In formal terms, you are confusing the abstraction of a mathematical construct with an application of a number system.
In regular English:
The fraction 1/3 is a perfectly legitimate number in every math sense. It is the multiplicative inverse of 3. Three has an inverse because the integers form a ring (an abstract algebra term).
Thus, the fraction 1/3 has a purpose and a value in mathematics.
To measure it outright, on the other hand, is difficult. It is, however, by no means impossible. No more so than measuring 10.9 or 45.63 or even 5.0 What happens is that we realize that we will never have precision that is infinitely small (precision is the possible error in measurement). Another way of saying this is that we will always have limited accuracy.
Thus, we accept that a (wide) range of exact values be represented by a number of a particular precision.
A perfect example of this is when we buy gas. The pumps measure the price of gas in tenths of a cent, but we round to the nearest cent. We accept this "inaccuracy" because our smallest coin is a penny (one cent).
And so it is with other measurements. We accept an approximation for 1/3, not because it is a paradox or doesn't make sense, but because we realize that we can never measure it perfectly anyways.
For that matter, we can never measure ANY NUMBER perfectly. Try measuring 1.0 inches (or centimeters). All you really know is that you have approximated that length to a precision beyond which you cannot discern.
In other words, "it is close enough to your eyes". That is the way we apply these perfect notions we often find in mathematics. But more often than not, these applications work and work well.
Something else that may boggle your mind is to take your decimal representation of 1/3 = 0.333... and multiply it by three:
3*(1/3) = 3* (0.333333.......)
1 = 0.99999999999........
Isn't math fun?