[summary: Subtraction is "adding anti-apples", or equivalently removing some apples you already have.]
So far, we have met the idea of a rational number, treating them as chunks of apples, and how to add them together. Now we will discover how the idea of the anti-apple (by analogy with the integers' anti-cow) must work.
The anti-apple
Just as we had an anti-cow, so we can have an anti-apple. If we combine an apple with an anti-apple, they both annihilate, leaving nothing behind. We write this as 1+(−1)=0.
A very useful thing for you to ponder for thirty seconds (though I will give you the answer soon): given that an means "divide an apple into n equal pieces, then take a copies of the resulting little-piece", what would −1n mean? And what would −1n mean?
%%hidden(Show possible solution): −1n would mean "divide an apple into n equal pieces, then take −1 copies of the resulting little-piece". That is, turn it into an anti-little-piece. This anti-little-piece will annihilate one little-piece of the same size: −1n+1n=0.
−1n, on the other hand, would mean "divide an anti-apple into n equal pieces, then take 1 copy of the resulting little-anti-piece". But this is the same as −1n: it doesn't matter whether we do "convert to anti, then divide up the apple" or "divide up the apple, then convert to anti". That is, "little-anti-piece" is the same as "anti-little-piece", which is very convenient. %%
What about chunks of apple? If we combine half an apple with half an anti-apple, they should also annihilate, leaving nothing behind. We write this as 12+(−12)=0.
How about a bit more abstract? If we combine an apple with half an anti-apple, what should happen? Well, the apple can be made out of two half-chunks (that is, 1=12+12); and we've just seen that half an apple will annihilate half an anti-apple; so we'll be left with just one of the two halves of the apple. More formally, 1+(−12)=12; or, writing out the calculation in full, 1+(−12)=12+12+(−12)=12
Let's go the other way round: if we combine an anti-apple with half an apple, what happens? It's pretty much the same as the opposite case except flipped around: the anti-apple is made of two anti-half-chunks, and the half apple will annihilate one of those chunks, leaving us with half an anti-apple: that is, (−1)+12=−12.
We call all of these things subtraction: "subtracting" a quantity is defined to be the same as the addition of an anti-quantity.
General procedure
Since we already know how to add, we might hope that subtraction will be easier (since subtraction is just a slightly different kind of adding).
In general, we write −1n for "the 1n-sized building block, but made by dividing an anti-apple (instead of an apple) into n equal pieces". Remember from the pondering above that this is actually the same as −1n, where we have divided an apple into n equal pieces but then taken −1 of the pieces.
Then in general, we can just use the instant addition rule that we've already seen. %%note:Recall: this was am+bn=a×n+b×mm×n. Remember that the order of operations in the integers is such that in the numerator, we calculate both the products first; then we add them together.%% In fact, am−bn=am+(−bn)=a×n+(−b)×mm×n=a×n−b×mm×n
Why are we justified in just plugging these numbers into the formula, without justification? You're quite right if you are dubious: you should not be content merely to learn this formula%%note:This goes for all of maths! It's not simply a collection of arbitrary rules, but a proper process that we use to model our thoughts. Behind every pithy, unmemorable formula is a great edifice of motivation and reason, if you can only find it.%%. In the rest of this page, we'll go through why it works, and how you might construct it yourself if you forgot it. I took the choice here to present the formula first, because it's a good advertisement for why we use the an notation rather than talking about "1n-chunks" explicitly: it's a very compact and neat way of expressing all this talk of anti-apples, in the light of what we've already seen about addition.
Very well: what should am−bn be? We should first find a smaller chunk out of which we can build both the 1m and 1n chunks. We've seen already that 1m×n will work as a smaller chunk-size.
Now, what is am expressed in 1m×n-chunks? Each 1m-chunk is n of the 1m×n-chunks, so a of the 1m-chunks is a lots of "n lots of 1m×n-chunks": that is, a×n of them.
Similarly, bn is just b×m lots of 1m×n-chunks.
So, expressed in 1m×n-chunks, we have a×n lots of positive chunks, and b×m lots of anti-chunks. Therefore, when we put them together, we'll get a×n−b×m chunks (which might be negative or positive or even zero - after all the annihilation has taken place we might end up with either normal or anti-chunks or maybe no chunks at all - but it's still an integer, being a number of chunks).
So the total amount of apple we have is a×n−b×mm×n, just like we got out of the instant formula.
[todo: the rest of the page, including examples; make an exercises page, including subtracting negatives and so on]