In my quest for the perfect crystalline glaze porcelain body, I have tested a large number of materials. One problem I have is with the silica I use as a filler. Although it is advertised as being ground to pass 300s mesh, there are a frustrating number of oversized particles that make the tedious process of sieving even more time consuming.
A new source looked promising, it seemed more finely ground and as an added bonus was much whiter in colour than my usual one. There was one problem. On the 25kg bag the word ‘Cristobalite’ appeared. I ignored it – the bag was from Belgium, clearly there was a language difference and it didn’t mean what is does in English, besides, who would be crazy enough to sell cristobalite as a filler for a porcelain body? That was the first mistake: ignoring the obvious.
This was to be a casting body. I had mixed it many times before and the specific gravity always ended up at 1.83. This time it was lower. Odd, I must have put in more water this time. The second mistake: ignoring additional evidence.
The final piece of the puzzle was dramatic and impossible to ignore. The test piece came out of the kiln. There was a thrill – the crystals were exceptionally vivid. Putting the piece aside, I continued to unload the rest of the pieces, thankfully made from my more orthodox porcelain recipe. There was a loud bang and instead of the porcelain piece with nice crystals, there was a pile of broken porcelain fragments.
The explanation was now obvious. The bag really did contain cristobalite. The reason that the slip had a lower than expected specific gravity is because cristobalite is less dense than quartz, 2334 
compared with 2648 for alpha-quartz. When I did the calculation, the predicted difference in slip density was the figure I had observed.
The reason the piece had failed is that the cristobalite had remained, or at least partially remained, as cristobalite throughout the firing. Cristobalite has a higher thermal expansion coefficient than quartz and that was enough to produce an expansion mismatch between the porcelain and the glaze sufficiently great to cause it to fail spectacularly.
One lesson here is not to use cristobalite as a filler in a clay body, but that part is obvious. The real lesson is never to ignore observations, especially when they are at odds with expectation. It is so easy to fall into the trap of dismissing unexpected results.