Desperately seeking constellations?

What do you see? Photo by Andrew Coulter Enright.

I have a problem. I think I have an experiment to solve it, but I’d appreciate feedback.

I’ve been given a diagram of a potsherd from the Greek colony of Pithekoussai to look at. It’s quite dull, though on one side there’s some interesting scratches. These are two five-pointed star shapes and some lines which may be a figure, so this graffito could be one of the earliest drawings of a constellation. A top classicist has already looked at it and said that it’s probably a specific constellation, which I won’t name for reasons which will become obvious later. I’ve looked at it and I’m quite certain it’s another constellation. I prefer my explanation because it fits better with how astronomers work. However there’s a problem and it’s quite a big one.

Humans love to see patterns in things. As I said I’m fairly certain this is constellation X, but if I were given another potsherd with random scratchings on it and a couple of star symbols doodled to one side, would I say that this must be another constellation? Because if I’m the sort of person who pulls significant details out of random scratches then I can’t really be certain that I’ve identified the constellation correctly. What I’ll end up doing is building what could be a fantasy on a pattern I want to see. What I really need are expert opinions.

I might have access to these in a couple of weeks. I’m giving a talk at the National Astronomy Meeting where I’m expecting to find lots of astronomers. What I thought to do is ask their opinion. Asking them if this scratch is constellation X is a bit foolish. It’s a bit like finding a vaguely duck-shaped cloud and asking people if they can see a duck in it. If you point it out they might, but would that be meaningful? I did think about asking people to identify the constellation without telling them what it is, but even that is leading the witness. You’re telling them it is a constellation and asking them for the best fit. From any random scatter of dots you’ll always find a closest match to a constellation.

So here’s my experiment plan.

I design three glyphs. One shows this astronomical graffito, glyph A. The second is plotted randomly, glyph B. The third will either be random or a depiction of a another constellation. I haven’t yet which for glyph C. I then print out three batches of postcards. One will have glyphs A and B on it, another with glyphs B and C and a third with glyphs A and C. On the reverse I shall print instructions along the line of.

“An astronomical graffito has been recently discovered on a potsherd. This is an experiment to examine if these are random scratches or an depiction of a constellation. Overleaf are two diagrams. I would be grateful if you could identify which diagrams depict a Graeco-Roman constellation. Depending on which postcard you have this could be neither diagram, one or both. If you think a constellation is depicted I’d be grateful if you could write in the space provided which constellation you think it is. This experiment will be mentioned in the talk ‘Miscommunication in Archaeoastronomy’ in the Historical Astronomy session. If you wish to discover the results of the experiment please write in your email address below.” There may also be a plug for our Interdisciplinary Science course.

If A is convincing I’d hope for higher recognition scores than B. C would be the control sample. If it’s genuine then it should match A’s recognition score, if it’s random it should match B’s score. My aim is to try and see if the constellation on A is recognisable without the viewer being primed to see it. Hence the lack of illustrations or names in this post.

Can anyone spot any obvious flaws in this plan? The biggest one I can see is that A is only being compared with a small sample. Ideally I’d want lots of pairs for people to test, but there’s a limited amount of time and an even more limited budget.

The other problem is that it could prove me wrong. I have this beautiful scheme re-thinking how acculturation happened in the western Greek world and, if I’m right, this would help prove something that should be obvious. Fortunately it doesn’t all rest upon this one piece of evidence and seeing things my way. —–