In linguistic typology, mean value has many advangates. Mean value can be a reference point for surveying tendency of some linguistic properties, and used for deciding which is marked or unmarked. These merits stand out in phonological typology particularly. There are some methods of calculating mean value. It can be acquired by selecting languages at random and sampling with diversity value. In fact various mean value has been suggested by these methods. But each one has a few of problems, so it need a new approach. This paper offers another way of measuring a mean value. This new method removes a flaw of subjective data selection and at the same time prohibits a data bias into some language families. By this approach, we calculated mean values of consonant and vowel. The results tell us that mean values of consonant and vowel by new method are put between the maximum and the minimum of the existing results, which are closer to the maximum.