Lightfast tests - 5 months and updated June 2016 - that's 3 years of sunlight!

I tried  another method for this series of tests. Rather than painting all the swatches and cutting them up to expose half and protect the other half, I tried very hard to paint two swatches the same. (Nothing like a challenge!) They were pretty close but it is a less exact way of testing. However, it is much easier to manipulate these A4 300gsm watercolour pages than dealing with all the strips I had cut out of the last one.

In this test I just used the colours I most like. I painted all the ones in my palette or that I was considering using in my palette, and the 'extras' that I use occasionally...and a couple that hadn't made it into the previous tests that I was interested in trying. So you will notice there are a few versions of Quin Rose, Phthalo Blue, Phthalo Green, a few different crimsons and a few different Ultramarines etc as I tried one brand against another. I can't make the page any larger so if you are interested in a particular colour and can't read where it is, just ask me. Carmine, which wasn't in my other tests, is top left next to Buff Titanium. It is very similar in hue to W&N Permanent Alizarin which is third from the bottom on the right.

I really wanted to test the primateks too. Jadeite, Green Apatite and Piemontite are three that I love to use in my paintings that are framed and sold and of course I don't want them to fade. Since there isn't much research on these colours I wanted to do some more of my own.

The unlabelled swatches have been in a north facing window in my studio for 5 months so far. The labelled page has been in a dark drawer. Hopefully they will stay there for 2 years. So far I can see no change in any of them.

June 2016 update.
I had a good look at the pages today. They have now been in a north facing window in the harsh Australian sun for over three years. That's a pretty tough call for watercolour. I think a couple of the quinacridone colours have deepened slightly, but or lost some brightness, but I see no fading.
This is a photo since I don't have a large enough scanner. The exposed sheet is on the right.

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.