A couple years ago I wrote about how there are a few  people working on a new technology here that uses multiple lasers to analyze the chemical components of pretty much anything, and one of the things they could do is to test what’s in Yixing clay. I’ve been working with these folks since then to help them come up with ways that will have practical applications for people who use Yixing pots. They have also improved the technique they’re using as well as the sensitivity of the data, and I thought I can write an update on some of the things they’ve done recently with a few pots of mine.
Basically, I gave them four pots to test, without telling them previously what they were. The idea was to see if the analysis might yield any data that is interesting, and if so, what that might be. The pots I gave them were 1) a regular yixing pot I bought many years ago from a Shanghai tea market, 2) an antique that is an export to Japan, 3) a Japanese tokoname pot, and 4) a fake yixing (it’s so obvious it’s fake it’s pretty painful) but made in the style of a yixing pot, complete with “Zhongguo yixing” seal at the bottom, but the clay is obviously off, also bought from Japan. The experimenters also added one of their own, called “cheap” in the data you see below.
The way they do this analysis is to basically place the teapots on their testing platform, and do a series of laser shots to vaporize a little tiny bit of the teapot, then the second laser does a spectrum analysis of the puff that is created. It looks like this:
So the results of the tests on the five teapots, visualized for simplicity, is as follows:
The X and Y axis are simplifications of the actual data, of which they have about 51k datapoints for total for the five teapots. You can see that the yellow (Tokoname) and orange (fake yixing) almost completely overlap – and in fact if you go find the underlying data shows that there is basically no chance this is happening by accident. In other words, the fake yixing is probably of Japanese origin using Japanese clay that is substantially similar to tokoname clay.
This sort of thing is quite interesting, because if we can build up a database of teapots, then it’s possible to actually use the database to try to authenticate teapots, perhaps even periodize them if we have enough data. That’s for the long run, but it’s quite enticing a prospect.
In the short term though, there are other things that this can do – for example, testing for heavy metals. None of the teapots sampled had any traces of heavy metals, such as lead. Since the tests are conducted on multiple locations on the teapots, it is quite reliable and not down to a single datapoint. That in and of itself could be of interesting application as well, considering how so many people are worried about what’s in their teapots.
11 responses so far ↓
miig // November 2, 2015 at 10:12 am |
wow, fascinating!
Thank you for this great idea and the sharing of the results. It’s going to be very interesting to see what’s coming next… the coating of fake pots with genuine clay to fool the normal laser testing perhaps? 🙂
MarshalN // November 3, 2015 at 1:39 am |
If it’s a very thin film of coating then it gets burned off in the initial shot or two they do to clean the surface. If it’s deeper then that it doesn’t really matter, does it? That’s all your tea/water is coming into contact with anyway. In some ways what’s in the middle of the clay has very little bearing on how the teapot brews (heat retention is probably the only thing that may differ).
Cwyn // November 3, 2015 at 12:03 am |
Nice article. Really wanting these scales and data points defined. Looks like some kind of cluster analysis? I get that the variables are large, but the number of points here on the unlabeled scales are few enough to be labeled. Did they provide this info?
MarshalN // November 3, 2015 at 1:38 am |
They are not definable because the x and y axes here do not represent anything in particular – they are aggregates of data after some fancy math they used to aggregate the many data points into a few dots. What they are doing here is to show a visualization that highlight the differences (if any) between the samples. This is not your simple x-y graph.
Leopold Kong // August 20, 2020 at 8:12 am |
Looks like principal component analysis. Would love to see an update of this analysis.
Xiao Bai // November 3, 2015 at 6:44 pm |
Very nice article. I think this is a really exciting project and very much hope you will find the time to report on the results more often in the future. I have a couple of questions: Could possibly tell us something about the era in which the sampled pots were made (modern, 80s/70s/60/50s, ROC, Qing,…)? I guess that this may be difficult for the fake one, but do you have a rough idea of when the authentic, antique, and cheap Yixing pots could have been made? And what is the salient property or properties that place the data for the authentic teapot in the upper half plane, away from the other pots’ data (remarkably, also including the antique), located in the lower half plane of the plot? Thanks a lot in advance.
teaman // October 14, 2016 at 1:39 pm |
Great article! Was curious your take on the cast iron / fake cast iron pots from China – how would one test them or be assured they were safe to use?
MarshalN // October 29, 2016 at 2:59 am |
What are you worried about, exactly?
ayumianraku // December 19, 2019 at 8:25 am |
Oh, interesting to know that the cheap ones are likely to be made of Japanese clay. I have seen some Japanese shops selling teapots claiming that their teapots had very similar properties to the yixing ones. I was wondering how similar they are when I came into this article.
Kara Marie // January 28, 2021 at 5:23 pm |
I wondered how I might test my 3 cast iron Japanese teapots for lead content. I love brewing my loose tea in them, but 2 have glazing inside, and 1 does not. Is there an over-the-counter test for this I could buy please?
MarshalN // January 28, 2021 at 8:17 pm |
I think there are testing kits easily available if you google for it, but I don’t think iron and lead usually go together in an alloy – at least not in the “let me add a little lead” way