This leads into a nice discussion of what counts
evidence. It's all very well to say
"All swans are white": it sounds
pretty clear-cut, and all you have to do is wait until a single counterexample
comes along to disprove
What counts as a swan, and what counts as being non-white? What is the logic of our naming? The simple answer is that there is no one simple answer. "Swan" and "white" both vary somewhat depending on context.
The swan is a very distinctive bird. Although there are several species, they are all unmistakably like each other, and quite unlike geese or ducks. All swans are white, at least if we retreat to the England of the 1600s, where we are speaking English but have not yet been swamped by explorers such as Dirk Hartog (1616) or William Dampier (1688), who were the first to explore Western Australia, perhaps the first Europeans to see black swans. I don't know the true story*, but let's suppose we're in England in 1690, Captain Dampier is approaching Portsmouth after his navigation of the seven seas, and on board, pickled in rum, he has the corpses of several interesting New Holland animals, including a bird that looks almost exactly like a swan, except that it's black.
How do we assess "All swans are white" while we are still ignorant of these new-found creatures? How then do we re-assess it when Dampier docks and shows us his specimens?
First, the obvious dimension: pure white, snowy white, or an off-white or cream? This terrible word "really". Is an off-white swan really white? (Hint: If you use the word "really" as part of a philosophical argument you're probably on the wrong side.)
Does it make a difference if it's off-white because its feathers are that colour; - if it's greenish because it's been swimming in stagnant water? How about if it's pinkish because its diet consists of small crustacea that turn it pink? (I have an idea this is actually true of the flamingo, but even if it's not, it's enough that it be possible.) So this swan deprived of its habitat or diet, thoroughly washed and brushed, might be a somewhat different colour to its state out in the wild. (Which is its real condition?) Do we say "that's not a white swan, but it would be if you washed it"? Or do we say "that's a white swan but it's covered in green slime"? Does it matter?
Swans have two legs, two eyes, and a beak too. But if I chop off one of its legs, does it cease to be a swan? Of course not. Why not? Well instead of insisting "All swans have two legs" you could say "Most swans have two legs". Most? But they all do. Apart from (i) freak mutations, and (ii) accidental or deliberate mutilations. "All swans have two legs except those that have some other number"? No: there is a real truth about swans, a natural property of them, that they generically have two legs. Exceptions like these don't invalidate the rule.
Think about a case where it's clearly not true that "All X are Y". All cats are grey? No: you get grey, white, black, ginger, tortoiseshell, and tabby cats. You get yellow, green, and blue budgerigars. These are clear examples where species vary, in the way in which swans and ravens do not vary. A blue budgerigar is not a kind of queer exception to a general run of green budgerigars: you don't have to puzzle over whether it's "really" a green one with some odd dietary or environmental condition. But - of course you can get blue ones turned black with soot, etc.
The point I'm trying to make is that there's no simple one-size-fits-all answer to "What colour is X"? You can't just look at something and say what colour it is. The question needs context: do you mean, underneath, or at the moment, or during the day, or under this sodium light or in their deep-ocean habitat or...? A thing can be different colours in different circumstances (not just lighting but living conditions), and you just can't pick one of these out and say "that's its TRUE colour" (or REAL colour) or "its colour".
Dampier's ship arrives. He unloads these birds. They're remarkably like swans except they're black. Are they swans? Do we call them swans? We don't have to: Dampier might have learnt the native name (kookaburra, currawong, or whatever), and we might happily call it that. Then we say a currawong is very similar to a swan, but currawongs are black whereas swans are white.
Or we might choose to call this a "black swan" because it looks very similar to a swan, but is black, whereas swans are white. Why not? We did that with tigers. The Tasmanian tiger was so named because it looked (somewhat) and behaved like a tiger; but of course it's marsupial, not a close cousin to the Siberian tiger and Bengal tiger. It's just a name. The name "Tasmanian tiger" is fine, as long as you realize it's a descriptive, not a genetic name.
So this new West Australian bird, we can descriptively call it a name based on "swan": it looks like a black swan so let's call it "black swan": but that doesn't make it a swan. If this is why we named it that, the we have two animals, (i) the swan (= the traditional English white bird), and (ii) the black swan. All swans are still white, in this classification.
In fact, the black swan is of the same genus as the English swans, so it is reasonable to say they are swans. But a genus is not a natural kind: it doesn't automatically follow that you naturally lump anything in the same genus under the same common name "swan".
* sid's write-up on the black swan says the date of discovery was 1697; but there is an interesting earlier example: in 1660 there was a pub in Lincoln called the Black Swan. My source for this is some British postage stamps issued in August 2003 featuring pub signs. This does not imply prior zoological knowledge or even fortuitous guessing, since pub signs often feature heraldic beasts in unnatural colours.