Hi Dan,

Those are some interesting points. If there really were a very bright point right near Orion, I’m not sure whether we’d say that it was or wasn’t part of Orion. I suppose we could have a situation where Cassiopeia (the W-shaped thing) gradually drifted through Orion, and then we’d say that the stars belonged to one or the other, but in ordinary circumstances, I would have thought that proximity alone would make a bright point part of a (newly misshapen) constellation. But I’m not sure. I’ll have to think about that.

As for constellations being artifacts, that’s an interesting idea too. But I think you’re right that both points of view allow for that explanation. So in that case I’d have to rely more on arguments about what the sky is made of than the points about constellations themselves.

Really interesting post! You say that one of the main motivations for the collection-of-bright-points theory is that the location of a star is irrelevant to whether it’s part of a given constellation. I take it that what you have in mind is something like this: Let A be the set of stars associated with Orion and let B be some random set of stars that look from earth to be completely scattered but actually are no less scattered (relative to one another) than the ones in A. You want to say: there’s no ontologically significant difference between the stars in A and the ones in B that could explain why the ones in A compose something, and the ones in B don’t. So the only thing that could explain why the ones in A compose something is that we find their arrangement interesting. And that makes the facts about composition (and about what exists) objectionably mind-dependent. On the other hand, when it comes to bright points, at least we can appeal to their spatial proximity in explaining why them compose something.

But if that’s what you have in mind, I’m not sure how the move to bright points helps. After all, take some bright point in the vicinity of Orion’s bright points but that’s not part of Orion. Orion’s bright points together with this additional bright point don’t jointly compose a constellation. But again, there seems to be no difference between (i) Orion’s bright points and (ii) Orion’s bright points together with this extra one that could account for why the former but not the latter compose something, other than the fact that we find their arrangement interesting.

There are a few ways of dodging the problem, but they’re all going to be available to star-theorists too. For instance, suppose composition is unrestricted. Any plurality of bright points compose something — it’s just that which ones get classified as constellations is determined by how they look to us. But the star-theorist can say the same thing if they’re willing to accept unrestricted composition: whether some stars compose something isn’t at all mind-dependent; all that’s mind-dependent is which fusions meet the conditions for being a constellation (and it’s no surprise that that should be mind-dependent).

Another possibility is that constellations are artifacts. Just as states came into existence when we started drawing boundaries on a map of North America (or perhaps as soon as we started conceptually carving up North America), constellations came into existence when we started drawing boundaries on a map of the night sky (ditto). But again, both the point-theorist and the star-theorist can help themselves to this explanation.

Asterisms are actually closer to what I think has been discussed here. They are fixed bunches of stars: big dipper, “the” southern cross. If one picked out about a dozen stars and called it “Orion”, that would be an asterism. [this doesn’t settle whether it is bodies or lights that are being picked out by ‘star’, of course].

If we each have a stamp collection, I think we can trade some stamps without trading collections. But if I suddenly trade all my stamps for all of your stamps, I think we trade collections. Similarly, if Earth is suddenly transported across the galaxy, the collection of lines of sight is just swapped out altogether. Collections can’t undergo that kind of sudden, complete change.

It seems implausible to me that the collections of lines of sight would be destoyed if the earth were destroyed. We can imagine an astronaut might remain floating in space after earth is annihilated, and that she can see along the same lines of sight. The sky, I claim, would cease to exist. The collection of lines of sight would remain (if there ever was a collection of these in the ordinary sense of ‘collection’, anyway).

I think the point you’ve already made about collections saves the “sky as collection of lines of sight” thesis from your objections. Just as a stamp collection can survive replacement of some of its members (and even all of them, if the replacement is done in the appropriate way), and the stamp collection could also be destroyed without any of its members being destroyed (if say each stamp is sold off to a different collector in a different part of the world), so too could the sky be destroyed without any of the lines of sight being destroyed, or the sky could continue to exist with all the lines of sight changed if they were replaced by the appropriate means (like moving the Earth).

If constellations were not in space, then that would be a reason to deny that they are ordinary collections of stars, since ordinary collections of stars clearly are in space. But I don’t see why I should agree that constellations are not in space. Perhaps Kenny has in mind that constellations are not in space because they are in the sky and nothing in the sky is in space. But it seems pretty clear that there are stars in the sky. And certainly stars are in space. So it seems that being in the sky doesn’t stop something from being in space as well. So I say that constellations are ordinary collections of stars that are both in the sky and in space. (Perhaps ‘in’ means different things in the two claims.)

Finally, it doesn’t seem right to me that the sky is a collection of lines of sight. If earth were suddennly annihilated, the sky would cease to exist. But the relevant lines would remain. And, if the earth were moved across the galaxy, it seems to me that the sky would move with it. But the collections of lines of sight from the earth would be different after the move. So that collection must not be the sky.

That’s probably right about collections. But I now no longer think that the persistence conditions are the relevant argument (sorry for being a moving target). We definitely need to address persistence conditions for collections in general for the reasons both of you bring up, but the move to constellations not being composed of glowing balls of gas is motivated by other considerations.

First of all, as I mentioned in the post, the actual location of a glowing ball of gas is irrelevant to whether it counts as part of Orion or not – instead, all that matters is which direction it is located from Earth, and how bright it appears from Earth. (Moving the central star in Orion’s belt by 100 light years in any direction wouldn’t affect whether it’s part of Orion, but for some of the closer stars, even much smaller movements would affect their membership of Orion. There are hundreds of millions of stars in the relevant direction to be part of Orion, but only a dozen or two of them are bright enough when seen from Earth to count.)

Another issue is the fact that constellations are located in the sky, but they don’t seem to be located in space. I now think the question I should have started with is “what is the sky?” rather than “what is a constellation?” After all, not everything in space is in the sky – imagine a twin Earth that orbits the sun exactly opposite to Earth (which is a perfectly stable orbit). This twin Earth wouldn’t be in the sky at all, since it would always be blocked by the sun. Also, two things can be at the same place in the sky (as I believe Venus and Mars were a week or two ago), without being at the same point in space. My instinct is to repeat what I suggested above and say that the sky is just the collection of all lines of sight from the Earth, much as a visual field is the collection of all lines of sight from the retina (or is at least related to this collection of lines of sight in some close way). My main worry about this is that birds, planes, and Superman are sometimes in the sky, without being far enough away to make sense of which line of sight they occupy.

I’m still troubled by Jonathan’s points about ambiguity (especially given the way I’ve seemingly referred to the ball of gas star by means of the bright point in the sky star at several points above). But I’m not sure how serious a worry this is. Does a Cartesian dualist have to deny that the pronoun “I” can ever refer to my body rather than my soul? Then it would be false to say that I ran into a tree the other day when I wasn’t looking where I was walking. And if the word is treated as ambiguous, then it looks like Descartes should be able to say things like “I am connected to me by my pineal gland”, which sounds exceedingly awkward at best. I take it that there are fairly decisive arguments against Cartesian dualism, but that this isn’t one of them. Perhaps the better treatment here is to say that each of these words only literally refers to one of the things (“I” refers to the soul, “star” refers to the ball of gas) but that one can also refer to the other with the same word by means of metonymy. And it seems more difficult to use a word both metonymously and literally in a single sentence than to use an ambiguous word in two different ways. “Downing Street declared yesterday that Downing Street should be repaved.” “The White House told Congress that the White House should be repainted.” I think those sound awkward, though maybe they’re still not as bad as Jonathan’s examples.

But thanks for the comments!

This sentence sounds obviously false:

“Stars often emit stars, which travel to Earth.”

According to your view, there should be a true reading of this sentence. If there is a true reading, then, for Gricean reasons, we should expect it to be the dominant one. But it’s just clearly not.