It is a popular misconception that whenever a window function is used to ( ) ( ) treat a free induction decay FID , an improvement in the signal-to-noise ratio S / N is always in conflict with an improvement in resolution. Furthermore, it has also been assumed that it is impossible to treat noise differently than signal when using window functions, because of the multiplicative property of such functions. These misconceptions are discussed in this article, which shows that in special cases a process that uses window functions cannot only treat noise differently than signal, but it can also produce an improvement in S / N with no change in resolution. The process relies on using one function to leave the real part of the FID unchanged, but not the imaginary part, while increasing the S / N. A different, but similar, function is used to accomplish the reverse, that is, leave the imaginary part unchanged but not the real part, while also increasing the S / N. Then, the real part from the first operation is combined with the imaginary part from the second one to yield an unchanged decaying signal with less noise. Although, this procedure works perfectly only in special cases, it should provide the useful purpose of altering the way spectroscopists view the basic nature of window functions.