September 20th, 2012

December 2014, by PNH

...In Which We Learn by Resounding Failure

One of the facts of life that I face as a glaze developer is the possibility that a new glaze may appear to be just fine on a test tile, and then reveal some novel and unexpected bad behavior when I put it on an actual piece. Largely because of this, my boss (Doug Humphrey) came up with a wonderful idea that he calls “Teacups for Research”, or TFR. The idea is that before I risk a nice piece on a glaze that is only partly tested, I try it on something that is either slightly flawed in some way or is a shape that I’m not fully satisfied with, rather than something that is cracked or otherwise trashy (which is what I would ordinarily use). I can sell these if they turn out well, and all of the proceeds go to support the nonprofit that I work for. [Further developments about that will appear in a future posting.]

My usual test batch size is 100 grams. That's enough for several test tiles, so I can (for example) check a glaze in both oxidation and reduction, or in several different kilns. When I get something I like, I mix up a smallish “bucket” batch, perhaps 2,000 grams, and I dip a TFR piece in it, along with at least one test tile so I will have a tangible record of what it’s like.


The bucket of the Rutile Blue recipe that I’ve been using for a while is now running low, and the pieces I’ve dipped into it recently have not looked quite right. This happens sometimes; imperfect mixing can cause changes in the composition of a bucket of glaze as dipped pieces remove unequal amounts of the various ingredients, and some glazes just change over time. A few days ago I decided to revamp the formula instead of just mixing up more of the same stuff. (There is nothing wrong with it, but I’m a tweak.) I came up with a number of alternatives, which I tested.

One of them showed a typical problem, which I refer to as “Southwest Red”. It looks about like this:

Not exactly bad, but certainly not blue. Experience suggests that this effect is caused by insufficient Rutile in the recipe; but because Rutile Blue is a complex set of issues it isn’t just a matter of having some particular weight percentage, and it isn’t even just a matter of the mole percentage of TiO2 in the fired glaze. Sometimes they work the first time, and sometimes they don’t. I have had successful Rutile Blues with a little over 4% Rutile in them, and others that needed about 7%. One of the most famous published ones, “Woo Blue” has only 3.6% Rutile, if you include the colorants in the recipe. (If you look at the published recipe, you’ll find that it lists 4% Rutile and 4% Red Iron Oxide; but these are added to the other ingredients, which already total up to 100%. Long story, having to do with the history of ceramic glazes.)

[[If you are a potter, btw, Do Not go by the recipe in John Britt’s book; it has little or nothing to do with Woo Blue, and has no business bearing that name.]]

The borderline between “SW Red” and Rutile Blue for a given recipe can be surprisingly abrupt, btw, which is one of the interesting characters of these glazes for those potters who take the trouble to develop them.

I tried adding another half-percent or so of Rutile to the test batch, but all it did was darken the color a little. I will add yet more, and fire another test.


The real star of the set, however, was this:

It was obviously a keeper, so last night I mixed up 2,000 grams and dipped a test tile and a TFR bowl. I also dipped a second test tile in the original 100-gram batch, as a comparison to reveal any possible mixing errors in the new batch, and also to reveal any potential sensitivity to changes in the firing conditions. (It is difficult to be too careful or rigorous in the pursuit of righteous glazes.) I fired the kiln in the evening, and took the test tiles out late at night, leaving the TFR bowl to cool more slowly. Here is the bowl as it came out of the kiln this morning:

This is so ridiculous that I had to laugh. (“Bad glaze; no donut!”) I’ve never ever had anything dunt like this, despite the fact that I routinely develop glazes that have far lower thermal expansion coefficients than those of most potters. In fact, I’ve only ever had one other piece dunt, as far as I can recall. That one came apart in many small chunks while I was carrying it away from the kiln; but I think the problem with it was related to the body clay rather than the glaze. This one is clearly and obviously caused by the glaze. [Apologies, btw, for the lousy color balance; I took the photo with my phone, under fluorescent lights.]

Interestingly enough, the test tiles have the same creamy look as the glaze on the bowl, and during the night some of the glaze spalled off the tile from the original 100-gram batch (though not until after I dropped and broke it). None of the glaze has spalled off the test tile from the “bucket” batch yet, but I’m not betting on it remaining intact.

I have revised the recipe to increase the thermal expansion a bit, and will be testing one or two new versions as time and tide permit. We’ll see whether they’re worth posting about.