Each name has a prefix (specifier) and a suffix (generic).

Either may be absent. The attributes of these prefixes and suffices are qualitative and multi-valued. They are also constant. Integer numbers allocated to each attribute will classify them. There are eight possible values assumed by the attributes. Therefore if the value of the prefix attribute is np, then 1<np<8. This is an extensive variable, that is, the value of np found from a small sample can be added to the value found from a larger sample to give the value for the sum of the samples. Moreover, the attributes are conditions. They depict a situation, not behaviour. The combination np.ns is of interest. If the probability of np or ns assuming any value between 1 and 8 is equal, then this probability is 1/8.

Consequently the probability of the combination np.ns occurring is 1/64. This is the case when the values of the attributes of the prefixes and suffixes are randomly distributed. If this is not the case, the probability of np assuming the ith value (1<i<8) is K1/8 when 1<K1<8. Likewise, we can state the probability of ns assuming the ith value of the attribute to be K2/8. Therefore the probability of the combination np.ns is (K1*K2)/64.

The first test must be on a large sample to check the hypotheses that the probability of np and ns having any value between 1 and 8 is
K1/8 for K1=1 and K2/8 for K2=1. If not, the values of K1 and K2 are good estimates for the values of the population. Any smaller sample exhibiting significantly different values will lead us to reject the uniform naming habit hypotheses.