But, as in any other science, some times the logic that have to be followed to prove a new theory can look perfect one day, and after admited as true the new theory, some one can pinpoint a weak steep on the demostration making it all to have to be proven again, avoiding or explainig correctly this steep.

It doesn't mean languaje is bad, is just that is not perfect in the sense that a bad prove of a theorem can look true to a the reader that don't have the appropiate background, and as new theries still lack of people that fully understand it, it is possible that latest demostrations finally is proven grown, like it happened with the last Fermat theorem proven some years ago: first prove was accepted for some moths, but it had a weak point, that was uncovered and part of the demostration had to be rewritting, with the risk of beeing even more complex to fix the problem than the theorem it self!

So languaje is not really perfect in maths, and it was proven than a perfect languaje for maths can NOT be found, due to de "incomplettness theorem" from Goedel, that comes to say that maths can never form a complete system in witch all possible theorems that the languaje can express have to be proven worng or right with the time... there always be theorems that can be expressed with maths but not be proveen (right or worng) using this languaje, so maths stop been called "exact science" from that point!