I was working with rational numbers and I found something which I consider interesting. That, a rational numbers' decimal expansion is periodic with its length AT MOST the value of the denominator.

The first part, that it is periodic I learned a long time ago. But the second about the size of the period never exceeding the denominator I never heard of before.

---------

To keep things simple, consider .

To find its decimal expansion you use an algorithm which you learned in the 5th grade. But I am going to be more formal, define integers as,

for

for

for

......

for

In other words, it is a infinite use of the division algorithm.

First, notice that for all because if then there is no way that because already .

My point is, that is thus, a valid decimal expansion because all lie between zero and nine. Next, I claim that .

We do this by showing that the infinite series,

converges to, .

We do this by looking at its partial sums,

Notice we can solve each equation obtained by the division algorithm to get,

Thus,

Recognizing the telescoping effect we have,

But, because

And since,

direct comparasion test states that as .

Thus, we proved that this decimal expansion converges to .

Next, we have some fun. Notice, that if any one of the is the same as then the algorithmic steps following need to be the same as the algorithmic steps following . For example, if then, Step 3 is same as Step 2 because it itself was same as Step 1. Thus, Step 4 is same as Step 3 and so one. I know it is difficult to say want I want to say, possibly because I did not find an elegant mathematical way of expressing my thought. Anyways, by the pigeonhole principle, if there are steps one of them if bound to match another one. Then by the previous explanation the steps would repeat, thus, fractional expansions must be periodic and further more CANNOT exceed the denominator.