On the bottom row is a Liquitex Hooker's free, some Art Spectrum colours and some Daler Rowney and lefranc et bourgeois colours that have survived fine so far.

So the main problem colours are PY40 Aureolin, Chinese and Japanese yellow, Prussian Blue, a number of Primatek colours (though MANY are fine) and some crimson reds. Surprisingly, the Opera Rose has not faded as expected...yet. I'll put them back in the window and report back in a few months.

September 2014 update here.