rainy seasons, which prevails among the Unyamwesi in Africa.[173] How powerful is the posthumous influence even on later times of the nomadic lunar division into weeks,—an influence which again and again obtained validity, even after it had been once supplanted by the solar reckoning by decads, we see best among the Romans. They had originally a consistent lunar computation; even their year consisted of ten months, the sun’s cycle of twelve months being ignored; and they divided the month into four weeks.[174] Later, this fourfold division gave way to a threefold division into three decads, _nonae_, _kalendae_, _idus_; but yet they returned at last to the week again, and called its seven days by the names of the sun, the moon and the five planets. However, the division of the month into three decads is not always connected with solar chronology; it is also found in combination with lunar reckoning, when three phases of the moon are acknowledged (as in the three-headed forms of the moon in the Greek mythology).[175] A _five-days’_ period has been proved to exist in many nations as the equivalent of our week (among the Chinese, Mongol tribes, Azteks, and Mexicans.)[176] But this division into pentads must be connected with an original quinary system of numeration, to the linguistic importance of which Pott has devoted a special treatise.[177] In Old Calabar on the west coast of Africa a week of _eight days_ occurs; most curiously, as the people cannot count beyond five.[178] _A priori_ this would seem impossible; but it is vouched for by an observer so accurate as Bastian.