Pieter Mostert

I'd love to hear what you learned. It's unlikely I'll have the opportunity to build a reduction electric kiln in the near future, but I'm still interested.

That last tile looks like a nice fake ash glaze. If your aim is to work with mostly local materials, you could try replace the Whiting with washed wood ash, but that's a whole other rabbit hole.

Mary, Ian Currie published his book on the grid method under a Creative Commons licence, so you can obtain a free electronic copy here.

Curt, I gave a description of the tile in my last post: Silica increases from left to right, and Whiting increases from top to bottom. So it isn't a Currie grid, but the bottom left corner still has the most flux. This is a useful test when your starting glaze doesn't have much clay, but has relatively high alumina. Mary, I fired the tile flat. This is not an example that shows increased fluidity, but I can dig up some if you're interested. I should add that the results of firing flat vs vertically can be fairly different. There's some discussion in the Currie thread about how to get the most info from flat tiles, including some indication of fluidity. By the way, the reason Insight gives different UMF numbers than what you calculated is that it doesn't include iron as a flux. Glazenerd, I haven't heard of iron being involved in shivering. I thought lithium was the main suspect.

That's not always the case. I've done several line blends (at cone 4) where I just added increasing amounts of Whiting to a glaze, resulting in an increase in fluidity, up to a point. If you keep adding Whiting, sooner or later your glaze will become underfired, since Whiting on its own has a pretty high melting point. What sticks out for me from the Insight analysis is the high UMF value of alumina. Have a look at the glazes plotted here (you can refine them to show only cone 6 if you like. I assume that's what you're firing to). There aren't many with Al2O3 as high as 0.99, and most of those that do, have high KNaO. Adding Wollastonite will bring down the Al2O3 UMF value, in addition to increasing to Al2O3 : SiO2 ratio and getting the flux ratio closer to 0.3 : 0.7 (Insight rounds off the numbers). But it might be more informative to do a biaxial test where you add both Whiting and Silica. I've attached a test I did where I increased Silica going left to right, and increased Whiting going top to bottom. I'm not claiming you'll get a stable glaze this way, since I think Matt recommends some boron for cone 6 glazes. But I also think the tests he did on stability didn't involve iron, so I'd be reluctant to extrapolate from them. I should point out that I haven't taken his course, so I could be completely wrong about this.

@Joseph F When I increased the phosphorus in my original glaze (see link in my reply to curt) there was a more abrupt transition between the orange centre and the background colour. (The application on this tile was way too thick, but you get the picture) I think you'd have a similar result if you increased the phosphorus in #9. @High Bridge Pottery Yes, alumina increases up and iron to the right. If the glazes in my grid behave similarly to the ones in the paper I mentioned, then for a given column, the ones lower down start softening at the same time as the ones higher up, but they're more fluid at the top temperature, so the rate at which the viscosity changes with temperature is greater. I think this sudden stiffening of the glaze as it cools is trapping the bubbles. After looking at the paper again, I realised that the conclusion I quoted was for clear glazes only (so alumina can't be too high, as in the glazes I tested). Even here, the data isn't totally convincing.

Sorry for not responding sooner; I've had internet trouble at home over the weekend. @Marcia Selsor Funny, I was looking at that paper on crystalline glazes yesterday. I know that the knowledge of crystalline glazes has advanced alot since then, but it's still a great source of data. I think Ferro 3191 is reasonably similar to the frit I was using, but I'd have to import it. Will try some local options first. @glazenerd I've noticed that in general, my tests on the red earthenware are more likely to pinhole than the ones on white stoneware, so that could well be the result of iron sulphide. But in this case I think the source of most of the gas and subsequent bubbling is from the decomposition of Fe2O3 to FeO. I can't explain why it would happen at 1130C instead of 1230C, which is the usual temperature given, but it still seems to me to be the most likely explanation. @curt The oxides other than iron and alumina are the same as in this recipe. The left-hand column has 5 - 6 % red iron oxide, which translates to 0.09UMF. The reason I think the bubbles may still be forming (or at least expanding and bursting) as the kiln cools, is that the later ones sometimes have different colour developing. For example, in the first close-up, the squashed bubble in the centre is more orange than the two adjacent to it, which must have formed later. This is assuming that the crystals giving the orange colour only form at lower temperatures. I may be (probably am) completely wrong about the effect of changing viscosity on the bubbles. I guess I was thinking something along the lines of what Tony Hansen talks about here: I really need to go back and read the bubble thread again.

Here's what I think may be happening with the bottom left corner. There's a 1914 BSc thesis by Sidney Sewell where he measures the viscosity of a family of glazes as a function of temperature. All glazes have 0.3 K2O, 0.7 CaO with Al2O3 ranging from 0.3 to 1 and SiO2 from 1.8 to 6 UMF. (Well, that was the plan, but the little slacker didn't finish testing all of them in time). He found that for fixed SiO2, decreasing Al2O3 increases the rate at which viscosity changes with temperature. So assuming this is also true for the family of glazes in my grid, as the kiln cools, the ones near the bottom increase in viscosity more rapidly than the ones near the top. So if there's a window of viscosities in which the glaze is too stiff to allow bubbles to grow, but fluid enough to heal over burst bubbles, the low alumina glazes may not be spending enough time in that window. Of course, this doesn't explain why adding iron makes the bubbles heal over. Maybe in this situation it's acting like alumina? I'd be interested in hearing any other explanations.

Here are the results of a Currie-type grid where I again varied iron and alumina. The proportions of everything else are fixed, and the same as my previous test, except that silica is now at 2.48 UMF (previously it was between 2.12 and 2.27). Based on the results of the first test, I reduced the range of iron slightly, to run from 0.09 to 0.22 UMF from left to right (previously 0.07 to 0.22). Alumina goes from 0.22 to 0.34 from bottom to top (previously 0.32 to 0.46) Below is a screenshot from Glazy showing the corner glazes (with Fe2O3 not regarded as a flux). The ones circled in red are my first test, and the ones circled in blue are the second. I kept essentially the same firing cycle (slow firing to 1132C, with a 45 min hold at 950C on the way down. Cone 4 buckled but not flat on the shelf the tiles were on). However this time I fired the white stoneware above the red earthenware tile. This time I also made a little hole in the glaze in each square (based on Curt's suggestion of scoring a 6 o'clock line), but I can't see any evidence of them. On red earthenware: On white stoneware: As in the first test, the tiles get progressively darker as you move from the bottom left to the top right corner. Unlike the first test, which had unhealed bubbles around the top right corner, here the bottom left corner had the most unhealed bubbles, which I don't know how to explain. The sort of stringy, collapsed bubble thing reminds me of another series of iron glazes I'm working on which are also relatively low in alumina, and form ugly blisters when thick. I'd have thought lower alumina would mean lower viscosity, and therefore give the glaze a better chance of healing over. While I was writing this, I thought of a possible explanation, but I'll put it in a separate post. The orange background appears with similar levels of Al2O3 for both tests. It would be interesting to do a grid varying silica and alumina around these points to see how far this region extends. I love the variety of colours and combinations of colours that this tile produced. It's hard to pick a favourite. The area in the middle is the most striking (and is what got me started on this series of tests), but I'm also drawn to the yellow-purpley combination in bottom right corner. In fact, if you look closely, the purpley background has tiny areas of red, which might be encouraged by changing the cooling cycle. The top left glaze and the one to the right of it (#1 and 2) are really interesting, even though I'm not mad about the dirty yellowish - orangey background. I'm intrigued by the blue though, and will probably play with the amount of phosphorus to see how it effects this. But the glaze I keep coming back to is the third from the top in the left-hand column (#11). I really love the subtle blue halos. Unfortunately the frit I used for these tests has been discontinued, and the replacement my supplier claims is the same is slightly different. Time to buy a big batch of frit from a supplier that provides the composition, if only I can find one.

Well, on Glazy you can specify the firing type and type of surface, although not everyone does this. I agree that the other variables are important, and should be included if the goal is for other people to replicate your glaze. On the other hand, I think of recipes I find online as starting points for tests, since most of the time I have to (a) tweak them to work at cone 4 instead of 6, and (b) reformulate them in terms of the ingredients I have available. Also, I'm limited by how fast and how high I can fire, so even if someone gave a firing schedule for cone 4, I might not be able to follow it. Back to Currie grids: I did another grid not that long ago which I'm really happy about. I'll post pictures once I have time to pull my thoughts together. In the same firing I tried out a couple of glazes from my first grid. The photos below are of the one in the middle of the bottom row (tile 33): The one on white stoneware went on thicker, and ran, as you can see. The one on red stoneware has fewer pinholes (not sure if that's due to thickness or claybody), but the spots are smaller. If there's one thing I've learnt from these tests, it's that I need to use more glaze and a bigger container to dip in. I had planned to dip in and out slowly to ensure that the top of the bud vases had a thicker application, but I was too busy trying to fit them into the container to do that. The other oilspot type glazes I tried were less successful. PS: Happy Thanksgiving, for those of you who celebrate it.

I'm going to have to start ditching my test tiles at some point too. I don't fire that often, but they're already staring to clutter up my studio. I'm just a bit wary of relying on photos, since there are some aspects of glazes that are hard to capture in a photo. I use Glazy for recording most of my tests, but a system that naturally caters for recording things like application thickness and firing cycles would be ideal. I know that Derek Au, the creator of Glazy, has thought about this, but at the moment he's busy rewriting the current version, so I don't think it'll happen any time soon. The new version of Glazy will be open-source, though, so if you're up for doing it yourself, this would be a good place to start.

I think this is a good direction to follow. For my regular test tiles I've been using something similar, except that instead of trenches I just have one or two supports on which I can lean the test tile. The supports are in the shape of triangles pointing upwards, if that makes sense. I put a layer a kiln wash over the support, so minor glaze runs don't ruin it. It could do with some improvement though, since my prototypes are pretty crude, and sometimes the tiles stick together. Anyway, for vertical tiles for a Currie grid, you could make 7 slabs, each with a row of 5 rectangle-shaped depressions, like the squares in a Currie grid, into which you could pour the individual glaze tests at the same time that you fill the corresponding squares in the regular grid. This way it'll be easier to keep the tests in the same order as the regular grid, and you won't have to dip them. A potential problem is that the very runny glazes will have a puddle of glaze attached to the bottom, assuming you can remove them from the support, which will make stacking them for storage difficult. But you could just omit the ones around corner C. Actually, it might be better to have 5 slabs with 7 depressions each, otherwise when you put the slabs together flat, the height of the whole thing will be much greater than the width.

Please keep the tiles coming. The quote from Tom Turner that Min shared seems to contradict my suggestion that reduction makes copper a more powerful flux. The only explanation I can think of is that perhaps with the finer mesh sizes of SiC, all the SiC is reduced earlier in the firing, so there's some re-oxidation happening.

Reduction makes iron a more powerful flux, so if the same is true for copper, that would explain why the SiC tiles appear more fluxed. But I don't know if this is the case.