Why do digits add down to 9?

CharlesW

New member
Joined
Aug 7, 2013
Messages
15
Hello everybody. Greetings from Yokohama.

I've started Scott Flansburg's book 'Math Magic' and am stumped from the start. On p. 3 he introduces the fact that numbers higher than 9 add down to nine. That is, if you add the digits (until you get single digits) and then subtract the result from the original number, you get 9.

11: 1+1 = 2; 11 - 2 = 9
18: 1+8 = 9; 18 - 9 = 9
160: 1+6+0 = 7; 160 - 7 = 153; 1+5+3 = 9
etc.

He then points out that this gives rise to a 'nines check' for addition. So if we want to add 8+8 to get 16, we can check that this is the correct answer by subtracting the number (16) as follows:

88 - 16 = 72; 7+2 = 9

Unfortunately, the author gives no explanation as to _why_ this works -- or is it too obvious for me to see? And how does the first principle give rise to the second? I don't get the connection. I could memorize it by rote, but I'll never be able to really use it unless I understand the principle.

Many Thanks
Charles
 
Last edited:
Well, the difference is a multiple of 9, not necessarily 9 itself. This is called "casting out 9s" and is true because our numeration system is "base 10" and 9 is one less than 10.
11= 10+ 1= 9+ 1+ 1= 9+ 2 so that 11- 2= 9.
18= 10+ 8= 9+ 1+ 8= 9+ 9 so that 18- 9= 9
160= 1(100)+ 6(10)= 1(99+ 1)+ 6(9+ 1)= (11+ 6)(9)+ 7 so that 160- 7= 17(9)
35= 30+ 5= 30+ 5= 3(10)+ 5= 3(9+1)+ 5= 3(9)+ 3+ 5= 3(9)+ 8 so that 35- 8= 3(9).
 
Hello everybody. Greetings from Yokohama.

I've started Scott Flansburg's book 'Math Magic' and am stumped from the start. On p. 3 he introduces the fact that numbers higher than 9 add down to nine. That is, if you add the digits (until you get single digits) and then subtract the result from the original number, you get 9.

11: 1+1 = 2; 11 - 2 = 9
18: 1+8 = 9; 18 - 9 = 9
160: 1+6+0 = 7; 160 - 7 = 153; 1+5+3 = 9
etc.

He then points out that this gives rise to a 'nines check' for addition. So if we want to add 8+8 to get 16, we can check that this is the correct answer by subtracting the number (16) as follows:

88 - 16 = 72; 7=2 = 9

Unfortunately, the author gives no explanation as to _why_ this works -- or is it too obvious for me to see? And how does the first principle give rise to the second? I don't get the connection. I could memorize it by rote, but I'll never be able to really use it unless I understand the principle.

Many Thanks
Charles
Consider the number

\(\displaystyle \displaystyle a = \sum_{i=1}^n10^{i-1}d_i,\ where\ n\ is\ an\ integer > 1\ and\ d_i\ is\ an\ integer\ such\ that\ 0 \le d_i \le 9.\)

If n = 2, this simplifies to \(\displaystyle a = 10d_2 + d_1.\) With me to here?

Now we want to subtract the sum of the digits from a, giving

\(\displaystyle a - (d_1 + d_2) = 10d_2 + d_1 - d_1 - d_2 = 10d_2 - d_2 = d_2(10 - 1) = 9d_2.\)

So if n = 2, a minus the sum of its digits is a multiple of 9.

If n = 3, then we can rewrite a as \(\displaystyle a = 100d_3 + b.\) Still with me?

Now we want to subtract the sum of the digits from a.

\(\displaystyle a - (d_1 + d_2 + d_3) = 100d_3 + (b - d_1 - d_2) - d_3 = 100d_3 - d_3 + (b - d_1 - d_2) = 99d_3 + (b - d_1 - d_2).\)

But \(\displaystyle b = 10d_2 + d_1 \implies b - d_2 - d_1 = 9d_2.\)

So \(\displaystyle a - (d_1 + d_2 + d_3) = 99d_3 + 9d_2 = 9(11d_3 + d_2).\)

So if n = 3, a minus the sum of its digits is a multiple of 9.

Notice what we did, we proved it was true for n = 3 because it was true for n = 2. But, if n = 4, we can use the same process of breaking a up into two pieces, one with 1 digit (times 1000 this time) and one with 3 digits. We will get a number that is

\(\displaystyle 999d_4 + 9x = 9(111d_4 + x).\)

We can keep doing this forever.

When the sum of the digits of a positive integer greater than 9 is subtracted from that integer, the difference is evenly divisible by 9. Hope this helps.
 
Last edited:
I still remember this; ages ago (all done by hand then)!
Checking if multiplication is correct:

673 times 987 = 664251 ; is that correct?

673: 6 + 7 + 3 = 16 ; 1 + 6 = 7
987: 9 + 8 + 7 = 24 ; 2 + 4 = 6 : 7 * 6 = 42 ; 4+ 2 = 6 ****

664251: 6 + 6 + 4 + 2 + 5 + 1 = 24 ; 2 + 4 = 6 ****

**** YES!
That is incorrect reasoning, Denis. It is necessary that the final sums of the digits match each other, but it is not sufficient.

Counterexample: Does 81 multiplied by 81 = 6,651? Do your calculations on that example to see how it is not always true.


Summary: if the final sum of digits of the numbers being operated on (basic arithmetic operations, as an instance)
does not match the final sum of the digits of the alleged answer, then the alleged answer is definitely wrong.


However, if there is a match, then the test is inconclusive.
 
Last edited:
Thanks

I appreciate all the enthusiastic comments, everyone. I can already tell this board will be a good place to get help. But please give me time to study your replies. For now, can we limit the focus to the "adding down" rather than the proof issue.


Flansburg really does give many examples, and they all do indeed add all the way down to 9, even large numbers:


2+3+4+3+7= 19 = 23,437 - 19 = 23,418 = 2+3+4+1+8 = 18 = 1+8=9


@JeffM, your answer is so kind and elegant I want to frame it and put it on the wall. I feel guilty to have to confess that, alas, it is way above my head. Is there an easier proof, I wonder?


I sense Hallsoflvy is right in saying it has something to do with the base ten system.


I wouldn't bother about this point if it weren't for the face that Flansburg at the outset of his book seems to stress that this is an essential basic principle: "If you understand this," he says on p. 3, "then you have just turned on the calculator in your brain."


But he seems to expect that readers will simple be able to intuit the underlying principle from the examples. I fear it's pointless to try to proceed in the book unless I understand this first.


I will appreciate and study any further comments, if you give me time.


Many thanks
 
I appreciate all the enthusiastic comments, everyone. I can already tell this board will be a good place to get help. But please give me time to study your replies. For now, can we limit the focus to the "adding down" rather than the proof issue.


Flansburg really does give many examples, and they all do indeed add all the way down to 9, even large numbers:


2+3+4+3+7= 19 = 23,437 - 19 = 23,418 = 2+3+4+1+8 = 18 = 1+8=9


@JeffM, your answer is so kind and elegant I want to frame it and put it on the wall. I feel guilty to have to confess that, alas, it is way above my head. Is there an easier proof, I wonder?


I sense Hallsoflvy is right in saying it has something to do with the base ten system.


I wouldn't bother about this point if it weren't for the face that Flansburg at the outset of his book seems to stress that this is an essential basic principle: "If you understand this," he says on p. 3, "then you have just turned on the calculator in your brain."


But he seems to expect that readers will simple be able to intuit the underlying principle from the examples. I fear it's pointless to try to proceed in the book unless I understand this first.


I will appreciate and study any further comments, if you give me time.


Many thanks
Halls of Ivy is absolutely right that it has to do with decimal notation.

And elegance that is incomprehensible is worthless. So rip that post of mine off the wall.

Let's start by explaining this in terms of two digit numbers. What does 43 mean? It means (4 * 10) + 3. Now what is the sum of the digits?
It is 4 + 3 = 7. Now let's subtract the sum of the digits in its expanded form from the actual number in its expanded form.

\(\displaystyle (4 * 10) + 3 - (4 + 3) = (4 * 10) - 4 + 3 - 3 = 4 * 10 - 4 = 4 * (10 - 1) = 4 * 9, which/ is/ evenly/ divisible/ by/ 9.\)

Now this is not an accident. Consider the number \(\displaystyle ab = (a * 10) + b.\)

That is ab is not meant as multiplication but the first and second digits respectively of a two digit number. You with me?

The sum of the digits is \(\displaystyle (a + b).\)

So the number minus the sum of its digits \(\displaystyle (a * 10) + b - (a + b) = (a * 10) - (a * 1) + b - b = a * (10 - 1) = 9a.\)

The reason that this all works is that the unit digit subtracts out, and we get 10 times the tens-digit minus the tens-digit which is always 9 times the tens-digit, so the number is ALWAYS divisible by 9. 10 - 1 = 9, now and forever.

You solid with that?

Now here is where things get a little tricky. There is a form of proof (called weak mathematical induction) that says if I can prove that something is true for some particular integer and if I can also prove that if that something is true for any integer equal to greater than that particular integer, then it is also true for the next higher integer as well, it is true for all integers equal to or greater than the particular integer. The intuition is this. In this case, I prove our adding up to a multiple of 9 idea is true for a two-digit number. Then I prove that if the adding up idea is true for for an n-digit number, where n is not smaller than 2, then it is also true for an (n+1)-digit number. Now this means that because it is true for 2, it is true for 3, which means that it is true for 4, but then it must be true for 5, and that in turn must mean it is true for 6,etc, world without end. The whole line of dominoes falls.

So let's say our divisible by 9 idea works for any number with n digits. Let's consider the number x expressible in (n + 1) digits.

\(\displaystyle Then\ x = (10^n * y) + z, where\ z\ is\ a\ number\ expressible\ in\ n\ digits.\)

\(\displaystyle Let\ u = the\ sum\ of\ the\ digits\ of\ z.\) But we know that z - u = 9v.

Still with me?

OK then the sum of the digits of x is y + u. So what is the difference between x and the sum of its digits?

\(\displaystyle (10^n * y) + z - (y + u) = (10^n * y) - (y * 1) + z - u = y(10^n - 1) + 9v.\)

Now when I subtract 1 from a power of 10 greater than or equal to 2 I get a number that is a string of 9's

10 - 1 = 9

100 - 1 = 99

1000 - 1 = 999

10000 - 1 = 9999.

So that number is evenly divisible by 9. (If I was giving a super formal proof I'd need to prove that, but it is obvious.)

\(\displaystyle Let\ 9w = (10^n - 1) \implies y(10^n - 1) + 9v = 9wy + 9v = 9(wy + v).\) Evenly divisible by 9.

I've given it my best shot.
 
Thanks

@JeffM


Thanks so much for your patient help. I do follow the two examples, for example rewriting -a as -(a*1) and then factoring it to end up with 9a. I can do this mechanically.


Maybe if I keep thinking about your explanation I may eventually understand the principle.


Thanks again for your patient help.


Charles
 
Top